Chemical-Physical Pretreatment of Phosphogypsum Leachate

Paolo Battistoni,*,† Enrico Carniani, † Valeria Fratesi,† Pietro Balboni,‡ and Pierluigi Tornabuoni‡

Institute of Hydraulics and Transportation Infrastructures, Engineering Faculty, Marche PolytechnicalUniVersity, Via Brecce Bianche, 60131 Ancona, Italy, and Syndial, P.za Boldrini 1,San Donato Milanese (MI), Italy

In landfills for phosphogypsum disposal, a large volume of leachate wastewater generated by the residualprocess water and by the rainfall percolation through the solid material must be managed. The leachate shallbe properly treated before its discharge to surface water or its reuse; it has a very low pH and high concentrationsof hazardous contaminants such as fluoride, nitrate, phosphate, ammonia, and heavy metals. On the basis ofa technical-economical evaluation, a membrane filtration process for the treatment of the leachate wasproposed. The membrane process produces a very good quality permeate and a concentrate that shall bereinjected in the stack. From the results of a calculation model that defines the thermodynamic saturationconditions of the salts, it can be expected that the high salt content of the wastewater will reduce the performanceof the membranes in terms of low recovery, higher operative pressure, and fouling problems. Moreover, asthe concentrate obtained in the membrane process results in metastable oversaturated conditions, it will causea rapid occlusion of the injection wells. In this study, a pretreatment of the leachate wastewater with limewas defined through laboratory batch tests, to obtain, under controlled conditions, the crystallization of lowsolubility salts and their separation from the effluent before entering the membrane plant. When operating ata pH of 6.7 with an addition of 10 g/L of lime, very high fluoride and phosphate removals were obtained(96% and 80%, respectively), similar to those predicted by the mathematical model. This approach was proposedas a pretreatment at demonstrative scale to evaluate its feasibility in a membrane process application.

Introduction

Phosphogypsum (PG) is the main byproduct generated duringthe wet process manufacture of phosphoric acid and fertilizers,wherein rock phosphate and sulfuric acid are used as rawmaterials. PG is an acid product composed of over 90% ofcalcium sulfate and lower percentages of silicon, aluminum,phosphate, and fluoride. PG is removed by the filtration step inthe phosphoric production process and is slurried in processwastewater onto one or more impoundments located nearby theproduction site, known as gypsum stacks. In the impoundments,the solids are allowed to settle and the process water is recycledto the plant to be reused. Periodically, the slurry is divertedfrom one impoundment to another and the first is allowed todry. The dried phosphogypsum is used to build up the dike thatforms the impoundment, and then it is returned to active service.In this manner, the stack increases in height and accumulatesadditional phosphogypsum. The rainwater leaches through thepores of the solid and is collected with process water in theperimeter drainage ditches. The water collected shall be properlytreated to reduce its pollution level, according to the existingstandards, before its discharge to surface water or its reuse. Thephosphogypsum leachate (PGL) shows a low pH (∼2.0) andhigh concentrations of phosphate, sulfate, chloride, fluoride,ammonia, nitrate, and heavy metals (Cd, Cu, Ni, Fe, Mn, Al,Zn, and Sr), including low concentrations of suspended solids(high concentrations in the case of rainfall).1

Because there is a lack of literature information on PGLtreatments, several processes used for the treatment of waste-waters with similar characteristics were preliminarily examined.Biological treatments cannot be applied, because this sample is

essentially inorganic with a low content of organic carbon.Among the possible processes, the following treatments wereselected:

• chemical precipitation with lime at pH 12 and oxidation ofammonia with hypochlorite; the disadvantage of this process isthat the residual concentrations of fluoride, phosphate, andnitrate exceed the limits for the discharge into surface waterand that the treatment produces a great amount of sludge thatis hard to dewater.

• natural or forced evaporation; this process is expensivebecause of the high energy cost and the large amount of coolingwater required. Moreover, the special materials necessary forthe distillation apparatus contribute to increase the treatmentcosts. A cost reduction may be obtained by using vapor for theleachate stripping.

• membrane filtration process; the membrane process pro-duces a good quality permeate and a concentrate that can bereinjected in the stack. Because of the high salt content of thePGL, a complex multistage system will be required; in addition,the presence of low-solubility salts (phosphates, fluorides, andsulfates) will cause membrane scaling and higher operative costs.

From a technical-economical evaluation of the previousprocesses, membrane filtration in situ is the only treatment ableto reach, although at high costs, the standard limits requiredfor the discharge of the final effluent into surface water or itsreuse for industrial or irrigation purposes. Moreover, theconcentrate reinjection in the landfill increases the technical andeconomical feasibility of the PGL treatment. In a membraneplant, it is necessary to decrease salts concentration beforefeeding, to increase the performances, and to reduce theoperative costs and the fouling problems.

The purpose of this study is to define, through laboratoryand pilot plant tests, a pretreatment of PGL with lime to obtain,under controlled conditions, the crystallization of low-solubilitysalts and their separation from the influent as a concentrate

* To whom correspondence should be addressed. E-mail:p.battistoni@univpm.it.

† Marche Polytechnical University.‡ Syndial.

3237Ind. Eng. Chem. Res.2006,45, 3237-3242

10.1021/ie051252h CCC: $33.50 © 2006 American Chemical SocietyPublished on Web 03/31/2006

sludge that is easily filterable.2 The precipitation process willalso reduce the salt content of the concentrate and, consequently,reduce the problems of occlusion of the reinjection wells.

Materials and Methods

Landfill Characteristics and Leachate Production. Theinvestigated phosphogypsum landfill has not been used since1992. The quantity of accumulated PG was some 4 000 000 tons,and the extension of the landfill area is∼55 hectares with anaverage depth of 6 m. The stack closure plan providescontainments and covering facilities to avoid the risk ofenvironmental contamination and to reach the minimization ofthe rainfall water infiltration.

The rainwater leachate and the process water accumulatedin the stack amount to∼250 000 m3. Moreover, a substantialreduction of the water depth at the bottom of the stack shall becarried out by draining∼50% of the existing wastewater in 2or 3 years.

Phosphogypsum Leachate.PGL samples used for laboratoryand pilot plant tests were collected on different dates by meansof submersible pumps from the perimeter drainage ditches ofthe stack. The analyses of the various leachate samples wereperformed according to standard methods.3

Laboratory Procedures.Several laboratory batch tests werecarried out to evaluate the best conditions for the precipitationof low-solubility salts with lime. The quantity of lime to beadded and the pH conditions were selected on the basis of thestoichiometry for the precipitation of low-solubility salts andof the results of the equilibrium model simulations.4 The testswere performed by introducing 1 L of PGL in a 2-Lglass beakerand submitting the sample to stirring using a vertical variablemixer IKA-WERK model RW 20 DZM and a stainless steelblade stirrer. The weighed quantity of lime powder was addedstep by step, maintaining the sample under fast agitation topermit the dispersion of the lime. For the pH correction at theprefixed value, a measured volume of 40% NaOH solution wasadded from a buret and the pH was determined using a pH-meter. After that, the suspension was maintained under slowagitation for∼1 h to permit the formation and the growth ofthe crystals; then the sample was left to settle under staticconditions to allow the solid separation. Finally, the sample wasfiltered on a paper filter, and the filtrate was analyzed for thevarious ionic species. The test temperature was maintainedconstant at∼20 °C. The solid fraction retained on the filterwas air-dried and weighed. A little quantity of air-dried samplewas introduced into a volumetric flask and dissolved in HClconcentrated for the quantitative analysis of the various ionicspecies. The analyses of Na, K, NH4, Cl, SO4, PO4, and NO3

were performed by using an ionic chromatograph DIONEXmodel DX-20. Heavy metals were determined by ICP (induc-tively coupled plasma spectrometer). An X-ray diffractometerPhilips model PW 1730 was used to analyze the crystallinecompounds contained in the solid fraction.

Particle-Size Distribution. The particle-size distribution ofthe air-dried solid fraction was performed by sieving using aseries of sieves with openings in the range 0.037-0.5 mm. Thesieves were arranged in downward decreasing mesh size andmechanically vibrated by an automatic vibrating machine for afixed period of time (∼15 min); the weight of the solid retainedon each sieve was measured and converted into percentcumulative undersize.5 From the previous data, it was possibleto design the sieve distribution curve that permits one to clearlyrepresent the sample size and its median diameter correspondingto 50 wt % of cumulative undersize.

Chemical Equilibrium Modeling. To establish the optimalpH for the precipitation of low-solubility salts, while minimizingthe coprecipitation of other crystalline phases, a calculationmodel named Visual MINTEQ (Version 2.02, EPA 1991) wasused.4 This model, based on the thermodynamic equilibriumbetween the species, allowed the creation of a personal databasein which the possible solid phases are inserted with theircharacteristic values of solubility product (Ksp), enthalpy, andstoichiometric coefficients. With this database, the program cancalculate the equilibrium concentrations of dissolved andprecipitated ionic species based on the input of component-ionconcentrations, pH, temperature, and ionic strength. In this study,a custom thermodynamic database was created for fluorite,fluorapatite, vivianite, CaHPO4‚2H2O, MgF2, hydroxyapatite,gypsum, and struvite.

Gravitational Behavior Test. Flocculant and hinderedsedimentation velocity was measured in a column settling test(1 L) where the surface separation between liquid and solidphases was time-followed.6

Results and Discussion

Leachate Chemical-Physical Characteristics. Leachatechemical-physical characteristics are reviewed in Table 1. Thedata show a quite constant composition, mainly characterizedby the following:

• acid condition (pH 2.3) and high dissolved solid content(TDS ) 26 500 mg/L) with a low concentration of organiccompounds (TOC) 23 mg/L);

• high content of sulfate, chloride, and fluoride (respectively,4 100, 4 200, and 800 mg/L) and a very high concentration ofphosphate (PO4 ) 16 000 mg/L).

The cationic composition is mainly constituted by Ca, Na,and Mg ions. Besides the macropollutant, a micropollutantpresence of heavy metals as As, Al, Cu, Cd, Ni, Zn, Cr, andPb, linked to phosphorite used as a mineral in phosphoric acidproduction, was determined (Table 2).

On the basis of their solubilty products, the possible saltsthat can be formed in an acid pretreatment process of PGL are

Table 1. Phosphogypsum Leachate Macropollutants

parameter m.u. average value standard deviation

conductivity µS/cm 24 070 3 730pH 2.33 0.14TSS mg/L 38 21TDS mg/L 26 500 1 268TOC mg/L 23 22Silica mg/L 317 54NH4 mg/L 1 145 265F mg/L 767 359NO3 mg/L 128 62PO4 mg/L 15 712 2 172Cl mg/L 4 177 1 051SO4 mg/L 4 088 471Ca mg/L 1 268 139Mg mg/L 480 75Na mg/L 2 712 621K mg/L 381 69Fe mg/L 50 15Sr mg/L 20 4

Table 2. Phosphogypsum Leachate Micropollutants

parameter m.u. range parameter m.u. range

As mg/L 1-2 Zn mg/L 5-15Al mg/L 21-55 Cr total mg/L 1-2Cu mg/L 0.4-1.3 Cr VI mg/L <0.02Cd mg/L 1-3 Pb mg/L 0.03-0.5Ni mg/L 1-4

3238 Ind. Eng. Chem. Res., Vol. 45, No. 9, 2006

fluorite, fluorapatite, hydroxyapatite, CaHPO4, vivianite, MgF2,struvite, and gypsum (Table 3). From the leachate composition,it can be observed that the amount of the main cations does notcover the stoichiometric request of different salts that can beformed. In particular, the concentration of calcium present inthe PGL sample is sufficient for the complete precipitation offluorite, but a large addition of this cation is required to obtainthe phosphate precipitation as fluorapatite.

Chemical Equilibrium Modeling. To define the best opera-tive conditions for the execution of laboratory batch tests, severalmodel simulations were carried out on laboratory PGL sampleat different pH values to calculate the possible solids inequilibrium with the dissolved phase and the performance inthe removal of low-solubility ionic species. The possibleequilibrium solids reported in Table 3 were selected on the basisof their solubility products and of stoichiometry and mineralcomposition of the water sample.The results of chemicalprecipitation simulations reported in Figure 1a show that theleachate sample at pH 2.2 is in thermodynamic oversaturatedcondition as fluorite concentration. At pH 4.5, the concentrationof fluorite rapidly decreases because of its solubity at higherpH values, while fluorapatite and MgF2 concentrations progres-sively increase; at neutral and alkaline pH, fluorite is absent,while fluoride and phosphate are present mainly as fluorapatite

and MgF2 (see Figure 1a). At pH> 7.5, struvite increases itsconcentration to the detriment of MgF2. From these results, itwas observed that fluoride removal is satisfactory at pH< 5.0and decreases with increasing pH; on the contrary, phosphateremoval is very low, even at higher pH values (maximum 27.5%at pH 9.0), because of the lack of calcium content (see Figure1b).

From a qualitative point of view, the addition of 10 g/L oflime, equivalent to 5410 mg/L of calcium, produces at differentpH values the same type of precipitate observed in the previoustest with the exception of gypsum, which is present only at pH< 3.0-3.5 (Figure 2a). As expected, the addition of calciumincreases considerably,>5×, the quantity of fluorapatite formed,while the concentrations of MgF2 and struvite do not changesignificantly. As a consequence, the removal of phosphatebecomes 85% at alkaline pH, while those of Mg and NH4 remainapproximately the same (Figure 2b). Fluorite removal decreasesprogressively at pH> 5.5-6.0 because of an increase ofsolubility of CaF2 in alkaline conditions. In conclusion, theoperative pH values just simulated (6.7, 7.5, and 8.9) with theaddition of a constant quantity of lime (10 g/L) must be testedkinetically with the aim to individuate the best conditions. Ofcourse, the success will be dependent on fluoride and phosphateremoval and reagent costs, together with the quantity and thesedimentability of the sludge formed.

Laboratory Tests. On the basis of simulation results,laboratory tests of chemical precipitation with lime wereperformed at three different pH values, 6.7, 7.5, and 8.9, inaccordance with the procedure previously described. To all thesamples, 10 g/L of lime together with a certain volume of 40%NaOH solution for pH correction, if necessary, were added. The

Figure 1. (a) Real PGLsmodel calculation of equilibrium solids at differentpHs; (b) real PGLsmodel calculation of removal of various ionic speciesat different pHs.

Table 3. Solubility Products of Solids at 25°C

salt formula pKsp

fluorite CaF2 10.5fluorapatite Ca5F(PO4)3 59.0hydroxyapatite Ca5OH(PO4)3 44.3monohydrogen calcium phosphate CaHPO4 19.3vivianite Fe3(PO4)2‚8H2O 37.8magnesium fluoride MgF2 8.1struvite MgNH4PO4‚6H2O 12.6gypsum CaSO4‚2H2O 4.6

Figure 2. (a) Real PGL+ 10 g/L of limesmodel calculation of equilibriumsolids at different pHs; (b) real PGL+ 10 g/L of limesmodel calculationof removal of various ionic species at different pHs.

Ind. Eng. Chem. Res., Vol. 45, No. 9, 20063239

analyses performed on the filtrates of the various samples arereported in Table 4. The results show that the filtrates have alow content of fluoride, phosphate, calcium, and magnesiumcompared to the input data.

As expected by the model, the highest fluoride removal isobtained at pH 6.7 (96%), while this percentage decreases withincreasing pH. Ammonia reduction from the solution is es-sentially due to struvite precipitation, and its removal is quitesatisfactory at pH> 7.5 (∼50%). Phosphates precipitate in allthe conditions tested, initially as fluorapatite and, subsequently,at pH values>7.0, also as struvite; the best experimentalphosphate reduction is obtained at pH 6.7 (80%).

Calcium removal is>94% at all the test conditions, notwith-standing the addition of 5410 mg/L of calcium as Ca(OH)2,while magnesium concentration decreases with increasing pHbecause of the precipitation of MgF2 and struvite; it is equal tozero at pH 8.9. Iron and the other heavy metals, with theexception of As and Zn, are substantially removed in everycondition. The choice to add calcium as lime determined inevery test a meaningful removal of salinity (48%) at pH 6.7;lower removals were observed at higher pHs, 7.5 and 8.9,because of the NaOH addition.

Table 5 reports the chemical analyses performed on the solidphase after filtration and air-drying of the sample to avoidammonia loss. Only salts obtained at pH 6.7 and 7.5 are

analyzed, because they are the most interesting as performances.The results, in accordance with those observed for the filtrates,indicate the presence of high quantities of calcium andphosphates, together with smaller concentrations of fluoride,sulfate, magnesium, and ammonium. Also, heavy metals,particularly iron and manganese, are present. The quantities ofthe precipitated salts measured as air-dried solids were∼20 gTS/L. The water content of the air-dried sample measured at 105°C was in the range 15-25%.

The mineralogical analyses of solids performed by X-raydiffractometry (Table 6) show that, at pH 6.7, fine crystals ofcalcium phosphate, together with smaller quantities of struvite,hydroxyapatite, and gypsum, are present. The analyses relevantto the solid obtained at pH 7.5 indicate a reduction of crystallineminerals and an increase of struvite, hydroxyapatite, and gypsumpercentage with respect to calcium phosphate. At pH 8.9, a solidthat is substantially amorphous is observed. The tests confirmedthat, by increasing the operative pH, the percentage of amor-phous precipitates increases. As a consequence, at high pHvalues, the solids obtained in the chemical-physical pretreat-ment proposed will be characterized by a low sedimentationvelocity. On the basis of the above results and also consideringthe high quantities of NaOH necessary for pH correction, whichnegatively affects the performance of the membranes, it wasdecided to drop definitively the test at pH 8.9.

In Table 7, a comparison between the composition of theprecipitated salts obtained by the thermodynamic model andthat calculated by the analyses of solids (Table 5) is reportedfor pH 6.7 and 7.5 samples. For the calculation of saltcomposition, it was assumed, according to the model, that allammonia should precipitate as struvite and the remaining Mgshould combine with fluorides as MgF2. Calcium should

Table 4. Analyses of Filtrates at Different pHs

parameter m.u. leachate pH) 6.7 E% pH) 7.5 E% pH) 8.9 E%

lime g/L 10 10 1040% NaOH sol. mL/L 2.25 10 16volume increment % 0.23 1.0 1.6conductivity µS/cm 20 140 18 100 22 000 25 400pH 2.2 6.7 7.5 8.9TSS mg/L 4.9TDS mg/L 27 000 14 000 48 18 000 33 19 000 29NH4 mg/L 849 651 23 405 52 414 51F mg/L 1 171 52 96 203 83 229 80NO3 mg/L 69 69 65 6PO4 mg/L 13 237 2 671 80 4 073 69 3905 70Cl mg/L 3 025 3 074 3 285 3 176SO4 mg/L 3 545 3 123 12 3 956 3 779Ca mg/L 1 325 83 94 26 98 48.5 96Mg mg/L 431 120 72 13 97 0 100Na mg/L 3 346 4 006 6 049 7 172K mg/L 346 332 4 201 42 201 42Fe mg/L 63.4 1 99 0.2 100 0.32 99Mn mg/L 11 0.4 96 <0.1 100 <0.1 100As mg/L 1.1 1.1 1.1 1.1Al mg/L 21.5 1.9 91 1.7 92 <0.1 100Cu mg/L 0.4 <0.1 82 <0.1 <0.1Cd mg/L 2.4 0.03 99 <0.01 100 <0.01 100Ni mg/L 2.5 0.17 93 <0.03 99 <0.03 99Zn mg/L 12.3 10 19 10 19 10 19Cr mg/L 0.9 0.12 87 <0.06 <0.06Pb mg/L <0.1 <0.1 <0.1 <0.1

Table 5. Analyses of Air-Dry Precipitates at Different pHs

parameter m.u. pH) 6.7 pH) 7.5

moisture at 105°C wt % 15.1 24.8moisture at 180°C wt % 20.0 31.5quantity of solid air-dry kg/m3 20.2 20.5NH4 mg/kg 4 992 8 062F mg/kg 41 245 33 261NO3 mg/kg 0 0PO4 mg/kg 411 783 318 092SO4 mg/kg 14 718 11 841Ca mg/kg 252 528 235 359Mg mg/kg 12 973 16 165Na mg/kg 5 614 11 766K mg/kg 1 297 1 298Fe mg/kg 2 818 3 177Mn mg/kg 403 482

Table 6. Diffractometric Analyses of Precipitate

testCaHPO4‚2H2O

wt %struvitewt %

hydroxyapatitewt %

gypsumwt %

pH ) 6.7 89 4 4 3pH ) 7.5 64 11 14 11pH ) 8.9 sample essentially

amorphous

3240 Ind. Eng. Chem. Res., Vol. 45, No. 9, 2006

partially combine with sulfates to form gypsum and withphosphates and fluorides to form fluorapatite. Even though themodel calculates the thermodynamic equilibrium concentrationswhile the analyses represent the results of a kinetic precipitationprocess, it can be observed that the kinetic and thermodynamicdata are quite comparable and confirm that, at pH 6.7,phosphates are precipitated mainly as fluorapatite while, athigher pH, struvite concentration increases. At both pH values,the MgF2 is quite low (∼2%). A little quantity of gypsum seemsto be present only in the laboratory samples, while the modelcalculation indicates that this salt should precipitate at pH values<3.5. The presence of gypsum in the laboratory samples wasconfirmed also by the diffractometric analyses reported in Table6.

A comparison between the removal of various ionic speciesin the filtrates at pH 6.7 and 7.5 measured by laboratory testsand calculated by the model is reported in Table 8. The resultsconfirm that, for fluoride, phosphate, calcium, magnesium, iron,and manganese, there is a quite good agreement between thetwo figures. The differences for NH4 might be caused bypossible ammonia losses during the execution of the laboratorytests, expecially at higher pH. The comparison with sulfateswas not considered, because the thermodynamic model doesnot provide the presence of gypsum at pH> 3.0-3.5.

The particle-size distribution of two different solid samplesobtained at pH 6.7 was performed by dry sieving, as previouslydescribed. The results were reported in Figure 3 by putting inthe ordinate the percent cumulative retained at various sievesas a function of particle size in mm. The sample is mainlycomposed of fine crystalline particles with a median diameterof ∼0.085 mm.

The gravitational behavior of the precipitate separated at pH6.7 and air-dried has been determined by a laboratory sedimen-tation test using a suspension of 10 000 mg/L. The test has beenperformed by introducing in a glass cylinder of 1-L capacity aweighed amount of precipitate and measuring at different timesthe depth of the interface between the suspension and the clearsupernatant water. Two different types of test were carried out:the first followed the clean separation surface between liquidand solids defined as the lowest velocity and the second followedthe sedimentation of bulk solids without considering if fewparticles remained in suspension, defined as the highest velocity.The lowest velocity was 0.003 m/h, and the highest was 0.33m/h (Figure 4).

Pretreatment Process Definition.Results obtained in chemi-cal-physical tests can be utilized to define the best configurationand design conditions for pretreatment before the membraneprocess. A classical flow scheme of a PGL treatment plant ispresented in Figure 5, where the chemical-physical processhas the role to reduce the particulate and soluble solids in PGLwith the aim to reduce pressure or energy consumption andincrease working time for the membrane process. The membraneprocess must be performed according to two steps in series.The first step uses a nanofiltration membrane to preserve thereverse osmosis (RO) membrane from clogging, and the secondstep uses further two steps of RO (RO1 and RO2) membraneswhich are necessary to produce a permeate to dischargeaccording to CE DIR 91/271.

In this scheme, the chemical pretreatment must be designedto perform the coagulation-flocculation process and precipitatesedimentation; the sedimentation tank can be designed using ahydraulic load of 0.33 m3/(m2 h), and the supernatant effluent

Table 7. Laboratory and Model Prevision of Main PrecipitatedSalts

salt

lab testpH 6.7%w./w.

modelpH 6.7%w./w.

lab testpH 7.5%w./w.

modelpH 7.5%w./w.

struvite 9.0 3.6 16.7 8.8MgF2 2.1 3.0 2.1 1.5fluorapatite 82.3 92.5 72.8 88.7gypsum 3.5 0 3.2 0

Table 8. Laboratory and Model Prevision of Main Ionic SpeciesRemoval

pH ) 6.7% removal pH) 7.5% removal

parameter lab. model lab. model

NH4 23.2 10.2 52.3 26.0F 95.6 90.8 82.7 77.3PO4 80.0 76.5 69.2 81.9Ca 93.7 100 98.1 100Mg 72.2 77.0 97.0 95.8Fe 98.9 99.8 99.6 100Mn 96.4 100 99.7 100

Figure 3. Particle-size distribution of solids at pH 6.7.

Figure 4. Gravitational test of precipitate.

Table 9. Operative Parameters of Precipitation Pretreatment

lime addition kg/m3 10NaOH addition

(40% solution)L/m3 2.25

operative pH 6.7decanter surface

hydraulic loadm3/(m2 h) 0.33

decanter solid flux kgTS/(m2 h) 6effluent filtration vacuum filtration on filter

screen with cellulose precoatdry sludge quantity (dry) kgTS/m3 20.2sludge TS % 45sludge quantity kg/m3 45sludge thickening anionic

polyelectrolyte addition% TS 0.4

Ind. Eng. Chem. Res., Vol. 45, No. 9, 20063241

will be filtered in a vacuum filter before feeding the membraneprocess, with the aim to remove 60-70 mg/L of suspendedsolids lost in sedimentation. Design and operative parametersderived from this work are reviewed in Table 9, where two datapoints are particularly meaningful: the amount of lime to beadded (10 kg/m3) and that of sludge to be handled (45 kg/m3).These two conditions suggest, for their technical and economicalimpact, a field experimentation in a large-scale pilot plant forprocess optimization. Furthermore, the disposal of membraneconcentrate in landfill, from which leachate is provided,frequently determines the well clogging due to the supersatu-ration condition of the concentrate. The sludge precipitation andrecovery proposed by chemical process must be considered asa reduction of salt landfill feedback and well clogging.

Conclusions

The paper has examined the chemical-physical treatment ofphosphogypsum leachate. The main results are as follows:

• The chemical composition of leachate suggested thepossibility to form different types of salts, but its acidity requiresthe addition of an external calcium source to gain a higher pHand thermodynamic solubility conditions of the salts.

• The choice of Ca(OH)2 can be considered optimum toreduce total dissolved solids according to two mechanisms:precipitation and neutralization.

• The best operative conditions can be tested using the VisualMINTEQ model. The selected pH values represent a compro-mise between fluoride and phosphate removal.

• The experimental results substantially confirm those pre-dicted by the thermodynamic model, indicating, at an operativepH of 6.7 and with 10 g/L of lime addition, 96% of fluorideand 80% of phosphate removal; besides, heavy metals arecompletely precipitated.

• A modified flow scheme of the membrane process toproduce an effluent congruent with CE 91/271 can be individu-ated using experimental results, but a mass balance for sludgeproduction and chemical addition is suggested to verify on alarge-scale pilot plant its feasibility and convenience.

• The impact of membrane concentrate feedback in landfillwells must be carefully considered, as the precipitate formationin the chemical-physical process represents a consistent reduc-tion of mass load to be reinjected in the wells.

Literature Cited

(1) Luther, S. M.; Poulsen, L.; Dudas, M. J.; Rutherford, P. M. Fluoridesorption and mineral stability in an Alberta soil interacting with phospho-gypsum leachate.C. J. Soil Sci.1995, 83-91.

(2) Battistoni, P.; De Angelis, A.; Prisciandaro, M.; Boccadoro, R.;Bolzonella, D. P removal from anaerobic supernatants by struvite crystal-lization: Long-term validation and process modelling.Water Res.2002,36, 1927-1938.

(3) APHA. Standards methods for the examination of water andwastewater, 16th ed.; American Public Health Association: Washington,DC, 1985.

(4) Environmental Protection Agency (USEPA).Visual Minteq, AGeochemical Assessment Model for EnVironmental Systems, Version 2.02;EPA/600/3-91/021; KTH, Swedish Royal Institute of Technology: Stock-holm, Sweden, 1991.

(5) Battistoni, P.; Paci, B.; Fatone, F.; Pavan, P. Phosphorus removalfrom supernatants using packed and fluidized-bed reactors.Ind. Eng. Chem.Res.2005, 44, 6701-6707.

(6) Dick, R. L.; Ewing, B. B. Evaluation of activated sludge thickeningtheories.J. Sanit. Eng. DiV., Am. Soc. CiV. Eng.1967, 93, SA-4.

ReceiVed for reView November 11, 2005ReVised manuscript receiVed March 3, 2006

AcceptedMarch 3, 2006

IE051252H

Figure 5. Membrane plant block diagram with precipitation pretreatment.

3242 Ind. Eng. Chem. Res., Vol. 45, No. 9, 2006