CHEMISTRY OF SILICA IN CERRO PRIETO BRINES

Oleh Weres and Leon Tsao Lawrence Berkeley Laboratory University of California Berkeley California 94720

ABSTRACT

We have studied the precipitation of amorphous silica from synthetic geothermal brines that resemble the flashed brine at Cerro Prieto We found that part of the dissolved silica quickly polymerizes to form suspended colloidal silica The colloidal silica flocculates and settles slowly at unmodified brine pH values near 735 Raising the pH of the brine to about 78 by adding base and stirring for a few minutes causes rapid and complete floculation and setting These results have been conshyfirmed in the field using actual Cerro Prieto brine Several commercially available flocculating agents were also tested Both in the Laboratory and in the field we found quaternary amines to be effective with some brine comshyposition but not with others Polyacrylamides do not work at all

These results suggest the following simple preinshyjection brine treatment process age the brine for 10 to 20 minutes in a covered holding tank add 20 to 30 ppm lime (CaO) stir for 5 minutes and separate the floccushylated silica from the brine using a conventional clarifier The brine coming out of such a process will be almost completely free of suspended solids

The pilot plant tests needed to reduce this conmiddot ceptual process to practice are discussed

In a separate study we researched the rate of deposition of silica scale from synthetic brines We found that a modest decrease in pH could significantly reduce the scaling rate at a reasonable cost

INTRODUCTION

The work reported here was conducted in support of the ongoing studies on brine reinjection at Cerro Prieshyto being conducted by the personnel of the Comisi6n Federal de Electricidad (CFE) and the Instituto de Inshyvestigaciones Eh~ctricas (liE) The work which was conducted at the Lawrence Berkeley Laboratory (LBL) is part of the CFE-DOE Cooperative Program at Ceshyrro Prieto

For the purposes of this study we assumed that the brine to be reinjected would be flashed and separatshyed from steam twice and that it would be delivered to the brine treatment facility at atmospheric pressure and a temperature near 100degC These assumptions are consistent with CFEs longterm field development plan Under these conditions Cerro Prieto brine is supersaturated relative to amorphous silica by about three-fold and the excess silica in solution is rapidly converted to susshypended colloidal silica by homogeneous nucleation and growth of colloidal particles The flocculation of this colloidal silica and its adhesion to solid surfaces creates the massivemiddot white amorphoJs silica deposits that form throughout the existing surface brine disposal system at Cerro Prieto

Injecting brine that contains 400 to 500 mgI-1 suspended solids into a reservoir having pore-dominatshyed permeability like that at Cerro Prieto will cause severe formation damage Therefore this colloidal silshyica must somehow be removed prior to reinjection

The major task assigned to us was to research methods to remove silica from flashed brine to make it fit for reinjection A secondary task was to study the rate of silica scale deposition from flashed brine at about 100degC and to find a means of controlling it

The experimental techniques we employed in this work are an extension of the techniques developshyed during an earlier study of the chemical kinetics of silica polymerization in geothermal-brine-like solutions (Weres et aI 1980b)

The staff of the liE and CFE laboratories at Ceshyrro Prieto have recently performed similar experiments and confirmed our results (Hurtado et al 1979) They are presently testing the brine treatment process that we recommended to them on a pilot plant scale and the results of this work should be available shortly

The carbonates and sulfates of calcium stronshytium and barium are the materials other than silica

568

most likely to precipitate and cause formation damshyage after reinjection This possibility was studied theoretshyically The results of this work are reported elsewhere (Iglesias and Weres 1980)

The contents of this paper and that by Iglesias and Weres (1980) are included in the report by Weres (1980) are included in the report by Weres et al (1980a) which also describes some additional work not discussed here

Calgon Magnifloc Nalgene Purifloc Separan Teflon and Tygon are all registered trademarks although they are not always identified as being such in the text and figures

REAL BRINES AND SYNTHETIC BRINES

We studied the precipitation of amorphous silica from Cerro Prieto brines by preparing synthetic brines that closely resemble them in our laboratory The approxishyrnate chemical compositions of two such synthetic brines which were extensively used in the silica removal studies are presented in T abl e 1 The silica concentrations and pH vallJes shown in the table were those we used most often However some work was also done at different silica concentrations and pHs The concentrations of all other components were always approximately as shown in the table (The concentratlon of sodium varshy

ied slightly with the silica concentration and pH Also evaporation tended to concentrate the brines slightly during the experiments)

These synthetic brines were formulated to have concentrations of their major components in the same range as real Cerro Prieto brines that have been flashshyed down to atmospheric pressure The high-Ca forshymulation was more or less arbitrarily chosen to resemble a typical flashed brine at Cerro Prieto

The low-Can formulation approximately corshyresponds to a mixture of the waste brine streams from Cerro Prieto wells M-8 M-27 M-31 M-35 and M-46 This composition was chosen because the waste brine line shared by these wells is the source of brine used for most pilot plant work at Cerro Prieto Chemical components that we consider unlikely to directly affect the chemical behavior of the silica in the brine under the conditions of greatest interest were left out of our synthetic brines

The pH value 73 which we chose for the synshytheticl)rines is typical of the pH values of freshly flashed brine specimens determined in the field with still~hot brine These range between about 70 and 76 and are most often about 73 or 74 (private communication from A Manon M and our own field measurements)

TABLE 1 NATURAL AND SYNTHETIC CERRO PRIETO BRINES

Natural brines Synthetic brines Well i4 Well 30 Low Ca

mgl-l mmole mg I-I mmolel- l mg mg I-I mmolemiddotl

Na 7079 308 7809 340 6970 333 6610 288 K 1439 37 1833 47 1502 38 1436 37 Ca 445 11 596 15 501 12 329 8 CI 13113 370 15173 428 12940 365 12089 341 Si02 960 16 1077 18 1000 17 1094 18 B 18 17 175 16 11 10 131 12 F 24 013 13 007 Sulfate 11 011 13 014 Carbonate + 50 082 35 057

Bicarbonate As 15 002 047 0006 Ba 107 0078 80 0058 Cu lt005 lt0001 i 002 Fe 08 0014 05 0009 Mn 02 0004 26 0047 Sr 176 020 188 021 Zn 001 00002 001 00002 Barbital buffer 20 20 pH 72-74 72-74 T 950 C 95 0 C

Values quoted and recalculated from Cosner and Apps (1978) Records 68 and 106 This data was generated by the CFE laboratory at Cerro Prieto and kindly provided to lBl by A Maii6n M

569

Also shown are typical analyses of flashed brine from wells M-14 and M-30 We used brine from these wells in our limited field work with freshly flashed brines at Cerro Prieto

The brines at Cerro Prieto are weakly buffered The major buffer at temperatures near 100degC and pH values between about 72 and 95 is monosilicic acid Si(OHk (Henceforth monosilicic acid will be abbreshyviated as MSA) At lower pH values the carbon dioxshyide-bicarbonate buffer system is dominant The nature of our experiments was such that it would have been very hard for us to adjust and control the pH of our synthetic brines by relying on these buffer systems alone Therefore we included a small amount of the buffering compound barbital (5 5-diethylbarbituric acid) in our synthetic brine formulation The pka of barbital is about 74 Bicarbonate was usually left out of the synthetic brines despite being a relatively imshyportant real brine component because it has no direct effect upon the behavior of silica and its presence would have complicated pH adjustment Boric acid was includshyed because we know that it can have mild inhibiting effect upon silica polymerization under some conditions (see Weres et aI 1980b Section 315)

In all experiments the artificial brine was conshystituted and silica polymerization initiated by mixing together two preheated solutions one acid and one alkashyline During the experiments the brine temperature was always between about 92 and 97degC

SILICA REMOVAL STUDIES EXPERIMENTAL METHODS

In these experiments the alkaline solution consisted of a premeasured amount of dilute sodium metasilshyicate sOlutiOn The acid solution contained the apprOshypriate amOunts Of the majOr compOnent salts sOdium barbital acid to adjust the buffer pH and neutralize the silica and water to make up the tOtal desired volshyume A tOtal volume Of 300 ml was usually employed

ThrOughout Our wOrk with simulated brines all vOlumes were measured at rOOm temperature ThereshyfOre when we say 300 ml we mean that the vOlume WOuid be that if the solutiOn were coOled down to room temperature When applied to salt concentrations and initial silica cOncentrations units Of moles or grams per liter are alsO used in this sense Because of the reshylatively low salt content of these solutions and the semishyquantitative nature of mOst of the synthetic brine data this is for all practical purposes the same as moles or grams per kilogram of water

The sOdium metasilicate stock solution contained about 2 gI-1 Si02 Because it was hard to make up

this solution to an exact predetermined concentration it was always standardized by comparison with a standshyard silica solution (Fischer AA standard)

The desired amount of sodium metasilicate sOshylution was put into a plastic bottle some of the remainshying air squeezed out and the cap closed tightly It was then put into a boiling water bath to preheat (This procedure ensured that the bottle would neither rupshyture nor lose water when heated) A separate 50-ml aliqUot of the sodium metasilicate solution was titratshyed with 05 M HCI using a pH meter to determine the volume of HCI solution needed to neutralize the given amount of sodium metasilicate sOlution

The acid starting solutiOn was prepared from a stock solution of concentrated brine a stock SQshy

lutiOn of sodium barbital 05 M HCI and water The concentrated brine was a mixture of sodium potassium and calcium chlOrides and boric acid in water formushylated SO that putting 3 parts of it into 20 parts of the final synthetic brine would give the desired final mashyjor component concentrations For example in the case of the high Ca synthetic brine presented in Table 1 the concentration brine cOntained 1667 M NaCI0256 M KCI 00833 M CaCI 2 and 00066 M H3 B03 The conshycentrations of the last three components in this concenshytrate are simply 203 times their desired concentrations in the synthetic brine The concentration of NaCI is somewhat smaller because the sodium metasilicate and buffer stock solutions also contain sodium

In the case of the Low Ca synthetic brine the concentrated brine contained 1547 M NaCI 0245 M KCI00547MCaClzandO008l M H3BOa

The barbital buffer stock solution was prepared to be 05 M in barbital and 05 M in NaOH Purified grade Mallinckrodt barbital intended for pharmaceushytical use was employed (All chemicals used other than the barbital and the commercial flocculants were of reagent grade)

To prepare the acid solution measured amounts of the brine concentrate buffer stock solution and the amount of water needed to bring the final volume up to that desired were mixed in a beaker suspended in a boiling water bath The beaker was temporarily covered with a watchglass and its cOntents allowed to heat up to above about 90degC Next the mixture was titrated with 05 M HCI to adjust the pH value to that desired in the synthetic brine Finally the amount of additional 05 M HCI needed to neutralize the sodium metasilicate solution as determined by the earlier tishytration was added to the beaker and the resulting mixshyture was again allowed to heat up

570

To start the reaction the preheated alkaline soshylution was poured into the beaker containing the preshyheated acid solution with rapid stirring The stirring was stopped after a few seconds the pH measured and periodic 1-ml samples withdrawn for analysis from then on

Both dissolved silica and total silica were detershymined for by the molybdate yellow method To deshytermine the molybdate active (ie dissolved) silica the molybdate yellow method was employed directly To determine the total silica (Le including the susshypended colloidal silica) the specimen was first reacshyted with NaOH to digest and dissolve the colloidal sishylica (Henceforth molybdate active silica will be abbreviated as MAS which should not be confused with MSA MSA is molybdate active but so are some other small molecular and ionic silica species)

The difference between the empirically detershymined molybdate active and total silica concentrations is approximately equal to the amount of colloidal sishylica suspended in the brine This differential techshynique is sensitive down to about 10 mgI-1 suspended silica

The measured silica concentrations reported in the figures and tables were not corrected for the effect of temperature Therefore the measured silica concenshytrations really are in units of gp at the stated tempershyature To approximately convert these values to units of grams (kgHz0)-1 multiply them by 105

The sampling syringe was always preheated by rapidly drawing brine in and out of it once or twice before actually taking the sample to be analyzed

The molybdate yellow method is subject to seshyrious interference by the traces of dissolved silica and inorganic phosphate that are usually present in tap Washyter To eliminate this problem all solutions were preshypared using water that was twice doubly deionized (01) with mixed anioniccationic ion exchange resins The water coming out of our laboratorys 01 water tap was already once doubly deionized as a precaution we sent it through our own column filled with mixed resshyins for the second double deionization step

As in our previous work we used the molybdate yellow procedure recommended by R K lIer (1979 p 97) The molybdate reagent stock solution was preshypared by mixing 100 g ammonium molybdate [(NH)6 Mo 0Z44HzO] 470 ml concentrated NH4 0H (28 NH3) and water to make 10 1 of solution The final working molybdate reagent solution was prepared by

mixing 100 ml of the molybdate stock 200 ml of 15 M HZ S04 and 500 miDI water

This working molybdate reagent solution was measured out in 400-ml aliquots and placed in 500-ml volumetric flasks To determine dissolved silica 10 ml of the solution to be analyzed was added to the molybshydate reagent in the flask and the volume made up to 500 ml with 01 water The flask was then inverted and shaken three times and the color allowed to deshyvelop for 35 minutes (Shaking three times was found to improve the reproducibilitv of the results by ensurshying more complete mixing) The OPtical densitY was then measured at 400 nm using a dOUble-beam spectroshyphotometer (Perkin-Elmer 550)

To determine total silica a 10-ml aliquot was drawn and placed in a 50-ml plastic volumetric flask containing 10 ml of 10 N NaOH The flask was stopped shaken and set in a hot water bath for 10 minutes to allow the colloidal silica to be digested by the NaOH After the collidal silica had been digested 400 ml of the working molybdate solution and water to make the total volume up to 500 ml were added and the procedure continued as described above

In most of our experiments we treated the brine in one of two ways after 20 or 30 minutes had elapsed One treatment was to increase the pH by adding NaOH Another was to add a few mgmiddotlmiddotl of one of a number of commercial synthetic flocculants In either case the chemical addition was followed by a few minutes of rapid stirring After stirring was stopped pH was detershymined again and periodic sampling resumed

We also considered the possibilitv that the rapid stirring itself rather than the chemical addition that i~ accompanies is what causes the colloidal silica to flocculate This hypothesis was disproved by an expershyiment in which the brine was vigorously stirred for several minutes at the proper time without adding chemicals this treatment was found to have no significant effect on the amount of silica that remained in suspension at the end of the experiment

At the very end of the experiment the suspendshyed silica remaining in the brine was determined by an absolute method First about 150 ml of the brine was decanted into a 200 ml beaker being careful to leave as much of the settled silica floc as possible beshyhind The decanted brine was held in a water bath at about 85degC for a few minutes to allow the small amount of coarse floc not removed by the first decantation step to settle out Using a preheated 1oo-ml pipette a 100-ml sample was thencarefully withdrawn and placed in a preheated pressure filter funnel Ordinarily a 4oo-nm

571

pore size polycarbonate membrane filter was used The filter funnel was then sealed and the fluid was driven through it with compressed nitrogen gas at about 13 bars gauge pressure

The filter was then removed and placed in a plasshytic beaker with 30 ml of 10 N NaOH and heated in a boiling water bath for 10 minutes After the silica was digested in this way the filter was rinsed into the beaker and the contents of the beaker were then rinsed into a 500 m volumetric flask containing 400 ml of the molybdate reagent for silica determination Apshypropriate calculations then gave us the residual conshycentration of suspended silica in milligrams per liter of the clarified synthetic brine

When implemented as described above this techshyniqUe can not be used to determine suspended silica concentrations greater than about 32 mgmiddot-1 because higher concentrations cause the spectrophotometer to go off scale This is why final suspended silica is reshyported as gt 32 in Tables 2 and 3 for some experiments However high concentrations of suspended silica were usuallY obvious from the amount of floc accumulatshyed on the filter In such cases the digested floc was diluted up to 50 ml with 01 water and then a l-ml aliquot of this diluted solution was withdrawn and the concentration of silica in it was determined in the usual way This extra dilution step had the effect of extending the range of the method upward by a facshytor of 50

In cases in which the brine treatment was apparshyently successful the final suspended silica values were usually in the range of 2 to 5 mgmiddot-1 These suspiciously repetitious values may be an artifact caused by slightly stirring up the floc during the second decantation step Therefore the amount of suspended silica remaining in the brine after a better executed settling process (as in a properly designed settling tank in a pilot or commercial-size brine treatment facility) may prove actually to be less than 3 mgI-1 all else being equal

Of coursemiddot this method can only determine the concentration of particles and flocs large enough to be stopped by a 400-nm filter In practice this probshyably means all particles and flocs whose greatest linshyear dimension exceeds some size considerably smaller than 400 nm Although the elemental colloidal particles are probably only between 20 and 100 nm in diameter under these conditions the strongly flocculating nature of the treated brine probably ensures that most of the suspended silica is in the form of flocs large enough to be stopped by a 400-nm filter Going to a finer filter did not seem warranted in the case of the laboratory

tests but probably would be in the case of pilot plant experiments

Sometimes we also determined the floc setting rate This was done by remixing the floc with the brine after the filtration procedure quickly pouring the mixture into a graduated cylinder and measuring the rate at which the floc settled with a stopwatch The values obtained in this way were only approximate because the rate of sedimentation was strongly affected by conshyvection currents and the like

SILICA REMOVAL BY INCREASING pH

It is well established that increasing pH can destabilshyize colloidal silica suspended in salt solutions and cause it to flocculate or coagulate (Allen and Matijevic 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et aI 1980b Sections 221 and 22) Increasing the pH increases the negative charge on the surface of the particles and cations are adsorbed These adsorbed cations serve as bridges between the particles and this bridging effect is what causes the particles to adhere and flocculate (Coagulate is probably more appropriate than floccushylate in this particular case but we use flocculate throughout this paper to be consistent with common usage in the water treatment field) The flocculating ability of cations increases rapidly with their charge In Cerro Prieto brines calcium is probably the main flocculatshying ion

150r-----------------------------

95degC O33gl-1 Co+2

~ 0 100 E-~

-0 ltI) -0 C ltI) oQ Ill

(j)

OL-__-L____L-__-L____~__~____~

72 74 76 78

oH XBL 799middot2957

Figure 1 Suspended silica vs pH The pH of this synthetic brine was progressively increased by adding NaOH and the suspended silica determined after each inshycrease

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

568

most likely to precipitate and cause formation damshyage after reinjection This possibility was studied theoretshyically The results of this work are reported elsewhere (Iglesias and Weres 1980)

The contents of this paper and that by Iglesias and Weres (1980) are included in the report by Weres (1980) are included in the report by Weres et al (1980a) which also describes some additional work not discussed here

Calgon Magnifloc Nalgene Purifloc Separan Teflon and Tygon are all registered trademarks although they are not always identified as being such in the text and figures

REAL BRINES AND SYNTHETIC BRINES

We studied the precipitation of amorphous silica from Cerro Prieto brines by preparing synthetic brines that closely resemble them in our laboratory The approxishyrnate chemical compositions of two such synthetic brines which were extensively used in the silica removal studies are presented in T abl e 1 The silica concentrations and pH vallJes shown in the table were those we used most often However some work was also done at different silica concentrations and pHs The concentrations of all other components were always approximately as shown in the table (The concentratlon of sodium varshy

ied slightly with the silica concentration and pH Also evaporation tended to concentrate the brines slightly during the experiments)

These synthetic brines were formulated to have concentrations of their major components in the same range as real Cerro Prieto brines that have been flashshyed down to atmospheric pressure The high-Ca forshymulation was more or less arbitrarily chosen to resemble a typical flashed brine at Cerro Prieto

The low-Can formulation approximately corshyresponds to a mixture of the waste brine streams from Cerro Prieto wells M-8 M-27 M-31 M-35 and M-46 This composition was chosen because the waste brine line shared by these wells is the source of brine used for most pilot plant work at Cerro Prieto Chemical components that we consider unlikely to directly affect the chemical behavior of the silica in the brine under the conditions of greatest interest were left out of our synthetic brines

The pH value 73 which we chose for the synshytheticl)rines is typical of the pH values of freshly flashed brine specimens determined in the field with still~hot brine These range between about 70 and 76 and are most often about 73 or 74 (private communication from A Manon M and our own field measurements)

TABLE 1 NATURAL AND SYNTHETIC CERRO PRIETO BRINES

Natural brines Synthetic brines Well i4 Well 30 Low Ca

mgl-l mmole mg I-I mmolel- l mg mg I-I mmolemiddotl

Na 7079 308 7809 340 6970 333 6610 288 K 1439 37 1833 47 1502 38 1436 37 Ca 445 11 596 15 501 12 329 8 CI 13113 370 15173 428 12940 365 12089 341 Si02 960 16 1077 18 1000 17 1094 18 B 18 17 175 16 11 10 131 12 F 24 013 13 007 Sulfate 11 011 13 014 Carbonate + 50 082 35 057

Bicarbonate As 15 002 047 0006 Ba 107 0078 80 0058 Cu lt005 lt0001 i 002 Fe 08 0014 05 0009 Mn 02 0004 26 0047 Sr 176 020 188 021 Zn 001 00002 001 00002 Barbital buffer 20 20 pH 72-74 72-74 T 950 C 95 0 C

Values quoted and recalculated from Cosner and Apps (1978) Records 68 and 106 This data was generated by the CFE laboratory at Cerro Prieto and kindly provided to lBl by A Maii6n M

569

Also shown are typical analyses of flashed brine from wells M-14 and M-30 We used brine from these wells in our limited field work with freshly flashed brines at Cerro Prieto

The brines at Cerro Prieto are weakly buffered The major buffer at temperatures near 100degC and pH values between about 72 and 95 is monosilicic acid Si(OHk (Henceforth monosilicic acid will be abbreshyviated as MSA) At lower pH values the carbon dioxshyide-bicarbonate buffer system is dominant The nature of our experiments was such that it would have been very hard for us to adjust and control the pH of our synthetic brines by relying on these buffer systems alone Therefore we included a small amount of the buffering compound barbital (5 5-diethylbarbituric acid) in our synthetic brine formulation The pka of barbital is about 74 Bicarbonate was usually left out of the synthetic brines despite being a relatively imshyportant real brine component because it has no direct effect upon the behavior of silica and its presence would have complicated pH adjustment Boric acid was includshyed because we know that it can have mild inhibiting effect upon silica polymerization under some conditions (see Weres et aI 1980b Section 315)

In all experiments the artificial brine was conshystituted and silica polymerization initiated by mixing together two preheated solutions one acid and one alkashyline During the experiments the brine temperature was always between about 92 and 97degC

SILICA REMOVAL STUDIES EXPERIMENTAL METHODS

In these experiments the alkaline solution consisted of a premeasured amount of dilute sodium metasilshyicate sOlutiOn The acid solution contained the apprOshypriate amOunts Of the majOr compOnent salts sOdium barbital acid to adjust the buffer pH and neutralize the silica and water to make up the tOtal desired volshyume A tOtal volume Of 300 ml was usually employed

ThrOughout Our wOrk with simulated brines all vOlumes were measured at rOOm temperature ThereshyfOre when we say 300 ml we mean that the vOlume WOuid be that if the solutiOn were coOled down to room temperature When applied to salt concentrations and initial silica cOncentrations units Of moles or grams per liter are alsO used in this sense Because of the reshylatively low salt content of these solutions and the semishyquantitative nature of mOst of the synthetic brine data this is for all practical purposes the same as moles or grams per kilogram of water

The sOdium metasilicate stock solution contained about 2 gI-1 Si02 Because it was hard to make up

this solution to an exact predetermined concentration it was always standardized by comparison with a standshyard silica solution (Fischer AA standard)

The desired amount of sodium metasilicate sOshylution was put into a plastic bottle some of the remainshying air squeezed out and the cap closed tightly It was then put into a boiling water bath to preheat (This procedure ensured that the bottle would neither rupshyture nor lose water when heated) A separate 50-ml aliqUot of the sodium metasilicate solution was titratshyed with 05 M HCI using a pH meter to determine the volume of HCI solution needed to neutralize the given amount of sodium metasilicate sOlution

The acid starting solutiOn was prepared from a stock solution of concentrated brine a stock SQshy

lutiOn of sodium barbital 05 M HCI and water The concentrated brine was a mixture of sodium potassium and calcium chlOrides and boric acid in water formushylated SO that putting 3 parts of it into 20 parts of the final synthetic brine would give the desired final mashyjor component concentrations For example in the case of the high Ca synthetic brine presented in Table 1 the concentration brine cOntained 1667 M NaCI0256 M KCI 00833 M CaCI 2 and 00066 M H3 B03 The conshycentrations of the last three components in this concenshytrate are simply 203 times their desired concentrations in the synthetic brine The concentration of NaCI is somewhat smaller because the sodium metasilicate and buffer stock solutions also contain sodium

In the case of the Low Ca synthetic brine the concentrated brine contained 1547 M NaCI 0245 M KCI00547MCaClzandO008l M H3BOa

The barbital buffer stock solution was prepared to be 05 M in barbital and 05 M in NaOH Purified grade Mallinckrodt barbital intended for pharmaceushytical use was employed (All chemicals used other than the barbital and the commercial flocculants were of reagent grade)

To prepare the acid solution measured amounts of the brine concentrate buffer stock solution and the amount of water needed to bring the final volume up to that desired were mixed in a beaker suspended in a boiling water bath The beaker was temporarily covered with a watchglass and its cOntents allowed to heat up to above about 90degC Next the mixture was titrated with 05 M HCI to adjust the pH value to that desired in the synthetic brine Finally the amount of additional 05 M HCI needed to neutralize the sodium metasilicate solution as determined by the earlier tishytration was added to the beaker and the resulting mixshyture was again allowed to heat up

570

To start the reaction the preheated alkaline soshylution was poured into the beaker containing the preshyheated acid solution with rapid stirring The stirring was stopped after a few seconds the pH measured and periodic 1-ml samples withdrawn for analysis from then on

Both dissolved silica and total silica were detershymined for by the molybdate yellow method To deshytermine the molybdate active (ie dissolved) silica the molybdate yellow method was employed directly To determine the total silica (Le including the susshypended colloidal silica) the specimen was first reacshyted with NaOH to digest and dissolve the colloidal sishylica (Henceforth molybdate active silica will be abbreviated as MAS which should not be confused with MSA MSA is molybdate active but so are some other small molecular and ionic silica species)

The difference between the empirically detershymined molybdate active and total silica concentrations is approximately equal to the amount of colloidal sishylica suspended in the brine This differential techshynique is sensitive down to about 10 mgI-1 suspended silica

The measured silica concentrations reported in the figures and tables were not corrected for the effect of temperature Therefore the measured silica concenshytrations really are in units of gp at the stated tempershyature To approximately convert these values to units of grams (kgHz0)-1 multiply them by 105

The sampling syringe was always preheated by rapidly drawing brine in and out of it once or twice before actually taking the sample to be analyzed

The molybdate yellow method is subject to seshyrious interference by the traces of dissolved silica and inorganic phosphate that are usually present in tap Washyter To eliminate this problem all solutions were preshypared using water that was twice doubly deionized (01) with mixed anioniccationic ion exchange resins The water coming out of our laboratorys 01 water tap was already once doubly deionized as a precaution we sent it through our own column filled with mixed resshyins for the second double deionization step

As in our previous work we used the molybdate yellow procedure recommended by R K lIer (1979 p 97) The molybdate reagent stock solution was preshypared by mixing 100 g ammonium molybdate [(NH)6 Mo 0Z44HzO] 470 ml concentrated NH4 0H (28 NH3) and water to make 10 1 of solution The final working molybdate reagent solution was prepared by

mixing 100 ml of the molybdate stock 200 ml of 15 M HZ S04 and 500 miDI water

This working molybdate reagent solution was measured out in 400-ml aliquots and placed in 500-ml volumetric flasks To determine dissolved silica 10 ml of the solution to be analyzed was added to the molybshydate reagent in the flask and the volume made up to 500 ml with 01 water The flask was then inverted and shaken three times and the color allowed to deshyvelop for 35 minutes (Shaking three times was found to improve the reproducibilitv of the results by ensurshying more complete mixing) The OPtical densitY was then measured at 400 nm using a dOUble-beam spectroshyphotometer (Perkin-Elmer 550)

To determine total silica a 10-ml aliquot was drawn and placed in a 50-ml plastic volumetric flask containing 10 ml of 10 N NaOH The flask was stopped shaken and set in a hot water bath for 10 minutes to allow the colloidal silica to be digested by the NaOH After the collidal silica had been digested 400 ml of the working molybdate solution and water to make the total volume up to 500 ml were added and the procedure continued as described above

In most of our experiments we treated the brine in one of two ways after 20 or 30 minutes had elapsed One treatment was to increase the pH by adding NaOH Another was to add a few mgmiddotlmiddotl of one of a number of commercial synthetic flocculants In either case the chemical addition was followed by a few minutes of rapid stirring After stirring was stopped pH was detershymined again and periodic sampling resumed

We also considered the possibilitv that the rapid stirring itself rather than the chemical addition that i~ accompanies is what causes the colloidal silica to flocculate This hypothesis was disproved by an expershyiment in which the brine was vigorously stirred for several minutes at the proper time without adding chemicals this treatment was found to have no significant effect on the amount of silica that remained in suspension at the end of the experiment

At the very end of the experiment the suspendshyed silica remaining in the brine was determined by an absolute method First about 150 ml of the brine was decanted into a 200 ml beaker being careful to leave as much of the settled silica floc as possible beshyhind The decanted brine was held in a water bath at about 85degC for a few minutes to allow the small amount of coarse floc not removed by the first decantation step to settle out Using a preheated 1oo-ml pipette a 100-ml sample was thencarefully withdrawn and placed in a preheated pressure filter funnel Ordinarily a 4oo-nm

571

pore size polycarbonate membrane filter was used The filter funnel was then sealed and the fluid was driven through it with compressed nitrogen gas at about 13 bars gauge pressure

The filter was then removed and placed in a plasshytic beaker with 30 ml of 10 N NaOH and heated in a boiling water bath for 10 minutes After the silica was digested in this way the filter was rinsed into the beaker and the contents of the beaker were then rinsed into a 500 m volumetric flask containing 400 ml of the molybdate reagent for silica determination Apshypropriate calculations then gave us the residual conshycentration of suspended silica in milligrams per liter of the clarified synthetic brine

When implemented as described above this techshyniqUe can not be used to determine suspended silica concentrations greater than about 32 mgmiddot-1 because higher concentrations cause the spectrophotometer to go off scale This is why final suspended silica is reshyported as gt 32 in Tables 2 and 3 for some experiments However high concentrations of suspended silica were usuallY obvious from the amount of floc accumulatshyed on the filter In such cases the digested floc was diluted up to 50 ml with 01 water and then a l-ml aliquot of this diluted solution was withdrawn and the concentration of silica in it was determined in the usual way This extra dilution step had the effect of extending the range of the method upward by a facshytor of 50

In cases in which the brine treatment was apparshyently successful the final suspended silica values were usually in the range of 2 to 5 mgmiddot-1 These suspiciously repetitious values may be an artifact caused by slightly stirring up the floc during the second decantation step Therefore the amount of suspended silica remaining in the brine after a better executed settling process (as in a properly designed settling tank in a pilot or commercial-size brine treatment facility) may prove actually to be less than 3 mgI-1 all else being equal

Of coursemiddot this method can only determine the concentration of particles and flocs large enough to be stopped by a 400-nm filter In practice this probshyably means all particles and flocs whose greatest linshyear dimension exceeds some size considerably smaller than 400 nm Although the elemental colloidal particles are probably only between 20 and 100 nm in diameter under these conditions the strongly flocculating nature of the treated brine probably ensures that most of the suspended silica is in the form of flocs large enough to be stopped by a 400-nm filter Going to a finer filter did not seem warranted in the case of the laboratory

tests but probably would be in the case of pilot plant experiments

Sometimes we also determined the floc setting rate This was done by remixing the floc with the brine after the filtration procedure quickly pouring the mixture into a graduated cylinder and measuring the rate at which the floc settled with a stopwatch The values obtained in this way were only approximate because the rate of sedimentation was strongly affected by conshyvection currents and the like

SILICA REMOVAL BY INCREASING pH

It is well established that increasing pH can destabilshyize colloidal silica suspended in salt solutions and cause it to flocculate or coagulate (Allen and Matijevic 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et aI 1980b Sections 221 and 22) Increasing the pH increases the negative charge on the surface of the particles and cations are adsorbed These adsorbed cations serve as bridges between the particles and this bridging effect is what causes the particles to adhere and flocculate (Coagulate is probably more appropriate than floccushylate in this particular case but we use flocculate throughout this paper to be consistent with common usage in the water treatment field) The flocculating ability of cations increases rapidly with their charge In Cerro Prieto brines calcium is probably the main flocculatshying ion

150r-----------------------------

95degC O33gl-1 Co+2

~ 0 100 E-~

-0 ltI) -0 C ltI) oQ Ill

(j)

OL-__-L____L-__-L____~__~____~

72 74 76 78

oH XBL 799middot2957

Figure 1 Suspended silica vs pH The pH of this synthetic brine was progressively increased by adding NaOH and the suspended silica determined after each inshycrease

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

569

Also shown are typical analyses of flashed brine from wells M-14 and M-30 We used brine from these wells in our limited field work with freshly flashed brines at Cerro Prieto

The brines at Cerro Prieto are weakly buffered The major buffer at temperatures near 100degC and pH values between about 72 and 95 is monosilicic acid Si(OHk (Henceforth monosilicic acid will be abbreshyviated as MSA) At lower pH values the carbon dioxshyide-bicarbonate buffer system is dominant The nature of our experiments was such that it would have been very hard for us to adjust and control the pH of our synthetic brines by relying on these buffer systems alone Therefore we included a small amount of the buffering compound barbital (5 5-diethylbarbituric acid) in our synthetic brine formulation The pka of barbital is about 74 Bicarbonate was usually left out of the synthetic brines despite being a relatively imshyportant real brine component because it has no direct effect upon the behavior of silica and its presence would have complicated pH adjustment Boric acid was includshyed because we know that it can have mild inhibiting effect upon silica polymerization under some conditions (see Weres et aI 1980b Section 315)

In all experiments the artificial brine was conshystituted and silica polymerization initiated by mixing together two preheated solutions one acid and one alkashyline During the experiments the brine temperature was always between about 92 and 97degC

SILICA REMOVAL STUDIES EXPERIMENTAL METHODS

In these experiments the alkaline solution consisted of a premeasured amount of dilute sodium metasilshyicate sOlutiOn The acid solution contained the apprOshypriate amOunts Of the majOr compOnent salts sOdium barbital acid to adjust the buffer pH and neutralize the silica and water to make up the tOtal desired volshyume A tOtal volume Of 300 ml was usually employed

ThrOughout Our wOrk with simulated brines all vOlumes were measured at rOOm temperature ThereshyfOre when we say 300 ml we mean that the vOlume WOuid be that if the solutiOn were coOled down to room temperature When applied to salt concentrations and initial silica cOncentrations units Of moles or grams per liter are alsO used in this sense Because of the reshylatively low salt content of these solutions and the semishyquantitative nature of mOst of the synthetic brine data this is for all practical purposes the same as moles or grams per kilogram of water

The sOdium metasilicate stock solution contained about 2 gI-1 Si02 Because it was hard to make up

this solution to an exact predetermined concentration it was always standardized by comparison with a standshyard silica solution (Fischer AA standard)

The desired amount of sodium metasilicate sOshylution was put into a plastic bottle some of the remainshying air squeezed out and the cap closed tightly It was then put into a boiling water bath to preheat (This procedure ensured that the bottle would neither rupshyture nor lose water when heated) A separate 50-ml aliqUot of the sodium metasilicate solution was titratshyed with 05 M HCI using a pH meter to determine the volume of HCI solution needed to neutralize the given amount of sodium metasilicate sOlution

The acid starting solutiOn was prepared from a stock solution of concentrated brine a stock SQshy

lutiOn of sodium barbital 05 M HCI and water The concentrated brine was a mixture of sodium potassium and calcium chlOrides and boric acid in water formushylated SO that putting 3 parts of it into 20 parts of the final synthetic brine would give the desired final mashyjor component concentrations For example in the case of the high Ca synthetic brine presented in Table 1 the concentration brine cOntained 1667 M NaCI0256 M KCI 00833 M CaCI 2 and 00066 M H3 B03 The conshycentrations of the last three components in this concenshytrate are simply 203 times their desired concentrations in the synthetic brine The concentration of NaCI is somewhat smaller because the sodium metasilicate and buffer stock solutions also contain sodium

In the case of the Low Ca synthetic brine the concentrated brine contained 1547 M NaCI 0245 M KCI00547MCaClzandO008l M H3BOa

The barbital buffer stock solution was prepared to be 05 M in barbital and 05 M in NaOH Purified grade Mallinckrodt barbital intended for pharmaceushytical use was employed (All chemicals used other than the barbital and the commercial flocculants were of reagent grade)

To prepare the acid solution measured amounts of the brine concentrate buffer stock solution and the amount of water needed to bring the final volume up to that desired were mixed in a beaker suspended in a boiling water bath The beaker was temporarily covered with a watchglass and its cOntents allowed to heat up to above about 90degC Next the mixture was titrated with 05 M HCI to adjust the pH value to that desired in the synthetic brine Finally the amount of additional 05 M HCI needed to neutralize the sodium metasilicate solution as determined by the earlier tishytration was added to the beaker and the resulting mixshyture was again allowed to heat up

570

To start the reaction the preheated alkaline soshylution was poured into the beaker containing the preshyheated acid solution with rapid stirring The stirring was stopped after a few seconds the pH measured and periodic 1-ml samples withdrawn for analysis from then on

Both dissolved silica and total silica were detershymined for by the molybdate yellow method To deshytermine the molybdate active (ie dissolved) silica the molybdate yellow method was employed directly To determine the total silica (Le including the susshypended colloidal silica) the specimen was first reacshyted with NaOH to digest and dissolve the colloidal sishylica (Henceforth molybdate active silica will be abbreviated as MAS which should not be confused with MSA MSA is molybdate active but so are some other small molecular and ionic silica species)

The difference between the empirically detershymined molybdate active and total silica concentrations is approximately equal to the amount of colloidal sishylica suspended in the brine This differential techshynique is sensitive down to about 10 mgI-1 suspended silica

The measured silica concentrations reported in the figures and tables were not corrected for the effect of temperature Therefore the measured silica concenshytrations really are in units of gp at the stated tempershyature To approximately convert these values to units of grams (kgHz0)-1 multiply them by 105

The sampling syringe was always preheated by rapidly drawing brine in and out of it once or twice before actually taking the sample to be analyzed

The molybdate yellow method is subject to seshyrious interference by the traces of dissolved silica and inorganic phosphate that are usually present in tap Washyter To eliminate this problem all solutions were preshypared using water that was twice doubly deionized (01) with mixed anioniccationic ion exchange resins The water coming out of our laboratorys 01 water tap was already once doubly deionized as a precaution we sent it through our own column filled with mixed resshyins for the second double deionization step

As in our previous work we used the molybdate yellow procedure recommended by R K lIer (1979 p 97) The molybdate reagent stock solution was preshypared by mixing 100 g ammonium molybdate [(NH)6 Mo 0Z44HzO] 470 ml concentrated NH4 0H (28 NH3) and water to make 10 1 of solution The final working molybdate reagent solution was prepared by

mixing 100 ml of the molybdate stock 200 ml of 15 M HZ S04 and 500 miDI water

This working molybdate reagent solution was measured out in 400-ml aliquots and placed in 500-ml volumetric flasks To determine dissolved silica 10 ml of the solution to be analyzed was added to the molybshydate reagent in the flask and the volume made up to 500 ml with 01 water The flask was then inverted and shaken three times and the color allowed to deshyvelop for 35 minutes (Shaking three times was found to improve the reproducibilitv of the results by ensurshying more complete mixing) The OPtical densitY was then measured at 400 nm using a dOUble-beam spectroshyphotometer (Perkin-Elmer 550)

To determine total silica a 10-ml aliquot was drawn and placed in a 50-ml plastic volumetric flask containing 10 ml of 10 N NaOH The flask was stopped shaken and set in a hot water bath for 10 minutes to allow the colloidal silica to be digested by the NaOH After the collidal silica had been digested 400 ml of the working molybdate solution and water to make the total volume up to 500 ml were added and the procedure continued as described above

In most of our experiments we treated the brine in one of two ways after 20 or 30 minutes had elapsed One treatment was to increase the pH by adding NaOH Another was to add a few mgmiddotlmiddotl of one of a number of commercial synthetic flocculants In either case the chemical addition was followed by a few minutes of rapid stirring After stirring was stopped pH was detershymined again and periodic sampling resumed

We also considered the possibilitv that the rapid stirring itself rather than the chemical addition that i~ accompanies is what causes the colloidal silica to flocculate This hypothesis was disproved by an expershyiment in which the brine was vigorously stirred for several minutes at the proper time without adding chemicals this treatment was found to have no significant effect on the amount of silica that remained in suspension at the end of the experiment

At the very end of the experiment the suspendshyed silica remaining in the brine was determined by an absolute method First about 150 ml of the brine was decanted into a 200 ml beaker being careful to leave as much of the settled silica floc as possible beshyhind The decanted brine was held in a water bath at about 85degC for a few minutes to allow the small amount of coarse floc not removed by the first decantation step to settle out Using a preheated 1oo-ml pipette a 100-ml sample was thencarefully withdrawn and placed in a preheated pressure filter funnel Ordinarily a 4oo-nm

571

pore size polycarbonate membrane filter was used The filter funnel was then sealed and the fluid was driven through it with compressed nitrogen gas at about 13 bars gauge pressure

The filter was then removed and placed in a plasshytic beaker with 30 ml of 10 N NaOH and heated in a boiling water bath for 10 minutes After the silica was digested in this way the filter was rinsed into the beaker and the contents of the beaker were then rinsed into a 500 m volumetric flask containing 400 ml of the molybdate reagent for silica determination Apshypropriate calculations then gave us the residual conshycentration of suspended silica in milligrams per liter of the clarified synthetic brine

When implemented as described above this techshyniqUe can not be used to determine suspended silica concentrations greater than about 32 mgmiddot-1 because higher concentrations cause the spectrophotometer to go off scale This is why final suspended silica is reshyported as gt 32 in Tables 2 and 3 for some experiments However high concentrations of suspended silica were usuallY obvious from the amount of floc accumulatshyed on the filter In such cases the digested floc was diluted up to 50 ml with 01 water and then a l-ml aliquot of this diluted solution was withdrawn and the concentration of silica in it was determined in the usual way This extra dilution step had the effect of extending the range of the method upward by a facshytor of 50

In cases in which the brine treatment was apparshyently successful the final suspended silica values were usually in the range of 2 to 5 mgmiddot-1 These suspiciously repetitious values may be an artifact caused by slightly stirring up the floc during the second decantation step Therefore the amount of suspended silica remaining in the brine after a better executed settling process (as in a properly designed settling tank in a pilot or commercial-size brine treatment facility) may prove actually to be less than 3 mgI-1 all else being equal

Of coursemiddot this method can only determine the concentration of particles and flocs large enough to be stopped by a 400-nm filter In practice this probshyably means all particles and flocs whose greatest linshyear dimension exceeds some size considerably smaller than 400 nm Although the elemental colloidal particles are probably only between 20 and 100 nm in diameter under these conditions the strongly flocculating nature of the treated brine probably ensures that most of the suspended silica is in the form of flocs large enough to be stopped by a 400-nm filter Going to a finer filter did not seem warranted in the case of the laboratory

tests but probably would be in the case of pilot plant experiments

Sometimes we also determined the floc setting rate This was done by remixing the floc with the brine after the filtration procedure quickly pouring the mixture into a graduated cylinder and measuring the rate at which the floc settled with a stopwatch The values obtained in this way were only approximate because the rate of sedimentation was strongly affected by conshyvection currents and the like

SILICA REMOVAL BY INCREASING pH

It is well established that increasing pH can destabilshyize colloidal silica suspended in salt solutions and cause it to flocculate or coagulate (Allen and Matijevic 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et aI 1980b Sections 221 and 22) Increasing the pH increases the negative charge on the surface of the particles and cations are adsorbed These adsorbed cations serve as bridges between the particles and this bridging effect is what causes the particles to adhere and flocculate (Coagulate is probably more appropriate than floccushylate in this particular case but we use flocculate throughout this paper to be consistent with common usage in the water treatment field) The flocculating ability of cations increases rapidly with their charge In Cerro Prieto brines calcium is probably the main flocculatshying ion

150r-----------------------------

95degC O33gl-1 Co+2

~ 0 100 E-~

-0 ltI) -0 C ltI) oQ Ill

(j)

OL-__-L____L-__-L____~__~____~

72 74 76 78

oH XBL 799middot2957

Figure 1 Suspended silica vs pH The pH of this synthetic brine was progressively increased by adding NaOH and the suspended silica determined after each inshycrease

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

570

To start the reaction the preheated alkaline soshylution was poured into the beaker containing the preshyheated acid solution with rapid stirring The stirring was stopped after a few seconds the pH measured and periodic 1-ml samples withdrawn for analysis from then on

Both dissolved silica and total silica were detershymined for by the molybdate yellow method To deshytermine the molybdate active (ie dissolved) silica the molybdate yellow method was employed directly To determine the total silica (Le including the susshypended colloidal silica) the specimen was first reacshyted with NaOH to digest and dissolve the colloidal sishylica (Henceforth molybdate active silica will be abbreviated as MAS which should not be confused with MSA MSA is molybdate active but so are some other small molecular and ionic silica species)

The difference between the empirically detershymined molybdate active and total silica concentrations is approximately equal to the amount of colloidal sishylica suspended in the brine This differential techshynique is sensitive down to about 10 mgI-1 suspended silica

The measured silica concentrations reported in the figures and tables were not corrected for the effect of temperature Therefore the measured silica concenshytrations really are in units of gp at the stated tempershyature To approximately convert these values to units of grams (kgHz0)-1 multiply them by 105

The sampling syringe was always preheated by rapidly drawing brine in and out of it once or twice before actually taking the sample to be analyzed

The molybdate yellow method is subject to seshyrious interference by the traces of dissolved silica and inorganic phosphate that are usually present in tap Washyter To eliminate this problem all solutions were preshypared using water that was twice doubly deionized (01) with mixed anioniccationic ion exchange resins The water coming out of our laboratorys 01 water tap was already once doubly deionized as a precaution we sent it through our own column filled with mixed resshyins for the second double deionization step

As in our previous work we used the molybdate yellow procedure recommended by R K lIer (1979 p 97) The molybdate reagent stock solution was preshypared by mixing 100 g ammonium molybdate [(NH)6 Mo 0Z44HzO] 470 ml concentrated NH4 0H (28 NH3) and water to make 10 1 of solution The final working molybdate reagent solution was prepared by

mixing 100 ml of the molybdate stock 200 ml of 15 M HZ S04 and 500 miDI water

This working molybdate reagent solution was measured out in 400-ml aliquots and placed in 500-ml volumetric flasks To determine dissolved silica 10 ml of the solution to be analyzed was added to the molybshydate reagent in the flask and the volume made up to 500 ml with 01 water The flask was then inverted and shaken three times and the color allowed to deshyvelop for 35 minutes (Shaking three times was found to improve the reproducibilitv of the results by ensurshying more complete mixing) The OPtical densitY was then measured at 400 nm using a dOUble-beam spectroshyphotometer (Perkin-Elmer 550)

To determine total silica a 10-ml aliquot was drawn and placed in a 50-ml plastic volumetric flask containing 10 ml of 10 N NaOH The flask was stopped shaken and set in a hot water bath for 10 minutes to allow the colloidal silica to be digested by the NaOH After the collidal silica had been digested 400 ml of the working molybdate solution and water to make the total volume up to 500 ml were added and the procedure continued as described above

In most of our experiments we treated the brine in one of two ways after 20 or 30 minutes had elapsed One treatment was to increase the pH by adding NaOH Another was to add a few mgmiddotlmiddotl of one of a number of commercial synthetic flocculants In either case the chemical addition was followed by a few minutes of rapid stirring After stirring was stopped pH was detershymined again and periodic sampling resumed

We also considered the possibilitv that the rapid stirring itself rather than the chemical addition that i~ accompanies is what causes the colloidal silica to flocculate This hypothesis was disproved by an expershyiment in which the brine was vigorously stirred for several minutes at the proper time without adding chemicals this treatment was found to have no significant effect on the amount of silica that remained in suspension at the end of the experiment

At the very end of the experiment the suspendshyed silica remaining in the brine was determined by an absolute method First about 150 ml of the brine was decanted into a 200 ml beaker being careful to leave as much of the settled silica floc as possible beshyhind The decanted brine was held in a water bath at about 85degC for a few minutes to allow the small amount of coarse floc not removed by the first decantation step to settle out Using a preheated 1oo-ml pipette a 100-ml sample was thencarefully withdrawn and placed in a preheated pressure filter funnel Ordinarily a 4oo-nm

571

pore size polycarbonate membrane filter was used The filter funnel was then sealed and the fluid was driven through it with compressed nitrogen gas at about 13 bars gauge pressure

The filter was then removed and placed in a plasshytic beaker with 30 ml of 10 N NaOH and heated in a boiling water bath for 10 minutes After the silica was digested in this way the filter was rinsed into the beaker and the contents of the beaker were then rinsed into a 500 m volumetric flask containing 400 ml of the molybdate reagent for silica determination Apshypropriate calculations then gave us the residual conshycentration of suspended silica in milligrams per liter of the clarified synthetic brine

When implemented as described above this techshyniqUe can not be used to determine suspended silica concentrations greater than about 32 mgmiddot-1 because higher concentrations cause the spectrophotometer to go off scale This is why final suspended silica is reshyported as gt 32 in Tables 2 and 3 for some experiments However high concentrations of suspended silica were usuallY obvious from the amount of floc accumulatshyed on the filter In such cases the digested floc was diluted up to 50 ml with 01 water and then a l-ml aliquot of this diluted solution was withdrawn and the concentration of silica in it was determined in the usual way This extra dilution step had the effect of extending the range of the method upward by a facshytor of 50

In cases in which the brine treatment was apparshyently successful the final suspended silica values were usually in the range of 2 to 5 mgmiddot-1 These suspiciously repetitious values may be an artifact caused by slightly stirring up the floc during the second decantation step Therefore the amount of suspended silica remaining in the brine after a better executed settling process (as in a properly designed settling tank in a pilot or commercial-size brine treatment facility) may prove actually to be less than 3 mgI-1 all else being equal

Of coursemiddot this method can only determine the concentration of particles and flocs large enough to be stopped by a 400-nm filter In practice this probshyably means all particles and flocs whose greatest linshyear dimension exceeds some size considerably smaller than 400 nm Although the elemental colloidal particles are probably only between 20 and 100 nm in diameter under these conditions the strongly flocculating nature of the treated brine probably ensures that most of the suspended silica is in the form of flocs large enough to be stopped by a 400-nm filter Going to a finer filter did not seem warranted in the case of the laboratory

tests but probably would be in the case of pilot plant experiments

Sometimes we also determined the floc setting rate This was done by remixing the floc with the brine after the filtration procedure quickly pouring the mixture into a graduated cylinder and measuring the rate at which the floc settled with a stopwatch The values obtained in this way were only approximate because the rate of sedimentation was strongly affected by conshyvection currents and the like

SILICA REMOVAL BY INCREASING pH

It is well established that increasing pH can destabilshyize colloidal silica suspended in salt solutions and cause it to flocculate or coagulate (Allen and Matijevic 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et aI 1980b Sections 221 and 22) Increasing the pH increases the negative charge on the surface of the particles and cations are adsorbed These adsorbed cations serve as bridges between the particles and this bridging effect is what causes the particles to adhere and flocculate (Coagulate is probably more appropriate than floccushylate in this particular case but we use flocculate throughout this paper to be consistent with common usage in the water treatment field) The flocculating ability of cations increases rapidly with their charge In Cerro Prieto brines calcium is probably the main flocculatshying ion

150r-----------------------------

95degC O33gl-1 Co+2

~ 0 100 E-~

-0 ltI) -0 C ltI) oQ Ill

(j)

OL-__-L____L-__-L____~__~____~

72 74 76 78

oH XBL 799middot2957

Figure 1 Suspended silica vs pH The pH of this synthetic brine was progressively increased by adding NaOH and the suspended silica determined after each inshycrease

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

571

pore size polycarbonate membrane filter was used The filter funnel was then sealed and the fluid was driven through it with compressed nitrogen gas at about 13 bars gauge pressure

The filter was then removed and placed in a plasshytic beaker with 30 ml of 10 N NaOH and heated in a boiling water bath for 10 minutes After the silica was digested in this way the filter was rinsed into the beaker and the contents of the beaker were then rinsed into a 500 m volumetric flask containing 400 ml of the molybdate reagent for silica determination Apshypropriate calculations then gave us the residual conshycentration of suspended silica in milligrams per liter of the clarified synthetic brine

When implemented as described above this techshyniqUe can not be used to determine suspended silica concentrations greater than about 32 mgmiddot-1 because higher concentrations cause the spectrophotometer to go off scale This is why final suspended silica is reshyported as gt 32 in Tables 2 and 3 for some experiments However high concentrations of suspended silica were usuallY obvious from the amount of floc accumulatshyed on the filter In such cases the digested floc was diluted up to 50 ml with 01 water and then a l-ml aliquot of this diluted solution was withdrawn and the concentration of silica in it was determined in the usual way This extra dilution step had the effect of extending the range of the method upward by a facshytor of 50

In cases in which the brine treatment was apparshyently successful the final suspended silica values were usually in the range of 2 to 5 mgmiddot-1 These suspiciously repetitious values may be an artifact caused by slightly stirring up the floc during the second decantation step Therefore the amount of suspended silica remaining in the brine after a better executed settling process (as in a properly designed settling tank in a pilot or commercial-size brine treatment facility) may prove actually to be less than 3 mgI-1 all else being equal

Of coursemiddot this method can only determine the concentration of particles and flocs large enough to be stopped by a 400-nm filter In practice this probshyably means all particles and flocs whose greatest linshyear dimension exceeds some size considerably smaller than 400 nm Although the elemental colloidal particles are probably only between 20 and 100 nm in diameter under these conditions the strongly flocculating nature of the treated brine probably ensures that most of the suspended silica is in the form of flocs large enough to be stopped by a 400-nm filter Going to a finer filter did not seem warranted in the case of the laboratory

tests but probably would be in the case of pilot plant experiments

Sometimes we also determined the floc setting rate This was done by remixing the floc with the brine after the filtration procedure quickly pouring the mixture into a graduated cylinder and measuring the rate at which the floc settled with a stopwatch The values obtained in this way were only approximate because the rate of sedimentation was strongly affected by conshyvection currents and the like

SILICA REMOVAL BY INCREASING pH

It is well established that increasing pH can destabilshyize colloidal silica suspended in salt solutions and cause it to flocculate or coagulate (Allen and Matijevic 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et aI 1980b Sections 221 and 22) Increasing the pH increases the negative charge on the surface of the particles and cations are adsorbed These adsorbed cations serve as bridges between the particles and this bridging effect is what causes the particles to adhere and flocculate (Coagulate is probably more appropriate than floccushylate in this particular case but we use flocculate throughout this paper to be consistent with common usage in the water treatment field) The flocculating ability of cations increases rapidly with their charge In Cerro Prieto brines calcium is probably the main flocculatshying ion

150r-----------------------------

95degC O33gl-1 Co+2

~ 0 100 E-~

-0 ltI) -0 C ltI) oQ Ill

(j)

OL-__-L____L-__-L____~__~____~

72 74 76 78

oH XBL 799middot2957

Figure 1 Suspended silica vs pH The pH of this synthetic brine was progressively increased by adding NaOH and the suspended silica determined after each inshycrease

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

bull bull bull bull

572

I n flashed Cerro Prieto brines the conversion of dissolved silica to colloidal silica is rapid but removal of the collidal silica from the brine by flocculation and settling is slow and incomplete Therefore the mashyjor task of preinjection brine treatment is to induce rapid flocculation and settl ing Because Cerro Prieto brine contains a substantial amount of calcium increasshying the pH recommends itself as the simplest way to induce rapid flocculation

Figure 1 semiquantitatively illustrates the effect of increasing the pH In this experiment the synthetshyic brine was originally constituted with a pH of about 73 and allowed to sit undisturbed for 25 minutes to let the concentration for MAS to drop to a nearly steadyshystate value Next the brine was briefly stirred the floc was allowed to settle for 2 minutes and the first two l-ml aliquots were withdrawn to determine the conshycentration of suspended silica by the differential method Then a small amount of 1 N NaOH was added to inshycrease the pH After brief additional stirring the pH was measured the floc was allowed to settle for 2 minshyutes without stirring and two more aliquots were withshydrawn for the determination of suspended silica This cycle was repeated four more times to generate the rest of the points

The data in Figure 1 are only semiquantitative in that the results obtained from such experiments are greatly affected by fine points of experimental design such as the duration of settling and the size and shape of the reaction vessel However it is obvious from Figshyure 1 that at pH 73 flocculation is poor and that inmiddot

[ I Add 95degC clgl-1

bull - NaOH I I I l1 degC-o--S_ --0 t I 0 Total Si02~ 1bull I 0 Molybdate

I I active Si02 I r 0 I I 05 gl -I Co+2 1

CJl

I Final suspended

deg 0 10

f I bull I

l I

I I

05 bull I ~o 01

I 9-- fI

bull il -e bull

bull I ~

00 pH~ 72 I 00 pH -78 J pH-73 i 33mgl-1

XBL 7992976

Figure 2 Synthetic brine flocculation experiments Solid symbols brine not treated Open symbols pH increased after 22 minutes by adding NaOH In this and further similar figures silica concentrations are not corrected for temperature effect

creasing the pH to about 78 removes most of the colshyloidal silica from suspension

The full detailed data obtained in two typical brine treatment experiments are presented in Figmiddot ure 2 In one of them (the solid symbols) the synthetic brine had an initial pH of about 73 and was maintained that way throughout the duration of the experiment In the other (open symbols) the synthetic brine inishytially had a pH of about 72 and the pH was increased by adding a small amount of 1 N NaOH after 22 minshyutes Except for the change in pH value and brief stirshyring in the second experiment the two experiments were the same The high Ca synthetic brine formulamiddot tion was used in both

In both cases a substantial fraction of the silimiddot ca had already polymerized by the time the first samshyples were taken for analysis At this time there was also a faint white haze in the brine indicating the beshyginning of flocculation (Unflocculated colloidal silishyca is invisible to the naked eye at these concentrations) The very rapid initial stage of the polymerizatio~ reacshytion had been completed by the time of the second point which was taken 5 minutes after the first This means that a 5-minute aging time would probably suffice to polymerize the silica in a brine like this inshystead of the 22 minutes allowed in this experiment

The 20middotminute total silica values show that by that time the colloidal silica had flocculated to the point that it was beginning to settle Flocculation and settling were slightly more pronounced in the synthetshyic brine that had the higher initial pH

At 25 minutes the MAS concentration in the unmodified brine had already fallen to a limiting value This value is somewhat higher than the equilibshyrium solubility level for bulk amorphous silica under these conditions which is about 033 gmiddotI-1 (See Weres et aI 1980b Sections 311 and 318 for detailed proshycedures to calculate the solubility of amorphous silishyca under various conditions) That the value observed is higher than the bulk solubility value is due to the small size of the colloidal silica particles that are creatshyed under these conditions of rapid nucleation Small particles are more soluble than bulk amorphous silica (AS) and their solubility determines the limiting MAS concentration

Up to the point that the pH of the treated brine was increased the MAS concentrations in the treated and untreated brines were equal However inshycreasing the pH and stirring noticeably increased the MAS concentration in the treated brine This is beshycause increasing the pH increases the solubility of amorshy

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

573

phous silica from about 033 to about 035 gI-1 in this case That the post treatment IIIIAS concentration was slightly above the corresponding equilibrium solushybility value for bulk AS was again due to the greater sol ubility of small particles If real the apparent slow increase of MAS concentration in the modified brine toward the end of the experiment is probably due to continuing reequilibration of the colloidal silica with its chemically modified environment

The partial redissolution of the colloidal silica during and perhaps after the flocculation process is a desirable phenomenon from the practical point of view It means that if the clarified brine is separated from the floc quickly enough the small amount of colloidal silica remaining in the brine will probably reshydissolve completely if given enough time to do so The final value of 33 mgl-l suspended Si02 was detershymined by the filtration technique As discussed above even this small value may be due to the difficulty of achieving good settling in a small container rather than to imperfect flocculation

The MAS concentration at the end of this expershyiment was about 042 gl-l which is still above the equishylibrium solubility of silica under the given conditions This means that the brine will slowly deposit vitreous silica on any surface with which it comes into contact Our best estimate of the molecular deposition rate as calculated using the methods presented by Weres

et al (1980b Section 314) is about 5 Jtmmiddotyr-1 bull This mayor may not be small enough to avoid injectivity decline over the life of an injection well depending on the pore size in the receiving formation and the exshytent of fracture permeability It would not effect fracture permeability but might damage pore permeability if the pores are small enough

Other experiments have shown that the floc proshyduced by increasing the pH settles at a rate of about 1 mmsec-1 and that its volume after having settled for about 100 sec is about 100 of the initial total volshyume of the brine Its volume would probably be conshysiderably smaller if it were allowed to settle for a longer period of time as would be the case in an actual setshytling tank

Experiments with initial dissolved silica concenshytrations of 08 and 09 gl-l gave essentially the same results except that the decline in MAS concentration to steady-state values required the full 20 minutes with the lower initial concentrations (See Table 2)

Similar experiments performed using the low Ca synthetic brine gave the same results (See Table 3)

In one experiment 005 gl-l of well-dispersed bentonite clay was added at the beginning In spite of this the treated brine was clear and the final suspended silica concentration was as low as usual This suggests

TABLE 2 FLOCCULATION EXPERIMENTS WITH HIGH Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Floccu1ant used Si02 pH (30 min) (40 min) Si02 at (min)

121278 none 1000 740 151 t 56 gt96 121178 none 1000 717 321 39 gt100 2 279 pH raised to 777 ROO 741 459 20 39 gt50

121978 pH raised to 779 800 718 lt10 lt10 17 gt70 1 579 pH raised to 774 900 729 44 15 33 gt60

122178 pH raised to 801 900 712 460 27t 35 gt48 121378 pH raised to 775 1000 724 15 lt10 33 gt60 61179 MF 572C 2 ppm 1100 723 69 t gt34 gt33 53079 MF 572C 2 ppm 1100 720 lt10 50 gt30 11079 MF 573C 1 ppm 1000 725 449 36 24 gt50 1 979 MF 573C 2 ppm 1000 727 576 lt10 10 gt60 42779 MF 573C 5 ppm 900 727 ltlOt 40 51 1 879 MF 585C 5 ppm 1000 717 34 13 18 gt50 61279 MF 591C 2 ppm noo 733 60t gt32 gt32

122078 MF 1563C 2 ppm 800 709 295 26 53 gt60 121878 MF 2535C 2 ppm 1000 723 64 31 13 gt67 2 779 MF 836A 2 ppm 1000 734 78 58 28 gt60 2 679 MF 836A 20 ppm 1000 725 39 40 22 gt50 43079 Ca1gon M 503 5 ppm 1100 731 18 30 gt30

NOTES Silica concentrations in ppm 1 ppm = 1 mgmiddot 1-1 MF American Cyanamid Magnifloc POint taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

574

TABLE 3 FLOCCULATION EXPERIMENTS WITH LOW Ca SYNTHETIC BRINE

Quantity Initial Initial Suspended Si02 at Final suspended Date Flocculant used Si02 pH 30 min 40 min Si02 at (min)

41379 none 1094 724 253 179 27 gt60 41079 none 1000 726 356 294 32 gt60 41179 pH raised to 790 1000 729 ltlOt 25 gt34 41779 pH raised to 783 1094 730 lt10 36 gt30 53179 MF 572C 2 ppm 1100 722 617 243 t 177 gt45 6 479 MF 572C 2 ppm 1100 734 530 31 26 gt40 6179 MF 572C 2 ppm 1100 732 278 303 gt30 42479 MF 573C 2 ppm 900 728 38st 310 65 41879 MF 573C 2 ppm 900 727 499t 32 52 6 579 MF 577C 2 ppm 1100 736 79t gt32 gt36 41679 MF 585C 2 ppm 1094 732 ltlOt 31 gt30 6 579 MF 587C 2 ppm 1100 732 94 gt32 gt31 6 679 MF 591C 2 ppm 1100 731 295 27 gt34 5 179 Calgon M 570 5 ppm 1100 725 92 133 gt31 5 279 Ca1gon M 580 5 ppm 1100 723 599t 144t 82 gt45 5 279 Calgon M 590 5 ppm 1100 730 149 174 gt30 41279 PuriF10c C31 2 ppm 1000 727 58t 30 gt32 42079 Separan CP7 2 ppm 900 725 503t 338 gt28 41979 Separan CP7 2 ppm 900 741 230t 55t 07 gt44 41179 Separan CP7 2 ppm 1000 737 39t 24 gt33 5 179 Separan CP7 5 ppm 1100 730 18t 21 gt28 42579 Separan CP7 7 ppm 1000 727 170t gt31 gt35

NOTES Silica concentrations in ppm ppm = mg- 1 MF = American Cyanamid Magnifloc Purifloc and Separan are Dow products Point taken before treatment tPoint taken within 5 min of stated time rather than at exactly that time

that the amorphous silica takes the clay particles with it when it flocculates and settles out We conclude that this process may give brine that is completely free of all suspended solids even if solids other than amorphous silica are initially present in it

In these experiments sodium hydroxide was used to increase the pH as a matter of convenience In pracshytice lime (calcium oxide or hydroxide) would be used because it is by far the cheapest base available We estishymate that only about 37 mg1-1 of calcium hydroxide would be needed to raise the pH to 78 (see Iglesias and Weres 1980)

Rothbaum and Anderton (1975) hilVe pilot testshyed a process for removing both suspended and dissolved silica from the brines at Wairakei and Broadlands New Zealand by adding a large amount of lime to them (up to 700 ppm CaO) The silica was precipitated out as an amorphous calcium silicate Adding lime in quanshytities as small as those proposed here apparently was not effective in flocculating the suspended silica beshycause the brines in these two fields are of much lower salinity than that at Cerro Prieto and contain much less calcium Sodium hypochlorite was also added to the brines and this caused the arsenic in it to precipishy

tate out with the silica Ali in all this is basically a difshyferent process from the one proposed here

EFECT OF SYNTHETIC FLOCCULANTS AND SUMMARY

Cationic polymers are known to be effective floccularits of colloidal silica So are quaternary amines with long side chains (general formula NR4 +X-l) For a discusshysion tf the general principles involved see lIer (1973 p 53-63 1979 p 384-396)

A variety of synthetic organic flocculants were obtained from three manufacturers and tested The results of these experiments are summarized in Tables 2 and 3 The flocculants tested are listed and briefly described in Table 4 The detailed results of a synthetic brine experiment with a synthetic flocculant are preshysented in Figure 3 t

The experiments in this series were like those discussed above in the section on silica removal by inshycreasing pH with one exception The requisite amount of the synthetic flocculant was added to the synthetic brine with stirring after about 30 minutes instead of adding NaOH

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

575

TABLE 4 SYNTHETIC FLOCCULANTS TESTED

Manufacturer Product Chemical nature

American Cyanamid Magnifloc S72C Probably a polymeric quaternaryammine

n It S73C do

n It S77C do

n n 58SC do

n It S87C do

n 591C do

It lS63C Cationic polymer otherwise mknown

n 2S35C do

Calgon MS03 Cationic polymer otherwise unkown

MS70 do

n M580 do

tI M590 do

Dow Sep~ran CP7 Probably a cationic polyacrylamidet

Purifloc C31 do

The American Cyanamid product literature did not identify the chemical nature of their products However the Magnifloc 500Cseries products are stated to be of low molecular weight and a shipping label on a package of samples we received identified the contents as quaternary ammonium compounds

tThe Dow product literature stated that most of the products in the Separan and Purifloc series belong to this chemical class

The most important data in Tables 2 Imd 3 are the final suspended silica values determined at the end of the experiment by the filtration method but these values cannot be considered in isolation The final susshypended silica values for the first two experiments in Table 2 (untreated brines) are low but only because of the abnormally long time that these experiments were allowed to run A treatment time this long would probshyably not be acceptable in practice The 30- and 40- minshyute suspended silica values determined by the differential method are high and the slow steady decline of suspendshyed silica concentration in the second experiment is shown in Figure 2 (solid symbols) The suspended silica conshycentrations in untreatedmiddot low-Ca synthetic brines likewise did not drop to acceptable values in a reasonable time (Table 3)

Raising the brine pH to about 78 quickly reshyduced the suspended silica concentration and gave a fishy

nal value below 5 mgl-l in all cases both with the highmiddot and low-Casynthetic brines (Tables 2 and 3 and Fig 2)

With the high-Ca synthetic brine several of the synthetic flocculants gave generally promising reSUlts Magnifloc 573C appears to be the most promising of them Of the flocculants tested with the high-Ca synshythetic brine only Magnifloc 836A which is an anionic polymer was confirmed to be definitely ineffective

As is evident from Table 3 synthetic flocculants in the concentrations used did not give good results with the low-Ca synthetic brine The one low final susmiddot pended silica value obtained with Dow Separan CP7 was probably anomalous because Ve were unable to reproduce it despite four attempts to do so Some of the entries in Table 3 do however suggest that at least some of the synthetic flocculants cause a significant

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

576

- -S o g 05 lfj

I~ 0 o

Add 1mg I-I MF573 C with brief stirring

96degC pH 725 Cj 19l1

05g[-1 Ca+2

I I I I I I I I I

o Total Si02 o Molybdate active

2 A mgl-I final suspended SiO

50

XBL 7992959 Time (min)

Figure 3 Synthetic brine flocculation experiment High-Ca synthetic brine Effect of Magnifloc 573C

decrease in suspended silica concentration even though the final values were still unacceptably high These flocshyculants might have been more effective if added in largshyer amounts but such a treatment would not be pracshytical given the relatively high price of these products and the very large amount of water that needs to be treated

We were unable to explain the difference in efshyfectiveness of the synthetic flocculants with the two different synthetic brines Nor were we able to correshylate it with any particular chemical difference between the two synthetic brine compositions

The advantages of using syntheticmiddot flocculants in practice would be

1 Relatively small amounts are used 2 The only preparation they need prior to use

is dilution with clean water to about 1 conshycentration

3 They cannot cause the precipitation of carbonshyate minerals as increasing the pH might

Their disadvantages are

1 Most of them are expensive 2 They slowly decompose especially at high

temperature 3 They do not increase the solubility of silica

as lime does 4 They are specialty products that might not

be conveniently and reliably available for use at Cerro Prieto

The advantages of using lime are

1 It is very cheap and universally available

2 It increases the solubility of silica in the treatshyed brine

The disadvantages of using lime are

1 Increasing the pH might induce the precIpIshytation of carbonate minerals see Iglesias and Weres 1980 or of an amorphous calcium silicate phase

2 The equipment used to produce lirne milk or gel from quicklime is somewhat complishycated -and expensive

FIELD VERIFICATION OF LABORATORY RESULTS

A few experiments were performed at Cerro Prieto to confirm the laboratory results These experiments were essentially the same as those discussed above except that freshly collected samples of actual geothermal brine were used instead of synthetic brine Also when the brine pH was raised a nearly saturated solution of calshycium hydroxide was used for this purpose instead of sodium hydroxide

These experiments were performed at the CFE Laboratory at Cerro Prieto with the help of J J Fausshyto L of CFE Brine from Cerro Prieto wells M-14 and M-30 was used because these wells are near the laborashytory The brine from M-30 is considered to be more representative of the field as a whole

Wells M-14 and M-30 are both producing wells that were steadily producing steam and brine when these experiments were performed These wells (and all others at Cerro Prieto) have a special smail-diameshyter brine-sampling line and miniature Webre separator

Add 20mgl-1

o Co (OH)2 o Total Si02~

I o I o Molybdate

active Si02I I

0 0 5 o I o2 o o I D o(i)

I l4mg 1-1 final

pH ~ 72 I suspended Si 02I Final pH not recorded

OL-~_~____~__~~____-L____~_____ I

a Time (min)

XBL 799-2975

Figure 4 Experiment performed at Cerro Prieto using freshly flashed brine from welt M-30

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

577

connected to the wellhead When this system was first turned on brine was allowed to flow through it for at least 20 minutes before sampling The sample was collected in a vacuum bottle which had first been rinsed with hot brine three times to bring it up to brine temmiddot perature Immediately after taking the sample the time and temperature were recorded and the first two 1-ml aliquots for the determination of total silica and MAS were withdrawn and put into prepared plastic 50middotml flasks that contained sodium hydroxide and ammomiddot nium molybdate reagent respectively The sample was then rushed to the laboratory by car The time from sampling to laboratory was usually about 8 minutes At the laboratory the whole sample was poured into a prepared beaker in a hot water bath and two more 1 ml aliquots were withdrawn for silica analysis

The colloidal silica that formed in the brine takshyen from Mmiddot30 usually began to flocculate and settle by the time that the brine was poured into the beaker However the amount of colloidal silica that remained in suspension after 30 minutes (gt10 mgI-) probably would not be acceptable for reinjection Figure 4 shows how well adding lime water to Mmiddot30 brine removes the suspended silica Adding 2 mgIl of Magnifloc 573 to M-30 brine did not reduce the suspended silica to acceptshyable levels (The results of these experiments are not shown here)

The colloidal silica in unmodified brine from M-14 does not flocculate even after the brine has been incubated for one hour Figure 5 shows this and the effect of adding a small amount of lime was added by mistake but there was no chance to repeat the expershyiment using more lime Even this small amount of lime caused most of the colloidal silica to flocculate and settle out Adding more lime say 30 mgl-l would

-~-95-0-C----r--A-dd-1-0-m--r-I- C-a-(-OH-)-2-] g I

I --0-_-----0---0-- pH 745

I 2 ~ i I o ~I~~D I-I~

(J5 ~ 23mg final pH 742 I suspended

silica

Time (min) XBL 799middot2960

Figure 5 Laboratory test at Cerro Prieto using brine from well M-14 One or both pH values may be in error

probably reduce the suspended sil ica in Mmiddot14 brine to acceptable levels The very small difference between the pre- and postmiddottreatment pH values in Figure 5 is probably due to experimental error

Adding 2 mgI of Magnifloc 573C to M-14 brine did reduce the suspended silica concentration to an acceptable level (Fig 6)

The vol ume and qual ity of the data we obtained in the field was reduced by the difficulty of working in an unfamiliar laboratory and time constraints How~ ever the results obtained do seem to confirm the reo suits of the laboratory work Increasing the pH seems to work with brine from either well while Magnifloc 573C works only with brine from Mmiddot14 this is exactly the same pattern as that observed working with the two different synthetic brines

RECOMMENDATION FOR A PILOT PLANT TEST

We doubt that bench tests with either synthetic brines or samples of real brine can profitably be carried much further The results reported here are probably about as complete and convincing as can be obtained in this way given the intrinsic limitations of bench tests and the variability of natural brines Further progress will only come from pilot plant experiments in the field at Cerro Prieto

Figure 7 is a schematic diagram of a pilot plant that would allow the brine treatment processes sugmiddot gested by the results of our bench tests to be evaluashyted in a practically meaningful way A design flow rate of 5 to 10 m3 hr- would probably be most convenient to work with This pilot plant resembles the one used by Rothbaum and Anderton (1975)

tAdd 2mgl 0 MF573C

0 I0

0 I

0 I0gt 00 0

0 I l g 0I (J5

o Total Si02 27mgl- final o Molybdate active Si02 I suspended oH ~ 74 I Si02

10 20 30 40 50 60 70

Time (min) XBL 799middot2977

Figure 6 Laboratory test at Cerro Prieto floc 573C on brine from well M-14

Effect of Magnishy

05

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

578

When operated without sludge recirculation this pilot plant would closely reproduce the processes evalshyuated in our bench tests As in the bench tests the function of the brine aging step is to allow time for the dissolved silica to be converted to colloidal silica However partial flocculation and settling of colloidal silica in the aging tank could lead to problems with sludge accumulation there If this proves to be a serious problem the aging step could be replaced by recirculation of part of the sludge coming out of the clarifier The large amount of colloidal silica in the sludge would react with the dissolved silica in the brine and decrease the dissolved silica concentration in the resulting mixture to its steadyshystate value almost instantaneously Sludge recirculation for this purpose is considered to be essential at Niland California because of the rather low rate of silica poshylymerization in the brine there (Quong et al 1978) However at Cerro Prieto the rate of silica polymershyization is high enough for brine aging to be a reasonshyable alternative As little as 5 minutes of aging might be enough

When this pilot plant is operated without sludge recirculation scale is most likely to form in the brine inlet pipe and the part of the aging tank nearest to it The aging tank should be designed with this in mind For example it might be possible to design it in such a way that the incoming brine does not come into conshytact with any part of the tank until a few minutes afshyter middotit enters the tank and is already relatively nonreactive

The sketch of the brine aging tank in Figure 7

Chemical supply

Lime slurry or synthetic flocClJlant solution

f lash mix tank

Sludge recircul3tionIoop optional

Brine aging tank

Sludge to evaporation pond

is meant to be a view from the top The brine flows from side to side around vertical baffles

A vertical laminar flow tank without baffles which the brine enters at the top and leaves at the bottom is another possibility All that is really required is that the aging tank provide approximately plug flow reacshytor conditions that sludge not accumulate in it and that it be easy to clean

In Figure 7 the clarifier mixing tank and sludge recirculation loop are shown as separate components If a process that includes sludge recirculation were to be ultimately implemented all three of these composhynents could be replaced by a reactor-clarifier which combines the functions of all three in a single unit Even if sludge recirculation is not required there are clarifiers available that incorporate the function of the mixing tank as part of the clarifier The separate comshyponent configuration shown in Figure 7 is recommendshyed for the pilot plant because it allows maximum exshyperimental flexibility and clearest separation of the effects of the various components on the overall pershyformance of the process

The pilot plant will require a dual chemical feed system to allow both pH increase and synthetic flocshyculant addition to be evaluated The feed system for the synthetic flocculant need consist only of a storage tank for the flocculant solution and a metering pump

In a full scale brine treatment facility pH would

COfe holders

etc

to pond

XBL 799 2849

Figure 7 Recommended brine treatment pilot plant facility for Cerro Prieto~

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

579

be increased by adding a hydrated lime slurry or gel to the brine The slurry or gel would most likely be produced on the spot by hydrating quicklime in anshyappropriate reactor However this would be completely impractical to implement on the small scale of the pimiddot lot plant An infinitely simpler and more convenient method to generate a hydrated lime slurry would be to mix metered solutions of NaOH and CaCI 2 just beshyfore they enter the mixing tank This would produce a suspension of Ca(OHh in NaCI solution Suitable experiments performed in our laboratory indicate that Ca(OHh scaling at the point of mixing would not be a problem It would be even simpler but less realistic to use NaOH alone to increase the brine pH in the pishylot plant tests

We doubt that a final brine filtration step will prove necessary at Cerro Prieto but it would be desirshyable to have that option available during the pilot plant work

The floc produced by the processes discussed in this report is mostly water and may actually represhysent as much as 5 to 10 of the total volume output

DraWing on related practical experience of Quong et al (1978) at Niland we recommend the following instrumentation and tests for monitoring the perforshymance of the process

1 A flow-through turbidimeter is needed to monshyitor the turbidity of the clarified brine This instrushyment would allow suspended silica concentrations above about 10 mgl-1 to be instantly detected and continshyuously monitored ina semiquantitative way This capashybility would be very helpful in coarse tuning the operation of the pilot plant

2 Some of the clarified brine should be pumped through a membrane filter The filtration equipment should include a precise constant flow rate pump and instrumentation for continuous monitoring of the presshysure drop across the filter This equipment would allow the formation plugging potential of the clarified brine to be determined via the method of Barkman and Davidshyson (1972)

3 The solid residue that collects on the filter should be analyzed to quantitatively determine the concentrations of suspended silica and calcium carbonshyate in the brine For details of these techniques see the section on experimental methods above and Weres et al (1980a Appendix 2) respectively There is no lower limit to the concentration of suspended solids that can be measured in this way Simply flow enough brine through the filter to accumulate the amount needshy

4 Clarified brine should be pumped though a sandstone core that is representative of the reservoir The core holder should be instrumented to allow the permeability to be followed continuously as with the membrane filtration apparatus High core confining pressure is not needed a simple Hassler sleeve-type core holder immersed in a boiling water bath would be completely adequate

5 Cores used in the core-flushing experiments should be sectioned and studied petrograph ically afshyterwards

6 Appropriate provision should be made to samshyple all streams in the pilot plant for analysis Total silshyica MAS and pH should be measured at all points and complete mass balances determined The brine aging tank and clarifier should each be provided with several sampling ports

7 Corrosion monitoring coupons should be disshytributed throughout the system

8 Quartz glass coupons should be distributed throughout the system to determine scale deposition rates The use of quartz glass coupons would allow the scaling rates to be quickly and quantitatively detershymined without the complicating effects of simultaneous corrosion processes The weight increases of the coushypons would give the rates of scale deposition and scale morphology could be determined by examining the specimens microscopically (Laboratory experiments analogous to this are discussed below)

Both in the pilot plant and in the ultimate pracshytical system all equipment should be designed to keep air out of the system Steam jacketing the major vesshysels is recommended for this purpose Particular efforts should be made to control heat loss from the small and complex clarified brine testing system discussed above

SCALE DEPOSITION FROM SYNTHETIC BRINES EXPERIMENTAL METHODS

Extensive solid and semisolid deposits of nearly pure colloidal amorphous silica form throughout the existshying waste water disposal system at Cerro Prieto Field experiments with low-pressure (Le second-stage) steam separator units have shown that rapid deposition of amorphous silica scale occurs within the separators if they are operated with an exiting brine temperature below about 130degC This scale is solid and forms at a rate of up to about 1 mmday- (R Hurtado J private communication and scale samples 1978) These deposits consist of colloidal amorphous silica that has been more or less cemented by the molucular deposition of disshysolved silica between the particles (See Weres et aI 1980a Section 10)

ed for convenient analysis this is typically only a few milligrams Silica scaling is presently only a minor problem

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

580

at Cerro Prieto because the existing very simple wasteshywater disposal system is not sensitive to it However second-stage steam separators are and a preinjection brine treatment system would be as well Therefore the kinetics of scale deposition and possible control methods are of interest These things may be studied in the laborashytory using synthetic brines precisely because scaling rates at Cerro Prieto are high enough to be conveniently measshyurable

For scaling rate data to be meaningful the scale must be deposited under steady state chemical condishytions This is clearly not possible in a beaker test like those discussed above and a continuous-flow kishynetic system must be used

The system developed for this experiment is scheshymatized in Figure 8 The compositions of the three solutions employed in most of these experiments are presented in Table 5 as well as the composition of the synthetic brine that is formed when two of them are mixed This synthetic brine has the same calcium conshycentration as does the highmiddotCa brine in Table 1 and the same sodium potassium and boron concentrations as the Low-Ca synthetic brine in Table 1 It differs from both of the synthetic brines in Table 1 in that the conshycentration of barbital in it is 2-12 times higher

The three solutions were pumped at equal rates by separate cassettes on a 10-channel Manostat Casshysette pump The cassettes were attached to the low speedshydrive shaft of the pump and 15-mm I D soft Tygon

0shy

f 0shyu ~ li

rf

Mixing manifold

Preheating coils

Water bath ~ 950 C

3mm LD f~sed quartz tubing

Air bubble trap

Barbital (Sol AI

Hel +

Salts (Sol BJ

tubing was used In most of the experiments the flow rate of each solution was 23 mlhr-1 and the total synshythetic brine flow rate was 46 mlmiddothr-1 bull

The whole kinetic system was immersed in a hot water bath whose temperature was nearly constant at about 95degC All of its components were compactly mounted on a supporting rack made out of silver solshydered 3-mm brass rod The maximum dimensions of the rack were about 26 x 16 x 16 cm

After leaving the pump the three solutions flowed through 1-m-long heat-exchange coils made out of thinshywalled Teflon tubing with an inner diameter of about

1 mm

The synthetic brine was constituted and the silica polymerization and scale deposition reactions were initiated by mixing (preheated) solutions A and B in a mixing manifold made out of a Teflon block Plugging of the mixing manifold by semisolid silica deposits was a recurrent problem This problem was ultimately reduced to manageable proportions by (1) reducing the silica concentration in the synthetic brine to 10 gI-1 and (2) drilling out the exit channel of the mixing manifold to about 2 mm I D Even so the exshyperiment often ended when the mixing manifold plugshyged up and one of the upstream tubing connections popped to relieve the pressure Careful operator atshytention was needed to accurately note the time at which this happened so that the total reaction time could be determined

Diluent brine stream

Neutral brine

XBL 799 shy 2851

Figure 8 Synthetic scale deposition apparatus There are actually four lengths of quartz glass tubing each of them consisting of several short segments that are held together by Tygon connectors An early version of the pH electrode manifold is shown here

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

581

TABLE 5 SOLUTIONS USED IN SCALE DEPOSITION EXPERIMENTS

Final synthetic Neutral Solution A Solution B brine brine (ppm) (mM) (ppm) (mM) (ppm) (mM) (ppm) (mM)

Si02 2000 333 1000 167 0

Na From Na 2Si03From NaBarb From NaCl

1530

2299

666

100 9389 4984

Total Na 3829 1666 6610 288 6610 288

K 2872 74 1436 37 1436 37

Ca 1002 25 501 125 501 125

B 262 24 131 12 131 12

Barbital 100 50 0

HCl To To

neut silo st ad] pH

666 118

Total HCl 784

Total Cl 21655 6108 10827 3054 12409 350

The scale was actually deposited inside of 3-mm 10 fused quartz tubing This material was chosen beshycause it is very nearly chemically identical to the silica scale being deposited Therefore the rare of deposition inside these tubes should not be very different from the rate of deposition on preexisting scale deposits ie it should closely approach the steady state deshyposition rate despite the very small amount of silica

scale actually deposited during a typical experiment

The fused quartz tubing run typically consistshyed of 10 short segments with a total length of 67 cm Before the experiment the cut ends of each segment were fire polished and the segments were rinsed thorshyoughly with 01 water and dried Then they were codshyed with indelible ink for identification and carefully weighed with an electronic balance to plusmn 01 mg These short segments were assembled into four straight lengths by joining them with short pieces of soft Tygon tubing slipped over the ends The ends of the segments within each length were brought together snugly within these Tygon connectors so that the fused quartz tubing withmiddot in each length was essentially continuous

One segmented length of 15 cm and two of 20 cm and one continuous length of 12 cm were used (Only three lengths are shown in Figure 8 and their segmented nature is ignored for simplicity) These parshy

ticular lengths were chosen for convenient assembly and handling The straight lengths are connected by longer pieces of soft Tygon tubing which is also used for connections elsewhere in the system The Tygon tubing used for this purpose has an inner diameter of about 2-12 mm The length of the flow path through each of the longer Tygon connectors between the lengths of quartz glass tubing was 5 cm

All joints between the longer Tygon connectors and glass or Teflon tubing were reinforced by wire ties wound around them Also the supporting rack was carefully designed to minimize forces on the connectors These precautions were necessary because Tygon sofshytens in hot water and the connectors would have freshyquently slipped off otherwise

The total length of the quartz glass tubing and Tygon connectors was 82 cm At a brine flow rate of 46 mlhr-1 this gave a total transit time of about 71 minutes

After leaving the quartz glass tubing the synthetshyic brine flows through a glass manifold that holds the tip of a pH electrode Before it enters the electrode compartment the synthetic brine is diluted by oneshythird with the neutral brine This solution is the same as the synthetic brine except that it does not contain

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

582

any silica or barbital (see Table 5) The purpose of dilutshying the brine at this point is to reduce the concentration of dissolved silica and thereby avoid silica deposition on and damage to the glass membrane of the electrode The neutral brine used for this purpose was formulated to resemble the synthetic brine in order to minimize the change in pH caused by dilution

An inverted Tit was built into the manifold tc remove bubbles from the brine before it reached the electrode This was necessary because bubbles of steam and air inevitably formed in the solutions during preshyheating and caused wild fluctuations in the pH readshying if allowed to reach the electrode compartment Both the main longitudinal member of the manifold and the air vent were made out of 3-mm I D glass tub ing A smeller air ventdiameter would have caused burpmiddot ing and the tube it was joined to could not have a smaller diameter than the vent

The brine flowed into the electrode chamber through the bottom up past the electrodes glass bulb and out through a side port In early versions of the manifold another side port was provided for a thershymoelectric temperature sensor which directly read the brine temperature (rather than the water bath tempershyature) Brine temperature monitoring was abandoned after it was establ ished that it was never more than 1degC Lower than the water bath temperature and the second side port was eliminated from later versions of the manifold The annulus around the electrode was sealed off with two O-rings

Maximizing the time resolution of the pH readshying requires that the volume of the manifold be minshyimized by careful design Its longitudinal member was only 8 cm long and we were able to achieve a time resolution of about one minute in the pH reading

The pH electrode was allowed to preheat before the start of the experiment It was calibrated by pumping a commercial standard pH 7 buffer solution (Mallinck- rodt Buffer) through all three pump channels After a stable signal had been achieved the meter was set to the pH of the buffer at the bath temperature The tubes were then rinsed off with water and set in the reactant and diluent solutions Leaving a bubble in each of these feed lines allowed the operator to follow the position of the reactants and note when they met in the mixshying manifold After the reactant solutions reached the mixing manifold the pump was adjusted to give the desired flow rate of 23mlhr-1 per channel

A typical experiment ran for 1-12 to 2 hours At the end of the experiment the reaction tubes were removed and rinsed thoroughly to remove any loose silica They were placed in an oven set at 110degC and allowed to dry overnight When dry they were weighed the difference in weight before and after the reaction was the amount of silica deposited which was usually a few milligrams per segment

In a typical experiment about 75 ml of synthetic brine containing a total of about 75 mg of Si02 flowed through the system About 10 mg of this was typicalshyly deposited inside the quartz glass tubing An unknown but probably much smaller amount was depositshyed inside the Tygon connectors Because only about 10 to 15 of the total silica in the brine was deposishyted it seems unlikely that the relatively large surfaceshyto-volume ratio of the tubes significantly affected the results

SCALE DEPOSITION EXPERIMENTAL RESULTS

A total of about 20 scale deposition experiments proshyduced usable results The range of brine pH values in these experiments was 695 to 750

Devising adequate means of controlling and monshyitoring brine pH presented the major challenges in develshyoping this technique Some degree of varying mismatch between the pump channels pumping solutions A and B is inevitable Because these solutions contain submiddot stantial concentrations of base and acid respectiveshyly pumping irregularities cause fluctuations in the pH of the resulting synthetic brine Also it was not posshysible to put the barbital in with the acid and salts and to pretitrate it because the constituent solutions were prepared and kept at room temperature attempts to do so inevitable caused precipitation of the sparingly soluble acid form of barbital This is why a higher conshycentration of barbital was needed here than in the beakshyer experiments discussed above and why pH had to be continuously monitored

A Lazar sealed pH electrode was used in this work

The values obtained for scaling rate vs position along the flow path for four experiments at pH near 735 are presented in Figure 9 Each point represents the average scaling rate within a segment of quartz glass tubing determined by dividing its weight gain by the product of its inside surface area and the duration of the experiment The distance along tube is the disshytance along the flow path including the Tygon connecshytors measured from the beginning of the first tube segment (The three connectors start at about the 15 40 and 65 cm positions respectively) The symbol representing the scaling rate in a given tube segment is plotted at the position of the midpoint of that segshyment Each value is the average of the deposition rate over the whole length of the given segment The zero of the time scale is slightly offset from that of the disshytance scale to accountmiddot for the 7 -second transit time

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

583

Time from mixing (min) 5

95degC pH~ 735 Cj = 1gl-1

5xl0-6

Scaling rate vs position7

E c

N

E u CJl ~

gt e CJl c o u

(f) O~--~__L-__~_~~__~__~L-__L-~~ o 20 40 60 80

Distance along tube (cm) XB L 799middot2964

Figure 9 Experimentallv determined scale deposition rates as a function of position along the quartz glass tube and the time from mixing

between the point of mixing and the start of the quartz glass tubing

The dissolved silica concentration as a function of position along the tube is unknown This means that the rate of molecular deposition cannot be calculated [By molecular deposition we mean the deposition of dissolved silica onto a solid surface one molecule at a time See Weres et al (1980b) regarding the nashyture and kinetics of this process] However it is easy to determine that the rate of scale deposition is one or two orders of magnitude greater than the rate of molecular deposition of dissolved silica under similar conditions This demostrates that the dominant mechshyanism of scale formation under these conditions inshyvolves deposition onto the tube wall of colloidal silica formed in the liquid phase

The deposition rate is high at first and drops off rapidly with time and distance along the tube Comshyparing this data with the curves of MAS vs time preshysented in Figures 2 and 3 shows that the decrease in scaling rate is roughly parallel to the decrease in disshysolved silica concentration Because only a modest fracshytion of the colloidal silica that is formed in the brine is deposited as scale the decrease in colloidal silica susshypended in the brine with time cannot be the cause of the rapid drop-off in the deposition rate We conclude the rate of molecular deposition must control the overshyall scaling rate under these conditions

Almost certainly the first step of scale deposishytion involves the attachment of colloidal silica partishycles and dumps formed in the liquid phase to the

tube wall by electrostatic forces identical to those that cause flocculation in the liquid phase The particles are then permaflently attached and the deposit made solid by molecular deposition of dissolved silica beshytween the particles This model was originally proposed by lIer (1973)

Practically speaking Figure 9 shows that aging newly flashed brine for 3 to 5 minutes should suffice to greatly reduce the scaling rate further along in the system

Figure 10 shows the effect of pH on the deposishytion rate in the first three tube segments Between pH 695 and 735 the effect of pH on the deposition rate is comparable to its effect upon the rate of molecushylar deposition under similar conditions (see Weres et al1980b Figure 34 or Table A32) This means that the trend can probably be meaningfully extrapolated to at least moderately lower pH values No experiments at pH values below 695 were attempted because of the inconveniently low scaling rates in this range Alshyso such work would have required a different buffer system -probably barbital and maleate or maleate ashylone- with all the additional developmental effort and risk of inconsistent results that would have entailed The unexpectedly lower deposition rates obtained in the one experiment at pH 750 mayor may not be corshyrect they were not verified

Figure 10 suggests that even a moderate pH deshycrease may be effective in controlling silica scale depshyosition For example extrapolating the trend to pH 65 suggests that at this pH the deposition rate will be only about one-fourth of that at pH 735 This is discussed further by Iglesias and Weres (1980)

10-5

7 c-E N

E

o First segment bull Second segment x Third segment

0 0

0

0

0 u

5xl0-6 Q)-shye CJl C

0 U

(f)

deg0

0

x

bull bull bull x x

bull

bull x

bullbullbull

x

o

bull

0 L

7 75 pH

XB L 799middot 2963

Figure 10 Effect of pH on scali ng rat

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2

584

Scanning electron micrographs of synthetic silishyca scale formed in our experiments and scale specimens collected in the field at Cerro Prieto are presented and discussed in Section 10 of Weres et al (1980a)

CONCLUSIONS

We have demonstrated that the chemistry of silica in Cerro Prieto brine may profitably be studied in the laboratory using synthetic brines Because some propshyerties of the brine vary greatly with its composition such results must be confirmed in the field using acshytual brines However the results of the synthetic brine work are good enough to allow practical brine treatshyment processes to be proposed and to contribute to the planning of a well-defined program of pilot-plant studies

The same techniques could be applied to the studshyy of silica chemistry in other silica-rich geothermal brines that precipitate relatively pure amorphous silica These include the brines at Wairakei and Broadlands (New Zealand) Los Azufres (Mexico) and Baca site (USA)

The detailed conclusions and recommendations in this report are specific to Cerro Prieto Experience has shown that even the variation between individual wells in one field can be great enough to make a treatshyment that works with brine from one well not work with that from another The differences among fields are so large that for all practical purposes the problems at each field must be addressed separately Of course this is not to say that the experience gained and methodshyology developed in one place cannot be transferred to another

Studying the precipitation of calcium carbonshyate from synthetic brines would be a trickier experimenshytal challenge because of the volatile nature of carbon dioxide and because heterogeneous nucleation is probshyably important in the case of carbonates However properly designed and carefully executed experiments with synthetic brines should be able to address this problem as well The same consideration applies to the other carbonates and sulfates

ACKNOWLEDGMENTS

This work was suppoted by the Division of Geothermal Enershygy of the United States Department of Energy under contract W-7405-ENG-48 It was part of the Cerro Prieto Research Project in this Laboratory and fell under the auspices of the OOECFE cooperative agreement Reference to a company or product name does not imply approval or recommendation of the proshyduct by the University of California or the U S Department of Energy to the exclusion of others that may be suitable

We wish to acknowledge the extensive help rendered to us by the chemical researchers at CFE A Manon M J Fausshyto L and F Garibaldi and those at liE R Hurtado J and S Mercado G Both R L Fulton and E Eno worked with us on the design of a pilot plant that was ultimately not built by us Their assistance and encouragement are duly acknowledged

Other persons who contributed assistance information advice or encouragement to this work include J A Apps R A Gray S M Klainer D L Lessor M J Lippmann L L Mink F Pearson M R Reed H P Rothbaum W H Somershyton J F Thomas A H Truesdell D White and A Yee

We also thank P A Butler who typed this manuscript and C Carnahan and M J Lippmann who reviewed it

REFERENCES CITED

Allen L H and Matijevic E 1969 Stability of colloidal silmiddot ica I Effect of simple electrolytes Journal Colloidal

and I nterface Science v 31 p 287-296 1971 a Stability of colloidal silica II Ion exchange

Journal Colloidal and Interface Science v 33 p 420-429 1971 b Stability of colloidal silica III Effect of hyshy

drolyzable cations Journal Colloidal and Interface Scishy

ence v 35 p 66-75 Alonso H Domfnguez B Lippmann M J Molinar R Schroeshy

der R C and Witherspoon P A 1980 Update on reservoir engineering activities at Cerro Prieto in Proshyceedings Fifth Annual Stanford Geothermal Reservoir Engineering Workshop December 12-14 1979 Stanshyford Un iversity (in press

Barkman J H and Davidson D H 1972 Measuring water quality and predicting well impairment Journal Petroshyleum Technology p 865-73

Cosner S R an~ Apps J A 1978 A compilation of data

on fluids from geoth~rmal resources in the United States Berkeley Lawrence Berkeley Laboratory LBL-5936

Hurtado R Garibaldi F D(az C R Fausto L J J and Mershycado S 1980 Geothermal brine treatment for reinjecshytion at Cerro Prieto in Proceedings Second Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Comision Federal de Electricidad (this volume

Iglesias E R and Weres 0 1980 Theoretical studies of Ceshyrro Prieto brines chemical equilibria in Proceedings Secshyond Symposium on the Cerro Prieto Geothermal Field Baja California Mexico October 1979 Mexicali Coshymision Federal de Electricidad (this volumel

lIer R K 1973 Colloidal silica in Matijevic E ed Surface and colloid science New York John Wiley and Sons

1975 Coagulation of colloidal silica by calcium ions mechanism and effect of colloid size Journal Colloidal

and Interface Science v 53 p 476-488 1979 The chemistry of silica solubility polymerizashy

tion colloimiddotd and surface properties and biochemistry

New York John Wiley and Sons Matijevic E 1973 Colloid stability and complex chemistry

Journal Colloidal and Interface Science v 43 p 217-245 Quong R Schoepflin F Stout N D Tardiff G and Mc

Lain F R 1978 Processing of geothermal brine effluents for injection in Geothermal energy -- novelty becomes resource Transactions of the Geothermal Resources Council Annual Meeting Hilo Hawaii 25-27 July 1978

V 2 p 551-554

585

Rothbaum H P and Anderton B H 1976 Removal of silshyica and arsenic from geothermal discharge waters by precipitation of useful calcium silicates in Proceedings Second United Nations Symposium on the Development and Use of Geothermal Resources San Francisco 1975 Washin~ton D C U S Government Printing Office v 2 p 1417-1425

Weres 0 Tsao l and Iglesias I 1980a Chemistry of sil ica in Cerro Prieto brines Berkeley Lawrence Berkeley Labshyoratory LBL-10166

Weres- 0 Yee A and Tsao l 1980b Kinetics of silica poshylymerization Berkeley Lawrence Berkeley Laboratory LBL-7033

QUIMICA DE LA SILICE EN SALMUERAS DE CERRO PRIETO

RESUMEN

Se estudio la precipitacion de sflice amorfa en salmueras geotermicas sinteticas parecidas a las salmueras separadas en Cerro Prieto Se encontr6 que parte de la sllice disuelshyta polimeriza rapidamente y forma silice coloidal en susshypension La sflice coloidal flocula y se asienta lentamente a valores de pH cercanos a 735 para las sal mueras inmoshydificadas AI aumentar el pH hasta aproximadamente 78 por adicion de base y agitarlo durante algunos minutos se obtiene rltlpida y completa floculacion y asentamiento Estos resultados se confirmaron en el campo al usar salshymueras autenticas de Cerro Prieto Tambien se probaron varios floculantes comerciales Se encontro tanto en el laboratorio como en el campo que las aminas cuatershynarias son efectivas en salmueras de ciertas composiciones pero que no 10 son en otras Las poliacrilamidas son toshytal mente inefectivas

Estos resultados proporcionan el siguiente proceso simple para el tratamiento de las salmueras previo a su reinyeccion envejecer la salmuera por 10 a 20 minutos en un tanque de retencion cubierto agregar 20 a 30 ppm de cal (CaO) agitar 5 minutos y separar la sClice floculada de la salmuera usando un clarificador convencional

Se analizan las pruebas de planta piloto necesarias para reducir este proceso conceptual a la pnlctica

En un estudio separado se investigo la tasa de deshyposicion de escalas de sllice en salmueras sinteticas_ Se encontr6 que una modesta disminucion del pH podrla reducir significativamente fa tasa de deposicion de escalas a un costa razonable

INTRODUCCION

EI trabajo del que aqu se informa se planeo para apoyar los estudios en marcha relacionados con la reinyeccion de salmueras en Cerro Prieto que efectua el personal de la Comision Federal de Electricidad (CFE) y del Instituto de Investigaciones Electricas (lIE) EI trabajo que se lIeva a cabo en el Lawrence Berkeley Laboratory (LBL) es parte del Programa Cooperativo de CFE-DOE para Cerro Prieto

Con este proposito se supuso que las salmueras al ser reinyectadas ser Ian flasheadas y separadas del vapor dos veces y que se entregarfan a la planta de tratamiento de salmueras a presion atmosferica y a una temperatura cercana a 100degC Estas hipotesis son compatibles con los planes de CFE en el desarrollo del campo a largo plazo En dichas condiciones las salmueras de Cerro Prieto se encuentran sobresaturadas con respecto a la sllice amorfa por un factor de aproximadamente tres el exceso de sllice en solucion se convierte rapidamente en sflice coloidal por la formaci on de nucleos homogeneos y por el crecimiento de particulas coloidales La floculacion de esta silice coshyloidal y su adhesion a las superficies solidas crea deposishytos blancos masivos de sflice que se forman en todas las superficies del sistema de deshecho existente en Cerro Prieto

La inyeccion de salmuera con un contenido de 50shylidos en suspension de entre 400 y 500 mgP en un yashycimiento cuya permeabilidad se debe mayormente a la porosidad como el de Cerro Prieto causaria un dano

586

severo en la formacion Por tanto dicha sllice coloidal debe removerse de alguna manera antes de la reinveccion

La principal tarea asignada fue investigar sobre los metodos de remocion de sllice de las salmueras flasheadas de modo que resultaran adecuadas para su reinveccion Una de las tareas secundarias fue estudiar la tasa de deshyposicion de escalas de sflice en salmueras flasheadas a aproximadamente 100degC V encontrar los medios para control aria

Las tecnicas experimentales usadas en este trabajo son una extension de las tecnicas desarrolladas durante un estudio preliminar de la cinetica qufmica de la polimeshyrizaci6n de sflice en soluciones del tipo de las salmueras geotermicas (Weres et al 1980b)

Recientemente personal de los laboratorios del II E V de la CFE en Cerro Prieto realizo experimentos similashyres V confirmo estos resultados (Hurtado et al 1979) Actual mente dicho personal prueba los procesos para el tratamiento de salmueras que se les recomendaron a escala de planta piloto Los resultados de este trabajo estaran disponibles a corto plazo

Aparte de la sllice los carbonatos V sulfatos de calshycio estroncio V bario son los candidatos mas probables para formar precipitados V danar las formaciones despues de la reinveccion esta posibilidad se estudio teoricamente Los resultados de dicha investigaci6n se informan en otro estudio (Iglesias V Weres 1980)

EI contenido de este trabajo V el del de Iglesias V Weres (1980) se incluven en el informe de Weres et al (1980a) que tambien describe trabajo adicional no anaHshyzado aqul

Calgon Magnifloc Nalgene Purifloc Separan Teflon V Tvgon son marcas registradas aunque no siempre se las identifique como tales en el texto V en las figuras

SALMUERAS REALES Y SALMUERAS SINTETICAS

Para estudiar la precipitaci6n de sflice amorfa en salmueshyras de Cerro Prieto se usaron en nuestro laboratorio salshymueras sinteticas muv semejantes a las naturales En la tabla 1 se presentan las composiciones qufmicas aproxishymadas de dos de las salmueras sinteticas que se usa ron extensamente en los estudios de remoci6n de sllice Las concentraciones de sflice V valores de pH mostrados en la tabla son los que se utilizaron mas frecuentemente Sin embargo parte del trabajo se realizo con diferentes valoshyres de concentraciones de sllice V pH

Las concentraciones de los componentes restantes resultaron siempre aproximadamente iguales a las mostrashydas en la tabla (La concentraci6n de sodio vario ligerashy

mente con la concentracion de sflice V con el pH Ademiddot mas la evaporacion tendio a concentrar Hgeramente las salmueras durante los experimentos)

Estas salmueras sint~hicas se idea ron de modo que las concentraciones de sus componentes principales se encontraran en la misma gama que las salmueras reales de Cerro Prieto que se evaporarOA a presion atmosferica La formulaci6n con alto contenido de Ca se eligio mas 0

menos arbitrariamente de modo que semejara una salmuemiddot ra separada tipica de Cerro Prieto

La formulacion con bajo contenido de Ca corresshyponde aproximadamente a una mezcla de las salmueras de desecho de los pozos M-8 M-27 Mmiddot31 M-35 V M-46 de Cerro Prieto Se eligio esta composicion porque la fuenshyte de salmuerasmiddotusada en la mavona de los trabajos deplanmiddot ta piloto en Cerro Prieto es la linea de salmueras de desecho compartida por dichos pozos No se incluveron en las salmueras sinteticas los componentes quimicos que se considero no afectarfan probablemente el comportashymiento qufmico de la sflice en las condiciones de mayor interes

EI valor de pH = 73 que se eligio para las salmueshyras sirlteticas es tfpico de los valores pH de especfmenes de salmuera recientemente flasheada que se determinashyron en el campo con salmuera aun caliente Estos varfan entre 70 V 76 V se hallan muy a menudo alrededor de 73 o 74 (comunicaci6n privada de A Manon M V medimiddot ciones de campo de los autores)

Tambien se muestran los analisis tfpicos de salmueshyra separada de los pozos M-14 V M-30 En este trabajo de campo Hmitado en Cerro Prieto se uso salmuera de dichos pozos con salmueras recientemente flasheadas

Las salmueras de Cerro Prieto son debilmente buffer EI buffer principal a temperaturas cercanas a los 100degC V valores de pH entre 72 V 95 es el acido monosil fcico Si(OH)4 (De aquf en adelante el acido monosilfcico se abreviara MSA) A valores de pH mas bajos domina el sistema buffer dioxido de carbono-bicarbonato La natushyraleza de estos experimentos fue tal que result6 muv dishyffcil ajustar V controlar el pH de las salmueras sinteticas si se dependfa solo de estos sistemas buffer Por tanto se agrego una pequefia cantidad del compuesto buffer barbital (5 5 - acido dietilbarbiturico) en la formulashycion de la salmuera sintetica de los auto res EI pKa del barbital es aproximadamente 74 A pesar de que es un componente de la salmuera real de relativa importancia el bicarbonato se dejo general mente de lado en las salshymueras sinteticas porque no tiene efecto directo sobre el comportamiento de la sllice V su presencia habrla complishycado el ajuste del pH Se incluvo el acido b6rico porque se sabe que puede tener un efecto inhibitorio suave sobre

587

la polimerizacion de ltIa sflice bajo ciertas condiciones (vease Weres et al 1980b Seccion 315)

En todos los experimentos se constituyo la salmuera artificial y se inicio la polimerizacion de la sllice al mezshydar dos soluciones precalentadas una acida y una alcalina Durante los experimentos la temperatura de la salmueshyra estuvo siempre alrededor de los 92 y 97degC

ESTUDIOS DE REMOCION DE SIUCE METODOS EXPERIMENTALES

En estos experimentos la solucion alcalina consta de una cantidad premed ida de metasilicato sodico diluido La soluci6n acida contenfa las cantidades apropiadas de los componentes principales sales sodio barbital acido para ajustar el pH buffery neutralizar la sflice y agua para preparar el volumen total deseado Generalmente se emshypleo un volumen total de 300 mililitros

En este trabajo con salmueras simuladas todos los volumenes se midieron a temperatura ambiente P~r conmiddot siguiente cuando se dice 300 ml significa que el volumen deberfa ser ese si la solucion se enfriara a temperatura ambiente Cuando se aplica a concentraciones de sal y de sflice iniciales las unidades de moles 0 gramos p~r litro tambien se usan en este senti do Esto es para fines practicos 10 mismo que moles 0 gramos por kilogramo de agua debido al contenido de sal relativamente bajo de esshytas soluciones y a la naturaleza semicuantitativa de muchos de los datos de la salmuera sintetica

La soluci6n comun de metasilicato sodico contiene alrededor de 2 Emiddot 11 de Si02 Debido a que resultaba muy diffcil preparar esta solucion en una concentraci6n exacta predeterminada siempre se la estandarizo p~r comshyparaci6n con una soluci6n estandar de sflice (Fischer AA estandarL

La cantidad deseada de soluci6n de metasilicato s6dico se coloc6 en una botella plastica se sac6 el aire restante y se cerr6 con fuerza Se coloc6 luego en bane de agua hirviendo para su precalentamiento (Dicho proceshydimiento asegur6 que la boteita no se rompiera ni perdieshyra agua al ser calentada) Una alicuota separada de 50 ml de soluci6n de metasilicato s6dico se titul6 con 05 M de HCI se us6 un medidor de pH para determinar el vomiddot lumen de la soluci6n de HCI necesaria para neutralizar la cantidad dada de la solud6n de metasilicato sedico

La soluci6n acida inicial se obtuvo de una solud6n stock de salmuera concentrada una soluci6n stock de sodio barbital 05 M HCI y agua La salmuera concenshytrada era una mezcla de cloruros de sodio potasio y calshycio y acido b6rico en agua formulada de modo tal que al poner 3 partes de ella en 20 partes de la salmuera sinteshytica final se obtuvieran las concentraciones finales deseashy

das de los componentes principales Por ejemplo en el caso de la salmuera sintetica con alto contenido de Ca preshysentada en la tabla 1 la salmuera concentrada conten (a 1667 M NaCI 0256 M HCI 00833 M CaCl z y 00066 M H2 803 Las concentraciones de los tres ultimos composhynentes de este concentrado son simplemente 203 veces las concentraciones deseadas en la salmuera sintetica La concentraci6n de NaCI es menor porque el metasilicato 50shydico y las soluciones buffer stock tam bien contienen sodio

En el caso de la salmuera sintetica de bajo conshytenido de Ca la salmuera concetrada contenfa 1547 M NaCI 0245 M KCI 00547 M CaCI 2 y 00081 M H3 H03 bull

La soluci6n buffer stock de barbital se prepare para que resultara 05 M en barbital y 05 Men NaOH Se emshypleo barbital Mallinckrodt en grado purificado de uso farmaceutico (Todos los productos quimicos que se usaron aparte del barbital y los floculantes comerciales eran de grado reactivo)

Las cantidades que se midieron de la salmuera conshycentrad-a la soluci6n buffer stock y la cantidad de agua necesaria para lograr el volumen deseado se mezclaron en un vasa vertedor suspendido en un bane de agua hirvienshydo para preparar la soluci6n acida El vase se cubrio temshyporalmente con un vidrio de reloj y su contenido se calent6 hasta sobrepasar 90degC lnmediatamente la mezcla se titulo con 05 M HCI para ajustar el pH de la salmuera sintetica al valor deseado Finalmente se agrego al vase vertedor la cantidad de 05 M HCI adicional necesaria para neutralizar la soluci6n de metasilicato sodico como se determin6 mediante una titracion anterior y se calenshyt6 nuevamente la mezda resultante

Para comenzar la reaccion la solucion alcalina precalentada se volc6 en el vaso vertedor que contenfa la soluci6n Bcida precalentada y se agito rapidamente Despues de unos pocos segundos se suspendio el mezclashydo se midio el pH y desde ese momenta se tomaron muestras de I-ml peri6dicamente para su analisis

Mediante el metodo de molybdato amarillo se deshyterminaron la sllice disuelta y la sflice total Para detershyminar la sflice molybdato act iva (disuelta) se empleo directamente el metodo de molybdato amarillo Para deshyterminar la sllice total (la sllice coloidal suspendida inclusive) el especimen se reaccion6 primero con NaOH para digerir y disolver la sHice coloidal (De aqui en adeshylante molybdato de silice activa se abreviara MAS que no debe confundirse con MSA MSA es molybdato activo aunque tambien 10 son otras especies moleculares pequenas i6nicas de slliceJ

La diferencia entre el molybdato activo que se deshy

588

termina en forma experimental y las concentraciones de sllice total es igual aproximadamente a la cantidad de sishylice coloidal suspendida en la salmuera Esta tecnica diferencial tiene una resolucion aproximada de 10 mg I-I de sllice suspendida

Las concentraciones de sllice medidas de las que se informa en las figuras y en las tablas no se corrigieron por efectos de temperatura Por consiguiente las concenshytraciones de silice medidas estan en realidad en unidades de gl-l a la temperatura expuesta Para convertir aproximashydamente estos valores a unidades de gramos (kg Hz 0)-1 deben multiplicarse por 105

La jeringa de muestreo se precalent6 siempre al introducir y expulsar rapidamente salmuera de ella una 0

dos veces antes de tomar la muestra a analizar

EI metodo de molybdato amarillo se sujeta a intershyferencias serias por los rastros de sllice disuelta y fosfato inorganico que a veces estan presentes en el agua de la canilla Para eliminar este problema todas las soluciones se prepararon con agua doblemente deionizada (01) con resinas mezcladas anionicascationicas de intercambio ionico EI agua que sale de esta canilla 01 dellaboratorio ya fue doblemente deiejnizada una vez como precaucion se envio a traves de una columna lIenada con resinas mezshycladas para un segundo paso de doble deionizacion

Se uso el procedimiento de molybdato amarillo recomendado por R K lIer (1979 p 97) al igual que en un trabajo anterior La soluci6n stock de molybdato reactivo se preparo al mezclar 100 9 de molybdato de amonio (NH4 )6 Mo 024 bull 4HzO 470 ml de NH4 0H concentrado (28 N H 3) Y agua hasta completar 10 I de soluci6n La soluci6n de molybdato reactive para el trabajo final se preparo mezclando 100 ml del stock de molybdato 200 ml de 15 M H2 S04 y 500 ml de agua 01

Oicha soluci6n de molybdato reactivo se midi6 en alicuotas de 400-ml y se coloco en frascos volumetricos de 5O0-ml Para determinar la silice disuelta se agrego 10 ml de la soluci6n que serfa analizada al molybdato reactivo en el frasco y el volumen se lIevo a 500 ml con agua 01 Luego el frasco se invirtio se agito tres veces y se permitio la fijacion del color durante 35 minutos (Se descubri6 que al agitarse tres veces se elevaba la reproducibilidad de los resultados al asegurar una meshyjor mezcla) Se midi6 luego la densidad optica a 400 nm con un espectrofot6metro de doble rayo (PerkinshyElmer 550)

Para determinar la silice total se retiro una alicuota de 10-ml y se coloco en un frasco phistico volumetrico de 50 ml que contenfa 10 ml de 10 N NaOH EI frasco se cerro se agito y se coloco en un bane de agua caliente

durante 10 minutos para permitir que el NaOH digiriera la silice coloidal Oespues se agregaron 400 ml de la soshylucion de molybdato y agua hasta totalizar un volumen de 500 ml y se continu6 el procedimiento como se describio con anterioridad

En la mayoria de estos experimentos se trato la salmuera en una de las dos formas despues de transcurrishydos 20 0 30 minutos Un tratamiento fue aumentar el pH al agregar NaOH Otro consistio en adicionar algunos mgl-l de uno de los varios floculantes sinteticos comershyciales En cualquiera de los casos a la adici6n qufmica Ie sucedieron unos pocos minutos de nipida mezcla Oespues de mezclar se determino el pH nueva mente y se reanudo el muestreo peri6dicamente

Tambh~n se considero la posibilidad de que la agishytaci6n rapida mas que la adici6n de los productos qufmishycos concomitantes pudiera causar floculacion de la sishylice coloidal Oicha hip6tesis se descart6 mediante un experimento por medio del cual la salmuera se agita vigorosamente durante varios minutos en el tiempo justo sin agregar productos qu(micos se encontrJ que este experimento no ten(a efecto significativo sobre la cantidad de srlice que permanecera en suspension al fishynal del experimento

AI finalizar el experimento se determin6 la sflice suspend ida que permanecfa en la salmuera mediante un metoda absoluto Primero alrededor de 150 ml de la salmuera se decant6 en un vaso vertedor de 200 ml teshyniendo cuidado de dejar atras tanta cantidad de silice floculada precipitada como fuera posible Se coloco la salmuera decantada en un bane de agua de alrededor de 35degC durante unos minutos para permitir a la pequeshyna cantidad de floculado grueso no removida asentarse en el primer proceso de decantacion Con una pipeta de 100 ml precalentada se retiro cuidadosamente una muestra de 100 ml y se la coloco en un filtro embudo de presion precalentado Corrientemente se uso un filtro de membrashyna de policarbonato de 400 nm de tamano de poro Luego el filtro embudo se sell6 y se forzo el fluido a traves del mismo mediante gas nitrogeno comprimido a una preshysion de 13 bars

Se retir6 luego el filtro para colocarlo en un vasa vertedor de plastico con 30 ml de 10 N NaOH y se 10 calento en un ban~ de agua hirviendo durante 10 minutos Luego de la digesti6n de la silice el filtro se enjuag6 denshytro del vasa vertedor y los contenidos del mismo seenshyjuagaron en un frasco volumetrico de 500 ml que conteshynfa 400 ml de molybdato reactivo para la determinacion de silice Mas tarde calculos apropiados dieron la concenshytraci6n residual de silice suspendida en miligramos por Iishytro de la salmuera sintetica clarificada

589

Cuando se l1eva a cabo 10 descrito can anterioridad se via la imposibilidad de aplicar esta tecnica para detershyminar concentraciones de sflice suspend ida mayores de 32 mgl-l porque hicieron que el espectrofotametro se saliera de escala Por esta razon para algunos expeshyrimentos la sllice final suspendida se informa como gt 32 en las tablas 2 y 3 No obstante las altas concentrashyclones de sllice suspend ida eran general mente obvias por la cantidad de floculado acumulado en el filtro En dichas casas el floculado digerido se diluyo hasta 50 ml can agua 01 y luego se retiro una alicuota de 1-ml de dicha solucion y se determino la concentracion de sllice en la misma par los medios habituales Este paso extra en la dilucion tuvo el efecto de extender la gama del metoshydo haciaarriba por un factor 50

En casas en los cuales el tratamiento de sal muera era aparentemente exitoso las val ores finales de sllice suspend ida estaban general mente en la gama de 2 a 5 mgmiddotl-l Estos valores que se repiten de manera sospeshychosa pueden ser UA artefacto causado por una minima agitacion del floculado durante el segundo paso de decanshytacion Por consiguiente la cantidad de sllice suspend ida que queda en la salmuera despues de un proceso de asentashymiento mejor ejecutado (como en un tanque de asenshytamiento adecuadamente disenado en una planta piloto a la facilidad de tratamiento de salmuera a escala comershycial) puede probar actual mente que es menor de 3 mgl-l e igual todolo demas

Por supuesto este metoda solo puede determinar la concentracion de part(culas y floculado suficientemente grandes para no pasar par un filtro de 400-nm En la pracshytica probablemente esto implica todas las particulas y floculado cuya dimension lineal mayor exceda alguna medida considerablemente menor que 400 nm Aunque bajo estas condiciones las partfculas coloidales elementashyles estan posiblemente entre los 20 y 100 nm de diametro la naturaleza muy floculante de la salmuera tratada Qarantiza que mucha de la sllice suspend ida se eneuentre bajo la forma de floculado 10 suficientemente grande eomo para no pasar un filtro de 400-nm En el caso de pruebas de laboratorio no proporciona ninguna garantla utilizar un filtro mas fino pero seria posible en el caso de experimentos de planta piloto

Algunas veees tambien se determina la tasa de asentamiento del floculado Esto se hizo al volver a mezelar el floculado con la salmuera despues del proshyeeso de filtracion se volco rapidamente la mezcla en un cilindro graduado y se midio la veloeidad a laque se asienta el floeuJado eon un eronometro Los valores obtenidos de esta forma solo fueron aproximados porque la velocidad de sedimentaeion se afecta grandemente par eorrientes de conveccion y otros fenomenas

REMOCION OE SILICE POR INCREMENTO OEL pH

Se sabe bien que el incremento del pH puede desestabil ishyzar la sllice coloidal suspend ida en soluciones de sal y oeasionar la floculacion 0 coagulacion (Allen y Matijevie 1969 1970 1971 Matijevic 1973 lIer 1975 Weres et a 1980b Seceiones 221 y 22) Si aumenta el pH se eleva la carga negativa sobre la superficie de las parshytlculas y se absorben los cationes Oiehos cationes sirven de puentes entre las particulas y este efecto es 10 que oeasiona que las partieulas se adhieran y floculen (En este easo especial coagular es quizas mas apropiado que flocular pero se uso flocular en este trabajo para ser congruentes can el usa comun que se Ie da en el campo de tratamiento del agua) La habilidad de floculacion de los cationes aumenta rapidamente con SU

carga En las salmueras de Cerro Prieto el calcio es proshybablemente el principal ion floculante

En las salmueras separadas de Cerro Prieto la conversion de sllice disuelta a sll ice coloidal es rapida pero la remocion de la sflice coloidal de la s_almuera por medio de floculacion y asentamiento es lenta e inshycompleta Por tanto la tarea principal del tratamiento de preinyeceion de la salmuera es inducir floculacion y asentamiento rapidos Se reeomienda el incremento del pH como laforma mas facil para inducir una flocushylacion rltjpida porque la salmuera de Cerro Prieto contiene una sustancial cantidad de calcio

La figura 1 muestra de manera semicuantitativa el efecto de incrementar el pH En este experimento la salmuera sintetica se constituyo original mente con un pH de alrededor de 73 y se Ie permitio asentarse sin perturbaciones durante 25 minutos para permitir que la concentraeion de MAS cayera a un valor cercano al del estado estacionario Luego la salmuera se agit6 brevemente se permitio al floculado asentarse por 2 minutos y se retiraron las primeras dos alicuotas de l-ml para determinar la concentracion de sflice suspendida medianteel metoda diferencial Enseguida se agreg6 una pequena cantidad de 1 N NaOH para elevar el pH Luego de una breve agitaci6n adicional se midio el pH se permitio que el floculado se asentara a 10 largo de 2 minutos sin agitarlo y se retiraron dos alieuotas mas para la determinacion de la sllice suspendida Este cicio se repitio cuatro veces mas para generar el resto de los puntas

Los datos de la figura 1 son solo semicuantitativos porque puntos finos de diseno experimental tales como la duracion del asentamiento y la medida y forma del reeipiente de reaccion afectan enormemente los resulshytados que se obtienen de tales experimentos Sin embargo en la figura 1 es obvio que con un pH de 73 la floculashycion es pobre y que el incremento del pH a alrededor de 78 remueve la mayor parte de la sllice coloidal de la suspension

590

En la figura 2 se presentan los datos completos detail ados que se obtuvieron en dos experimentos tishypicos del tratamiento de salmuera En uno de ellos (los simbolos Ilenos) la salmuera sintetica tenia un pH inicial de alrededor de 73 y se la mantuvo as( durante todo el experimento En el otro (simbolos abiertos) la salmuera sintetica tenIa un pH inicial de alrededor de 72 que se incremento al agregar una pequena cantishydad de 1 N NaOH despues de transcurridos 22 minutos Con excepcion del cambio en el valor del pH y una

breve agitacion en el segundo experimento los dos exshyperimentos resultaron iguales En ambos se uso la forshymula de salmuera sintetica alta en Ca

En ambos casos para el momento en que las prishymeras muestras se tomaban para analizarlas una fracoion sustancial de la s(fjce ya habia polimerizado En este momento tambien hubo una debil bruma blanca en la salmuera que indicaba el comienzo de la floculacion (La sllice coloidal no floculada es invisible a simple vista en estas concentraciones) La muy rapida etapa inicial de la reaccion de polimerizacion se habia compleshytado para el momenta del segundo punto que se tome 5 minutos despues que el primero Esto significa que un tiempo de envejecimiento de 5 minutos probableshymente serfa suficiente para polimerizar la silice en una salmuera como esta en lugar de los 22 minutos que se permiten en este expedmento

Los valores de silice total de 20 minutos muestran que para ese entonces la s(( ice coloidal habla floculado al punto que estaba comenzando a asentarse La flocushylacion y el asentamiento resultaron apenas mas pronunshyciados en la salmuera sintetica que tenIa el mas elevado pH inicial

A los 25 minutos la concentraci6n de MAS en la salmuera no modificada habia ya descendido a un valor I (mite Dicho valor es un poco mayor que el nivel de equilibrio de solubilidad para la mayor(a de la s(lice amorfa bajo estas circunstancias que es alrededor de 033 gI-1 (Ver Weres et al 1980b Secciones 311 y 318 por procedimientos detallados para calcular la solushybilidad de la sllice amorfa bajo varias condiciones) Que el valor observado sea mas alto que el valor de solubishylidad global se debe a unas partculas de s(fjce coloidal pequenas que se crean bajo estas condiciones de nucleashyci6n rapida Las particulas pequenas son mas solubles que la mayor(a de la s(lice amorfa (AS) y su solubilidad determina la concentraci6n limite de MAS

Hasta el momenta en que se increment6 el pH de la salmuera tratada las concentraciones de MAS en las salmueras tratadas y no tratadas eran iguales Sin emshybargo la concentracion de MAS en la salmuera tratada se incremento al aumentar el pH y agitar vigorosamente

Esto se debe a que al aumentar el pH se eleva la sol ubishylidad de la silice amorfa desde alrededor de 033 hasta mas 0 menos 035 gI-1 en este caso Si el postratamiento de concentracion de MAS estuviera escasamente por encima del valor correspondiente de solubilidad en equishylibrio para la mayor parte de AS se debio nuevamente a la solubilidad mas grande de part(culas pequenas en realidad el lento aumento aparente de la concentraci6n de MAS en la salmuera modificada hacia el final del experimento se debe probablemente al reequilibrio conshytinuo de la sllice coloidal con su ambiente qu(micamente modificado

La redisoluci6n parcial de la sllice coloidal durante y quizas despues del proceso de floculacion es un feshyn6meno deseabledesde un punto de vista prcktico Significa que si la salmuera clarificada se separa bastanshyte rapido del floculado la pequena cantidad de sflice coloidal remanente en la salmuera probablemente se redisolvera completamente si se Ie da suficiente tiempo EI valor final de 33 mgI-1 de Si02 suspendido se deshytermino mediante la tecnica de filtraci6n Como se ve con anterioridad aun este pequeno valor puede deberse mas a una dificultad para lograr un buen asentamiento en un contenedor pequeno que a una floculacion imperfecta

AI finalizar este experimento la concentraci6n MAS era alrededor de 042 gI-1 que esta aun par encima de la solubilidad de la sflice en equilibrio bajo las condishyciones dad as Esto significa que la salmuera depositara lentamente silice vitreo sobre cualquier superficie con la que tenga contacto La mejor estimaci6n del grado de deposicion molecular es de alrededor de 5 JlmanO l

como se calcul6 con los metodos presentados por Weres et al (1980b Seccion 314) Esto puede 0 no ser demashysiado pequeno para evitar que decline la inyectividad sobre la vida de un pozo de inyeccion depende del tamaRo de poro en la formacion receptora y la extensi6n de la permeabilidad de la fractura Podr(a no afectar la permeabilidad de la fractura pero danar(a la permeashyIlilidad de poro si los poros fueran demasiado pequenos

Otros experimentos demuestran que el floculado que se produce al incrementar el pH se asienta a raz6n de 1 mmseg-1 y que luego de un asentamiento de alremiddot dedor de 100 s su volumen es mas 0 menos 10 del volushymen total inicial de la salmuera Probablemente dicho volumen podr(a resultar considerablemente menor si se permitiera un tiempo mayor de asentamiento como seria el caso con un efectivo tanque de asentamiento

Los experimentos con concentraciones iniciales de silice disuelta de 08 y 09 gI-1 dieron esencialmente los mismos resultados excepto que la declinacion en concentracion MAS a valores de estado estacionario requirieron los 20 min completos con las concentrashyciones iniciales mas bajas (vease la Tabla 2)

591

Experimentos similares que se Ilevaron a cabo con salmuera sintetica baja en Ca dieron los mismos resultashydos (vease la Tabla 3)

Sa agrego al comienzo 005 gI-1 de arcilla bentoshynita bien dispersa en un experimento a pesar de esto la salmuera tratada fue clara y la concentracion final de sllice suspend ida resulto tan baja como siempre Esshyto indica que la s(lice amorfa toma las partrculas de arcilla cuando flocula y se asienta Se concluye que este procedimiento puede hacer que la salmuera este completamente libre de todos los solidos en suspenshysion aun si esos solidos con excepcion de la srlice amorshyfa estan presentes en ella desde el principio

En estos experimentos se utilizo hidroxido de 50shydio para aumentar el pH por conveniencia En la pracshytica podrra usarse cal (oxido 0 hidroxido de calcio) porque resulta con mucho la base disponible mas ecoshynomica Se estimo que solo sera necesario alrededor de 37 mgI-1 de hidroxido de calcio para elevar el pH a 78 (vease Iglesias y Weres 1980)

Rothbaum y Anderton (1975) han probado p~r vez primera un proceso para eliminar srlice disuelta y suspendida de las salmueras de Wairakei y Broadlands Nueva Zelandia agregandoles una gran cantidad de cal (hasta 700 ppm CaO) La sflice se precipito como un silicato de calcio amorfo EI agregado de cal en cantidades tan pequenas como las propuestas aca no resulto apashyrentemente efectivo para la floculacion de la snice en sUspension porque las salmueras en aquellos dos campos son de mucha menor salinidad que en Cerro Prieto y contienen mucho menos calcio Tambien se agrego hiposhyclorito de sodio a las salmueras y esto ocasiono que el arsenico existente se precipitara fuera de la sllice Sise toma todo en consideraci6n basicamente este es un proceso diferente del propuesto aquf

EFECTOS DE LOS FLOCULANTES SINTETICOS Y RESUMEN

Se sabe que los pol rmeros cationicos son floculantes efectivos de la sflice coloidal Tambien 10 son las amishynas cuaternarias con cadenas largas (f6rmula general NR4 + X -1) Vease Idler (1973 p 53-63 1979 p 384-396) para una discusion de los principios general~s relacionados

Se obtuvo de tres fabricantes una variedad de flocushylantes sinteticos organicos y se hicieron pruebas los reshysultados de estos experimentos se resumen en las tablas 2 y 3 Los floculantes probados estan Iistados y se deshycriben con brevedad en la tabla 4 En la figura 3 se preshysentan los resultados detail ados de un experimento en salmuera sintetica con un floculante sintetico

cion fueron como aquellos vistos con anterioridad en la seccion de remocion de sJice con aumento de pH Se agrego a la salmuera sintetica la cantidad necesaria de floculante sintetico y se agitaron despues durante 30 min en vez de agregar NaOH

Los datos mas importantes de las tablas 2 y 3 son los valores finales de sllice suspendida determinados al final del experimento mediante el metodo de filtracion dichos val ores no pueden considerarse aislados Los vac lores finales de sflice suspendida resultaron bajos en los dos primeros experimentos en tabla 2 (salmueras no tratadas) pero solo debido al largo tiempo tan poco usual que duraron estos experimentos En la practica no serra aceptable un experimento tan largo Los valoshyres de 30 y 40 min de sflice suspendida determinados mediante el rrModo diferencial son altos y los bajos declinacion constante de la concentracion de srJice suspenshydida en el segundo experimento se muestran en la fjgura 2 (sfmbolos lIenos) Las concentraciones de sfl ice suspenshydida en salmueras sinteticas no tratadas de bajo calcio no descendieron a valores aceptables en un tiempo razonable (Tabla 3)

AI elevar el pH de la salmuera a alrededor de 78 se redujo rapidamente la concentracion de sllice suspenshydida y dio un valor final de 5 mgI-1 en todos los casos con ambas salmueras sinteticas la alta y la baja en calcio (Tablas 2 y 3 Y Fig 2)

Con la salmuera sintetica alta en Ca varios de los floculantes sinteticos dieron resultados generales promeshyted ores Magnifloc 5730 aparece como el mas prometedor de ellos De todos los floculantes probados con la salshymuera sintetica alta en Ca solo Magnifloc 836A que es un pollmero amonico se confirmo definitivamente como inefectivo

Como es evidente en la tabla 3 los floculantes sinteticos en las concentraciones usadas no dieron buenos resultados con la salmuera sintetica baja en Ca EI unico valor bajo final de s(lice suspendida obtenido con Dow Separan CP7 probablemente fue anomalo porque no pudo reproducirse a pesar de haberlo intentado cuatro veces Sin embargo algunas de las entradas en la tabla 3 indican que al menos alguno de los floculantes sinteshyticos causa un descenso significativo en la concentracion de dlice suspendida aunque los valores finales fueran aun inaceptablemente elevados Dichos floculantes pushydieron resultar mas efectivos si se los agregara en grandes cantidades pero un tratamiento semejante serfa practico dado el relativamente alto precio de estos productos y la enorme cantidad de agua que necesita para tratarse

No pudo explicarse la diferencia de efectividad de Los experimentos en estas series con una excep- los floculantes sinteticos con las dos diferentes salmueras

592

sinteticas Tampoco pudo correlacionarsela con alguna diferencia qu imica particular entre las composiciones de las dos salmueras sinteticas

En la practica las ventajas de utilizar floculantes sinteticos serian que

1 Se uti I izan relativamente pequefas cantidades 2 La unica preparacion que necesitan antes de usarshy

se es la dilucion con agua limpia en una concenshytracion de al rededor de 1 _

3 No pueden ocasionar la precipitacian de minerales carbonaticos como serla si se aumenta el pH

Sus desventajas que 1 Muchos de ellos son caros 2 Se descomponen con lentitud especial mente

a alta temperatura 3 No elevan la solubilidad de la silice como la cal 4 Son productos de especialidad que podr(an no

estar disponibles conveniente y fidedignamente para usarse en Cerro Prieto

Las ventajas de usar cal serlan que 1 Es muy barata y disponible universal mente 2 Eleva la solubilidad de la s(lice en la salmuera

tratada

Las desventajas de usar cal que 1 EI aumento del pH podr(a inducir la precipitashy

cion de minerales carbonaticos (vease Iglesias y Weres 1980) 0 de una fase de silicato de calcio amorfo

2 EI equipo que se usa para producir leche de cal o gel Ide cal viva es de algun modo complicado y caro

VERIFICACION DE CAMPO DE LOS RESULTADOS DE LABORATORIO

Para confirmar los resultados de laboratorio se Ilevaron a cabo algunos experimentos en Cerro Prieto Estos fueron esencialmente iguales a los expuestos con anterioridad excepto que en lugar de salmuera sintetica se usaron muestras recien obtenidas de una salmuera geotermica real Igualmente cuando se eleva el pH de la salmuera se utilizo una solucian de hidraxido de calcio casi satushyrada en lugar de hidraxido de sodio

Estos experimentos se lIevaron a cabo en el laboshyratorio de la CFE en Cerro Prieto con la ayuda de J Fausto l Se usaron salmueras de los pozos M-14 Y M-30 porque estos pozos estan cerca del laboratorio La salmuera del M-30 se considero como la mas represhysentativa del campo en conjunto

Los pozos M-14 V M-30 ambos pozos en produccion Podujeron constantemente vapor y salmuera cuando se lIevaron a cabo estos experimentos Dichos pozos (y todos los otros de Cerro Prieto) tienen una I (nea de pequeno diametro para muestreo de salmuera y un separador Webre miniatura conectados al cabezal Cuando este sistema se encendio por vez primera se permitio que la salmuera fluyera a traves del mismo por 10 menos unos 20 minushytos antes de tomar la muestra Dicha muestra se guardo en una botella al vaclo que se habfa enjuagado primero tres veces con salmuera caliente para elevarla a la temperashytura de la sal muera En cuanto se tomo la muestra se reshygistraron el tiempo y la temperatura y se extrajeron las dos primeras alicuotas del 1-ml para determinar la silice total y el MAS y se colocaron en frascos plasticos prepashyrados de 50 ml que contenfan hidroxido de sodio y moshyIybdato de amonfaco reactivo respectivamente La muesshytra se traslado con rapidez al laboratorio en automovil EI tiempo transcurrido entre el muestreo y su lIegada al lashyboratorio fue de alrededor de 8 minutos Una vez en el laboratorio la mezda completa se volco en un vaso vertedor preparado en un banG de agua caliente y se retiraron dos alicuotas mas de 1-ml para amllisis de silice

La sflice coloidal que se forma en la salmuera exshytrarda del pozo M-30 comenzo general mente a flocular y a asentarse para el momenta en que la salmuera se volshyco en el vaso vertedor Sin embargo la cantidad de sllice coloidal que permanecio en suspension despues de 30 mishynutos (gt10 mgl-l) no serfa probablemente recomendashyble para reinyeccion La figura 4 muestra cuan bien el agregado de agua de cal a la salmuera de M-30 remueve la silice suspendida EI agregado de 2 mgI-1 de Magnifloc 573 a la salmuera de M-30 no redujo la s(ljce slJspendida a niveles aceptables (los resultados de estos experimentos no se muestran aqu f)

La sllice coloidal en la salmuera no modificada de M-14 no flocula aun despues de que la salmuera se incuba durante una hora esto se muestra en la figura as como el efecto de agregar un poco de agua de cal a la salmuera de M-14 Esta pequena cantidad de cal se agrego por equishyvocacion pero no hubo oportunidad de repetir el experishy

mento con mas cal Aun esta pequena cantidad de cal provoco que la mayor parte de la sflice coloidal floculara y seasentara Agregar mas cal por ejemplo 30 mgmiddotP reduciria probablemente la sflice suspend ida en la salmueshyra de M-14 a niveles aceptables La pequefHsima diferencia entre los valores de pH pre y postratamiento en la figura 5 se debe probablemente a un error experimental

EI agregado de 2 mg 1-1 de Magnifloc 573 C a la salmuera de M-14 redujo la concentracion de sllice suspenshydida a un nivel aceptable (Fig 6)

EI volumen y la calidad de los datos que se obtuvieshy

593

ron en el campose redujeron por la dificultad de trabajo en un laboratorio no familiar y por restricciones de tiemshypo Sin embargo los resultados que se obtuvieron parecen confirmar los resultados del trabajo de laboratorio EI incrementodel pH parece funcionar correctamente con salmuera decualquier pozo en tanto que el Magnifloc 573 C solo funciona con la salmuera del M-14 este es exactamente el mismo modelo que aquel que se observo conlasdos salmuerassinteticas diferentes

RECOMEN-DACIQNPARA UNA PRUEBA DE PLANTAPltOTO

Se duda que las pruebas de banco con las salmueras sinshyteticas 0 con muestras de salmuera real puedan realizarshyse con provecho en el futuro Los resultados que aqu ( se informanson probablemente tan completos y convinshycentes como los que pueden obtenerse de este mododashydasJas limitaciones intdnsecas de las pruebas de banco y la variabilidad de las salmueras naturales Progresos posshyteriores vendran solo de los experimentos en la planta pimiddot lotodel campo de Cerro Prieto

La figura 7 es un diagrama esquematico de una planshyta piloto que permitira que los procesos de tratamiento de lasalmuera que indican los resultados de las pruebas se evaluen de un modo practicamente significativo Probablemiddot mente ser(a mas conveniente trabajar con una proporshycion de diseno deflujo de 5 a 10 m3 h- i Esta planta piloto semiddot paTece a una que usaron Rothbaum y Anderton (1975)

Esta planta piloto reproducirfa aproximadamente los pro~esos evaluados en nuestras pruebas de banco si operara sin recirculacion de lodo Lo mismo que en las pruebas de banco la funcion del paso de envejecimiento de Ia salmuera es dar tiempo para que la sllice disuelta se convierta en silice coloidal Sin embargo la floculacion parcialy elasentamiento de la silice coloidal en el tanque de envejecimientopodrfanocasionar problemas con la acumulacion delodo ah L Si se prueba que esto serla un serio problema podrfa remplazarse el paso de envejecimiddot mientoporrecirculacion de parte del lodo que sale del claiificadormiddot La enorme cantidad de sllice coloidal en el lodo reaccionarfa con la sllice disuelta en la salmuera y disminuirfa casi instantaneamente la concentracion de sllice disuelta en la mezcla resultante a su valor de estado estacionarioLa recirculaciondellodocon este fin se considera esencial en Miland California debido a la tasa bastante bajaque hay alia de polimerizacion de sflice en la salmuera(Quong et al 1978) Sin embargo la tasa de polimerizacion de sllice es bastante alta en Cerro Prieto como para que el envejecimiento de la salmuera sea una opcion razonable Serlan suficientes 5 minutos de envejecimiento

de lodo es mas factible que depositos de escamas se forshymen en la entrada del can~ de la salmuera y en la parte del tanque de envejecimiento proxima a esta EI tanque de envejecimiento deberfa disenarse teniendo en cuenta esto Por ejemplo deberfa ser posible disenarlo de modo tal que la salmuera entrante no este en contacto con parte alguna del tanque hasta unos pocos minutos despues de que ha entrado en el tanque y sea ya relativamente no reactivo

EI bosquejo del tanque de envejecimiento de la salshymuera en la figura 7 es una vista en planta La salmuera fluye de lado a lado alrededor de deflectores verticales Otra posibilidad es un tanque de flujo laminar vertical sin deflectores en el cual la salmuera entre por arriba y salga por abajo Lo que se necesita real mente es que el tanque de envejecimiento provea condiciones aproximashydas de reactor de flujo que el lodo no se acumule en el mismo y que sea facil de limpiar

En la figura 7 el clarificador el tan que de mezcla y el circuito de recirculacion de lodo se muestran c~mo componentes separados Si ultimamente se implantara un proceso que incluyera recirculacion de lodo estos tres componentes podrian remplazarse por un reactor-clarifishycador que combinara la funcion de los tres en una unidad simple Incluso si no fuera necesaria la recirculacion de lodo habrfa clarificadores disponibles que incorporaran la funci6n deltanquede mezcla como parte del clarifishycador La configuracion de componente separado que se observa en la figura 7 se recomienda para una planta piloto porque permite la maxima flexibilidad experimenshytal y la mas clara separacion de los efectos de los varios componentes sobre el funcionamiento completo del proceso

La planta piloto va a necesitar un sistema de ali menshytacion quimico dual que permita evaluar el incremento del pH y la adicion del floculante sintetico Lo que necesita el sistema de alilflentacion para el floculante sintetico es un tanque de almacenamiento para la solucion de flocushylante y una bomba medidora

En una planta de tratamiento de salmuera a gran esshycala el incremento de pH se alcanzarfa agregando a la salshymuera lechada de cal hidratada 0 un gel La lechada 0 gel muy probablementese producirian aliI mismo al hidratar cal viva en un reactor apropiado Sin embargo no serla practico efectuarlo en pequefia escala en la planta piloto Un metodo infinitamente mas simple y mas conveniente para producir lechada de cal hidratada seria mezclar somiddot luciones medidas de NaOH y CaCIz justo antes de que enmiddot tren al tanque de mezcla Esto produciria una suspension de Ca (OHh en la solucion de NaC Experimentos adeshycuados lIevados a cabo en el laboratorio indican que no

Cuando se opera la planta piloto sin recirculaci6n existirfa problema de formacion de escamas de Ca(OH)2