a review of real-time particle size analysers

7/30/2019 A Review of Real-time Particle Size Analysers

1/11

A r e v i e w of real-time particle size analysersSYNOPSIS

A . L . H INDE,. B .S c., A .R .C .S ., D .I.C ., P h.D .

T he physical prin ciples o f different sizing tech niques are critically exam ined in o rder to evalu ate the suitabilityof the tech niques for real-tim e particle size assessm ent. E xam ples o f both successful an d unsuccessful attem pts atprod ucin g an acceptable real-tim e sizer are describ ed in ord er to delin eate the difficu lties invo lv ed in m akin g rapidand accurate size m easurem ents. O nly tw o of the exam ples described have prov ed to be suitable for o n-line particlesize m onitoring, nam ely the "M intekjRSM On-stream Slurry Sizer" m anufactured by Cartner Group, Ltd., and the"PSM System 100" m ade by A utom etrics Co. Both these com mercial system s are capable of assessing particle sizew ith an accuracy com parable with that obtained by routine sieving and w eighing. The "PSM System 100" has beenu sed su cc es sfu lly fo r a utoma tic c on tro l o f g rin din g circ uits . E co nomic b en ef its d eriv ed from th is sy stem, o pe ra tin gunder rigorous plant conditions fo r several m onths, have pro ved that real-tim e size analysers are of significan t valueto th e m in in g a nd m eta llu rg ic al in du str y.SINOPSIS

Die fisiese beginsels van verskillende grootte bepalings tegniek w ord krities ondersoek om sodoende die ge-skiktheid van die tegnieke vir werklike tyd korrelgrootte waardasies vas te stel. Voorbeelde van beide sukses-volle en onsuksesvo lle p ogings o m 'n aanvaarbare werklike tyd groottemetingsinstrument te produseer wordbeskryw e om sodoende die m oeilikhede uit te beeld w at gepaard gaan m et vinnige en akkurate grootte bepalings.Slegs twee van die voorbeelde beskrywe was geskik gevind vir op-Iyn korrelgrootte waarnem ing, naamlik die"M intekjRSM O n Stream Slurry Sizer" vervaardig deur Cartner Group, Ltd., en die "PSM S ys tem 1 00 " v erv aa r-dig deur Autometrics Co. Beide hierdie kommersii:He sisteme is instaat om korrelgrootte weer te gee met 'nakkuraatheid wat vergelyk kan word met die verkry deur roetine siffing en weging. Die "PSM System 100" isa lre ed s s uk se sv ol a an gewe nd v ir o utoma tie se k on tro le v an m eu le k rin glo pe . E ko nomie se v oo rd ele v oo rtv lo eie nd euit hierdie sisteem in werking in aanlegte oor verskeie m aande onder straww e toestande het bew ys dat w erkliketyd grootte analise van betekenisvolle w aarde is vir die m yn en m etallurgiese industriee.

INTRODUCTIONIn the mining and metallurgical industry, particle

size effects can influence the economics of extractingmetals from their ore-bodies. In order to liberate thevaluable mineral component it is necessary to grind theore to a fine size. The desirable size distribution of theproduct after comminution is determined by suchfactors as the value of the metal, the cost of grinding,and the kinetic response of milled product to chemicaland physical processes. The response of the processequipment to variations in particle size is also an im-portant parameter. The exact effect of particle size onthese factors and on the overall efficiency of the ex-traction process is not well understood. Therefore, itis generally believed that a rapid and accurate method ofsizing might prove to be a useful diagnostic tool instudies of the relationship between the numerousparameters involved in the extraction process.

Of particular importance is the need for rapid sizeanalysis to control m illing circuits. Surges in the circuitslead to fluctuations in particle size and result in losses inp ro du ctiv ity an d p ro fitab ility .

A study of the literaturel, 2 on the subject has shownthat research directed towards the development of real-time size analysers for the industry has been pursued forat least a decade. However, little success has beenachieved and few real-time sizers have been developed tothe stage at which commercial manufacture was war-ranted. A lthough many rapid sizing methods can giveresults sensitive to particle size variations, the specifi-cations of acceptability are stringent in that methodsmust have an accuracy as reliable as that of the routine*Research O rganisation, Cham ber of M ines of South Africa.~~ MARCH 1 973

sieving analyses normally practised by the industry.This criterion for accuracy is difficult to achieve. In orderto delineate these difficulties, the physical principles ofdifferent sizing methods have been examined critically.The principles upon which the methods are based are:(a) measurement of slurry viscosity.

(b) sedim entation and elutriation techniques.(c) electro-optical scanning.(d) optical grating.(e) light diffraction.(f) sieving techniques.(g) size segregation in a helical tube.(h) ultrasonic attenuation.A lthough the list is by no means complete it is thought

that it contains examples of most of the physical princi-ples which numerous potential real-time size analysershave utilised or attempted to utilise.SIZE ANALYSIS BY MEASUREMENT OF SLURRY

VISCOSITYIt is well known that the rheological properties ofvery dilute slurries are dependent upon the volumeconcentration of the solids present. The relative vis-cosities of such dilute suspensions can be represented bythe fam iliar E instein equation:

1)r=l+kOwhere 1)r=relative viscosityk =shape factor constant0 =fraction by volume of solids

It is generally believed that particle size has littleinfluence on the rheological properties of very dilutesuspensions. However, as the concentration of solidsincreases, electro-viscous and other interparticle effectsbecome significant. Although a satisfactory theory has

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY


2/11

Particle mesh 5;70 [I"'")20 30 40 50 50 70 8090 lOG .. 20 0

~~' ,,~ ..'~. '/;/j/ //,% . I /~5 0 // /' I~ "

/' / .. 1/.',: : r . @I~ 14 '

1

;20 I)J/L " /d!J/:. ,~/i:' ~,~I0J eE~X,~~.~, ~~. ~-0~ , , ' ,, 1,0 1,1 1,2 I,)

10 0f08070~ 60';;;

S pe ci fic g ra vit y o f slurryFig. I-Rheological properties of gold ore slurries

yet to be developed, it has been demonstrated3 that as aconsequence of these effects the fluidity of suspensionsdecreases as particle size decreases.

The influence of particle size distributions on therheological properties of gold ore slurries is shown inFig. 1. These results were obtained from measurementswith a capillary viscometer. It will be seen that changesin apparent viscosity of a slurry are very insensitive tosize variations, even at high solids concentrations. Thislack of sensitivity is a major disadvantage in sizeanalysis since extraneous factors, unless controlledrigidly, can produce apparent viscosity changes com-parable with those which result from expected changesin particle size. For example, large discrepancies canarise because of lack of homogeneity of the slurry,variations in temperature, and the presence of traceam ounts of chem ical reagents.

In spite of the inherent difficulties of the methodBrown4 attempted to show that monitoring of grindingcircuits by measurement of changes in viscosity couldbe a feasible proposition. He suggested the use of adensity-sensitive gauge in close proxim ity to the vis-cosity probe. The resulting signals would then be pro-cessed to yield a signal, dependent on average particlesize, which could be used to control equipment. Littlesuccess was achieved since accurate and meaningfulresults could not be obtained. Both theory and pre-lim inary attempts at practice thus make it clear that theobservation of changes of viscosity for monitoringparticle size effects is not likely to yield great success.METHODS OF SIZING BY SEDIMENTATION AND

ELUTRIATIONSedim entation and elutriation m ethods are based onthe property that solid particles w ill settle in a viscous

medium, such as water, at a velocity related to particlesize. The method is generally used for particles smallerthan standard sieve mesh sizes (less than 44 /Lm) sincethe settling velocity of such particles can easily berelated to particle size according to Stokes' law:

V-(ps-pw)d2g181]

where V = se ttling ve lo cityd =particle diam eterg =acceleration of gravity1] =viscosity of suspending m ediumps =density of solidspw=density of suspending m edium

Although larger particles do not obey this simple law,their settling velocities can nevertheless be calculated ordeterm ined empirically. The methods are thereforecapable in principle of coping with particles ranging insize from a few micrometres to a few millimetres. It isimportant to note from the Stokes' law equation thatsize is determined by the viscosity of the suspendingmedium which in turn is sensitive to changes in tempera-ture. Temperature control is therefore a necessarycondition for accurate size analysis, but it presentsproblems in rapid size assessment. It is important to notethat size is also determ ined by the density of the solidphase. In many situations ore-bodies can containseveral gangue m inerals of different densities. V ariationsin the relative proportions of these minerals can lead tochanges in average density and thus give inaccurateresults. A further disadvantage is that the equationholds for only very dilute suspensions, so that any real-time analyser must have available to it very largevolumes of clear water. Finally, it should be noted thatthe equation is valid only for spheres, and that theparticle sizes found by sedimentation methods aretherefore "spherical equivalent" sizes. For roughlycubical particles, it is not too difficult to relate sedi-mentation or elutriation "sizes" to sieve sizes, but forplate or rod-shaped particles the sedimentation methodmay yield particle sizes which differ significantly fromsizes estim ated by screening.The disadvantage of sedimentation methods for de-term ining cumulative size distributions is that theycannot be used on-line, and analysis must be done off-line with batch samples. A further disadvantage ofsedimentation methods is that considerable time isnormally needed to perform a size analysis. A sedi-m entation instrum ent is available com mercially5 (M icro-meritics Instrument Co.) by which the time for sizeassessment has been shortened considerably. This in-strument determ ines, by means of a finely collimatedbeam of X-rays, the concentration of particles re-maining in suspension at various sedimentation depthsas a function of time. To minim ise the time of analysisthe position of the sedimentation cell is changed con-tinuously so that the effective sedimentation depth isinversely proportional to elapsed tim e.

Elutriation methods are attractive in that they makeon-line size assessment possible. The concept behindsuch methods is that a stream of particles is fed into anelutriation column and separated into two streams at a

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY MARCH 1973 259


3/11

certain cut-off size. The cumulative oversize fraction isrelated to the mass flow rates in the feed and overflow.Because of the viscosity of the water flowing in theelutriation column a parabolic velocity front exists.This prevents the attainment of a sharp cut-point. Theeffect of the uneven cut is the removal of coarse abovethe theoretical cut-point while leaving behind some ofthe fines. Thus, although the separate fractions are notaccurately sized, the cumulative mass fraction is oftenreasonably close to the correct value. A similar com-pensating effect also applies if the density of individualparticles differs from the mean solids density. In spiteof the apparent simplicity of the elutriation techniqueits application to real-time size analysis has receivedlittle attention. Some of the problems likely to arise inapplication of the method stem from the need forcareful sample preparation. It is important to ensurethat the particles are suitably dispersed prior to theirentering the elutriation column. W ith water as thesuspending medium it may be necessary to preventagglomeration of fines by adding an electrolyte or byuse of ultrasonics. In order to prevent interparticle effectsinside the elutriation column it is imperative that thesolids concentration should be very low (at the most1 per cent by weight). For an on-line size analyser basedon this method such high dilutions give rise to problemsof statistics which make it difficult to monitor repre-sentative sam ples of the process stream.Nakajima, Gotoh and Tanaka6 describe an on-lineparticle size analyser based on an elutriation techniqueusing air rather than water as the suspending medium.The instrument was capable of automatically recordingcum ulative size distributions which agreed with sievingresults to within about 10 per cent. It is understoodthat the South African Atomic Energy Board is alsoattempting to develop an analyser based on elutriation.

SIZING BY ELECTRO-OPTICAL SCANNINGE lectro-optical scanning techniques involve autom aticcounting and sizing of individual particles. The method

has enjoyed considerable success for particles restrictedto a narrow size range (this method is utilised by the"Coulter Counter"). Unfortunately, crushed or finely-ground brittle material often has particle sizes spanningseveral orders of magnitude; and the number of smallerparticles greatly exceeds that of larger particles. Severeinstrumental problems arise when the closely-spacedsmall particles must be recognised by receivers ass ep ara te e ntitie s.

An example of an attempt at developing the methodfor mineral ore slurries has been described by Ricciand Cooper7. In their particle size analyser an upward-flowing slurry stream is scanned by a focussed horizontallight beam derived from a helium-neon laser. Scanningis achieved by reflecting the light beam off the surface ofan oscillating mirror. As the light beam traverses thesample its passage is interrupted by the presence of theparticles in the stream . The time and frequency ofinterception are related to the particle size distribution.These interruptions can be determ ined using a fastphotodiode, and with sophisticated electronics it is26 0 MARCH 19 73

..

3~

30 0E~.~ 25~~E022 0.~" 15 0

10

5 5 35 '---' 400 15 20 25 30Weight '/, + 10 0m.oh (oiovin g)

Fig, 2-Results of optical scanning w ith a laser beampossible to process the signals from the photodiode toyie ld th e pa rticle size distribution .

Results of prelim inary laboratory tests are shown inFig. 2. Although these results look prom ising, the proto-type instrument was unable to yield accurate resultswhen subjected to in-plant tests8. The presence ofminus 44-JLm material was found to be responsible forirreproducible results and output bore a direct relation-ship to the concentration of solids in the sample. Someimprovement in results was obtained by removing all ofthe m inus 44-JLm material and maintaining the solidsconcentration to between 0,5 per cent and 1 per centsolids by weight. In spite of such elaborate precautionsan accuracy of :t 8,8 per cent + 100 m esh was obtained.

OPTICAL GRATING M ETHODThis methodl has many of the characteristics of the

elutriation and scanning techniques, and consequentlysuffers from similar deficiencies. However, lessonslearnt from development of the method at the Chamberof Mines Research Laboratories are worthy of dis-cussion.In this method, which was based on the relationshipbetween particle size and settling velocities, particlestravel downwards in a stream of water flowing through atransparent flow cell at a velocity V 0 (2 cm . sec-I). T heparticles pass through an area illuminated by a col-limated beam of light and the shadows produced by theparticles pass over an optical grating composed of al-ternate horizontal opaque and transparent strips (eachof width d(2) and are projected onto a silicon photocell.Each particle passing through the beam of light willproduce an alternating response on the photocell ofam plitude proportional to the projected surface area andfrequency given by (V 0+ Vt)!d where Vt is the term inalvelocity of the particle in water. W ith N randomlydistributed particles of a given size range in the light

JO UR NA L O F T HE SOU TH A FR IC AN IN STIT UTE O F M IN IN G A ND META LLU R,G Y


4/11

beam at any time the amplitude of the signal at thephotocell is proportional to Ni. In the prototype in-strument (com posed of four frequency channels) signalsappropriate to a given frequency range are amplifiedand then processed in electronic squaring and integratingcircuits. The output from the analyser therefore con-sists of time-integrated readings from four channels,each of which is a measure of the number of particlesw ithin a given size range... 90'in>-tic:"~ 80..';;j

Cl>.. 60

G>


5/11

lOO90

'" 60N.in 70c"= 60enen 50~~:E 40'"ij; 30'" 20

100 12,5 25

w POver fl ow t ube -L

A"Baffle.

-, /'

Stirrer\ Ultrasonic head!i!

B eam exp ander

Slurry ~, F low cell/S patial filter , photodetector

/Dig it al vo ltme te r//,

Gas laser/Lens /Calculator

Rotating screen//Pump

Fig. 4-Size analysis by diffraction of light

.- Motor

Screen

Balance

P article flow-- Laser'" S.,eving Propell "r - -- /-

50 10 0 20 0 40 0 80 0P article size (pm)

Fig. 5-R esults of diffraction m ethod Fig. 6-A wet sieving device262 MARCH 1973 JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY


6/11

10 0

90

80

70

Cl 60c>-~'" - 50;E::0.1: ! 40Cl_SijI". 30;t-1:Cl-" 20

~10

70c:Cl >~UV1 60E:tLJ')0- 500'1c:U1U1"a. 40;!!-

3015 0 2

It will now be demonstrated that sieving need not beso time consum ing as it is commonly thought to be.Screen analyses can be accomplished in a few m inutesw ithout any need to dry samples either before or aftersievinglo. The method to be described has as its basisth e a ss es sm en t of the weight of the materia l on thescreen by Archimedes' principle. The method may bedemonstrated w ith reference to F ig. 6. Vessel A is f il le dwith water via pipe W to a known level L determ ined bythe overflow tube I, the valves V an d X being shut.The vessel is then weighed (let th is weight be W w)' Th elevel of water is then lowered to the vic in ity of thescreen S co ntain in g a pe rtures of known size, by openingV and/or X. The valves are then closed and the slurryto be sized is admitted via pipe P to the level L. Th evessel is again weighed (let this weight be W B)' Th evalve V is then opened and water is adm itted via pipeW to wash all partic les finer than the screen size throughthe screen. A propeller driven by a small D .G. motorfacilitates a rapid sieving action without excessiveblinding and pegging. In the absence of mechanicalagitation the screen blinds severely and sieving cannotbe effected by supplying water alone. The use of ultra-sonics for keeping the screen clear of p egg ed m ateria lcould also be used to reduce the time taken for screening.Once all partic les which can do so have passed thescreen, as show n by the em ergence of clear w ater throughthe valve V, the valve V is closed and water is addedvia pipe W up to the level L. The vessel is weighed athird time (let th is weight be W A ). T hen , by co nve ntio na lphysical principles, the weight of so lids a dde d in itia llyto the vessel is (WB- Ww )/(l-pw /p s) where pw an d psare the specific gravities of water and solids to be sized,respectively. S im ilarly the weight of s olid s remain in gafter the fine materia l has been washed out is given by(WA-Ww)/( l-pw/Ps). Thus the fraction by weight ofsolids coarser than the screen is given by (W A - W w)/

10 0 10 0

90 90c:Cl >Cl >~UU1 80

c:Cl >Cl >U 80'"E:t~ 70

E:t;::! 70

0'1c:'"'""a.60~

0'1c:U1'"" .. 60 . ,~

50 J30 050 0

0 15 20 70 800 40 50 60W eight ", passing 74 I'm ( we t s ie vin g fo r 5 m in s)

F ig . 7 -Resu lt s of dry and wet sieving(W B - Ww) . On comple tion of the sieving and w eighingoperation the solids coarser than the screen may beflushed out of the vessel A through the valve X.

5

80

I10 4 600Sieving time (m ins.) S ieving time (m ins.)ieving time ( mins.)

Fig. 8-Kinetics of w et sie vin gJO UR NAL O F THE SO UTH AFR IC AN IN STITUTE O F M IN IN G AND M ETALLUR GY M AR CH 1973 263


7/11

C' Ic:>J'V ic: 1/ 2I: jJ"'0 1/0JE.c . qa....>- 0/6.0"'0J 0/.IVIJVI q2I.I:j~e n 0.-J3 -02J J>..... -0 4:j I::JE -06:I IUc: -0 8.... I0........ -10J J~0

-1 2

The method was tested under laboratory conditionsusing 4 4-lL m, 7 4-lL m an d 105-lLm screens. Samples ofbetween 20 and 25 grams of m illed W itwatersrand goldore were sieved dry by machine until the sieving rate ofsolids through the screen reached 0,1 per cent of thetotal sample weight per minute. This was taken to bethe end-point for sieving. The samples were then madeinto slurries containing about 15 per cent solids byweight and sieved for fixed times in the wet sievingdevice. Results of dry and wet sieving on 74-lLm screensare shown in Fig. 7. The kinetics of the sieving action inthe wet sieving device are described in Fig. 8 whichshows the percentage of material passing the screen as afunction of time. Reproducibility of results was ac-ceptable and errors amounted to less than 1 per centpassing given mesh size.

It has been pointed out that ore bodies can be com-posed of a mixture of minerals of different densities. Ifthe average density of particles corresponding to adiscrete size fraction is a function of particle size in agiven sample, one would expect a method of size analysisbased on wet sieving and Archimedes' principle to yielddifferent results from the standard method of dry

26 4 MARCH 1 973

sieving and weighing. Fortunately this variation inaverage density with size for finely m illed gold-bearingrock is generally small and rarely exceeds two or threeper cent for particle sizes from a few tens of m icrons to afew hundred microns. It is instructive to give some in-dication of the sensitivity of the difference in resultsbetween dry sieving and the method of sieving des-cribed above to variations of average density withp article size.

Consider a sample which can be represented by thecumu la tiv e v olume-size d istrib utio n V(x)= f (x) and letthe average specific gravity of the particles be a functionps(x) of particle size x. For dry sieving the cumulativeweight fraction of particles less than size x is :

x 00W n(x)= [Jps(x) f '(x)d x]/[J p s(x) f '(x )d x]

0 0where f '(x)= :x[ f (x)]

Sim ilarly for the wet-sieving method the cumulativeweight fraction is

ps (X ) = opal X + 2/7

10 20 40 500Particle site ( pm )

60 70

ps (X ) = - JOOI X + 2J7

Fig. 9-Errors in weighing solids by Archimedes' principleJO UR NAL O F TH E SO UTH A FR IC AN IN STITU TE O F M IN IN G A ND META LLU RG Y


8/11


9/11

12 40 J:0'1QIJ:~e n >aIE Percent solids 30 aIf ) 9 en


10/11

The sensitivity of the instrument to variations inparticle size looks remarkably good, as shown in Fig. 11which represents the output from the system. Accuracyof results compared with sieve analyses is also verygood, as Fig. 12 indicates. At Magma Copper M ine Co. inArizona, U.S.A., 286 sam ples were analysed over a three-month period and results yielded a mean square error of:t 0,64 per cent+65 mesh.The use of the analyser for control of milling circuitshas resulted in a uniform grind (:t 0,5 per cent on thecontrol screen fraction) and improved throughputs ofbetween 4 and 12 per cent greater than similar circuitsoperating without the use of an on-line sizer. It has beencalculated13 approximately that with a 5 per cent in-crease in throughput and a uniform grind (obtained bysize control) the Inspiration Consolidated Copper Co.,U .S.A. could pay for a control system in less than a year(cost of "PS M System -lOO" about 30000 U.S. dollars).

CONCLUSIONSThis review has endeavoured to show that the questfor a successful real-time size analyser has been fraught

with difficulties. The ideal instrument which can de-term ine the complete size distribution of particulatematter in process streams has yet to be developed.However, a thorough paper study of the physicalprinciples involved in the design of a real-time analyserwill often indicate whether the principles selected arelikely to lead to a potentially useful design. Neglect of thisstep has in the past led to some abortive attempts tod ev el op ana ly se rs .The best instruments available currently are capableof determining accurately only one point on the cumu-

lative size distribution curve. Although knowledge of asingle point may be sufficient for significant improve-ments in the control of grinding circuits, it may not beadequate for a full understanding of the complex natureof large-scale extraction processes and the behaviour ofprocessing equipm ent handling large tonnages.Unfortunately the industry is now faced with the

dilemma that although on-line size analysers are avail-able, the benefits (other than benefits from control ofm illing circuits) which can be derived from their usehave still to be determined fully. Since commercial on-line sizers involve considerable capital outlay there hasbeen some reluctance by the industry to purchase theseinstruments until their usefulness can be evaluated.Since their usefulness can only be proved from intensivein-plant tests a stalemate has developed. It is hoped thatthe dilemma will be resolved in the near future and thatthe potential of real-time size analysers in the miningand m etallurgical industry will begin to be realised.

ACKNOWLEDGEMENTSThe author would like to express his thanks to Dr

P. J. Lloyd and Dr D. A. White for friendly and stimu-lating discussions and to the Chamber of Mines of SouthAfrica for permission to publish this paper.

APPENDIXS om e d efin itio ns o f te rm in olo gyR ea l-tim e size a na ly se r: size analyser characterised by aprocess time for yielding size distributions which is

sufficiently rapid to allow control of process stream s.On.line s i z e analyser: a real-tim e size analyser in w hichthe size sensor is located in the process stream or in aparallel sampling stream and by means of which con-tinuous m easurements are m ade of the size distribution.Off-lin e siz e a na ly se r: a real-time size analyser in whichsize assessment is done in a batchwise manner by ex-trac tin g rep re sen ta tive sample s interm itte ntly .

REFERENCES1. JACOBS. D . J. "An investigation of the possibility of de-term ining the size distributions of particles in the sizerange 20 I-'m to 0,5 cm in suspension by diffraction or ex-tinction of light". M .Sc. thesis, Univ. of S. A frica, PhysicsD ept., D ecem ber, 1968.2. D AV IS, R . "Particle size analysis. Literature on the sam pling,collection, and analysis of particulates is reviewed". Ind.E ng ng . O hem., 6 2, 8 7-9 3 ( 19 70 ).3. SWEENY, K. H. and GECKLER, R . D. "The rheology of sus-pensions". J. Appl. Phys., 2 5, 11 35-44 (1 95 4).4. BROW N, G . E . "M ethod of control of particle size utilisingviscosity". Canadian patent No. 794, 330. Filed 28th June,1965.5. OLIVIER, J. P., HICKIN, G. K. and ORB, C. Jr. "Rapid,

automatic particle size analysis in the subsieve range".Powder Technol ., 4 , 2 57 -6 3 ( 19 70 /7 1) .6. N AKAJIMA, Y ., G OTOH , K. an d T ANAKA, T. "O n-lin e p articles ize analyser". I nd . E ng ng . O h em ., Fu nd am ., 6 ,5 87 -9 2 ( 19 67 ).7. RICCI, R . J. and COOPER, H. R. "A method for monitoringparticle size distribution in process slurries". Trans. Inst.S tan d. A m., 9 , 2 8-3 6 ( 19 70 ).8. SC OTT , K . C .S.I.R ., Pretoria, Private com munication.9. C ORNIL LAUL T, J. "P article size a naly ser". Appl. O pt., 11 ,2 65 -2 68 ( 19 72 ).10 . C HAMBERO F M IN ES S ERVIC ES (P TY .) L TD . "W et S iev in g".Prov. S.A . Pat. 72/0191, (1972). Dr A. L . Hinde ceded toCham ber of M ines Services (Pty.) L td.11. HOLLAND-BATT, A . B. "Further development of the RoyalSchool of M ines on-stream particle size analyser". Trans.Inst. M in. Engrs., 7 7,0 18 5- 19 0 (1 96 8) .12. HATHAW AY, R. E . "A proven on-stream particle size m oni-toring system for automatic grinding circuit control".Paper presented at the 1972 M ining and Metallurgy In-dustries Group Symposium and Exhibit of the InstrumentSociety of America in Phoenix, A rizona, U .S.A .; April 24and 25, 1972.13. DIAZ, L . and M USGROVE,P. M . Jr. "On-line size analysis ingrinding circuit control". Paper presented at the annualmeeting of the Society of M ining Engineers of AIME inSeattle, September 22 to 24, 1971.

DISCUSSIONJ. H . TALBOTThe diffraction technique of particle size analysis had

its origin in South Africa, and it is here that it has beenmost actively developed and has reached its higheststandard of perfection. The first instrument to be avail-able commercially was the Talbot Disa (acronym fordiffraction size analyzer) manufactured by the CornerHouse Laboratories under an agreement with theChamber of M ines of South Africa.In the diffraction size analyzer the size distribution is

obtained by Fourier inversion and differentiation of thepower spectrum of the object. The operations are alldone optically and the size distribution is obtaineddirectly.About two and a half years ago, the development of

the diffraction method of size analysis took a newdirection in South Africa with the invention of thespatial period spectrometerl. To understand this newdevelopment, it must be appreciated that the con-ventional diffractometer yields a Fourier analysis of

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY MARCH 1973 267


11/11

the object in the form of a spatial frequency spectrum .In the spatial period spectrometer, Fourier analysis isdone in such a way as to yield a spatial period spectrum.The advantage is that spatial period has the dimensionsof length and the spatial period spectrum is, therefore,a true size spectrum. It avoids the problem of defining au nit e ntity .

Spatial period is a statistical parameter and thereforethe spatial period spectrum may be used, like theMaxwell-Boltzmann distribution of statistical mech-anics, to relate the macroscopic properties of a system toits m ic ro sc op ic s tru ctu re 2.

Two models of the spatial period spectrometer arealready available. These are the PM3 and PM31, manu-factured in South Africa by Talbot Research. The PM31is an attachment for use w ith a PM1 series ImageAnalyzer.

A real-time, on-line, size spectrometer, designatedM odel PM 303, has been under development for more thantwo years and is expected to be available soon. Thebasic instrument consists of a sensing unit and a displayunit. The sensing unit contains a diffractometer of thespatial period type with a flow cell of rectangular cross-section 400 mm X 5 mm. This flow cell allows 5 litres ofsuspension to be processed in one second. A typical millpulp sample containing 100 g of solids can be processed

in 30 seconds. To achieve m ore rapid processing, num bersof sensing units may be fastened together and theiroutputs connected in parallel to a single display unit.The sensing unit is 1 m long, 420 mm wide and 100 mmhigh.

Instruments can be provided for monitoring any of anumber of normalised moments such as specific surface,mean linear dimension, mean volume etc. or for recordingthe spatial period spectrum in any number of size inter-vals.Owing to the translational invariance of the spatialperiod spectrum, aggregation is not as serious a problemas with most other methods.

A novel feature of the PM303 is the use of variancedetection, our proprietary system of signal-to-noiseratio improvement. W ith this system the variance of theparameter is measured as an estimate of the parameteritself. This m ethod elim inates low -frequency noise suchas ambient diffracted light due to imperfect opticalcom ponents, the Schlieren effect and particles depositingon lenses and cell w indows, which would otherw iserequire frequent re-setting of the zero.

REFERENCES1. TALBoT, J. H . South African Patent No. 70/3652.2. SCARLETT, B. Particle Size Analysis 1970, 101-112 (TheSociety for A nalytical C hem istry: London 1972).

ORANGE FREE STATE BRANCHMINUTES OF THE COMMITTEE MEETING HELD IN THE WELKOMCLUB ON WEDNESDAY, 8th NOVEMBER, 1972

P r e s e n t :C. J. I s a a c (in the Chair), E. T .W ilson, J. M . M eyer, C . M ostert,P. L. N athan, D . A . Sm ith, R . Suther-land.A p o l o g i e s :G . You ng .M INUT ES O F THE PREV IOUSCOMMITTEE MEETINGThe minutes of the CommitteeM eeting held on the 19th January,1972 were taken as read and thei radoption was proposed by Mr J. M .Meyer and seconded by M r E. T.Wilson.There were no matters arisingfrom th es e m in utes .

GENERAL MEETINGS FOR THEENSUING YEARIt was agreed that the format ofmeetings used for the past year becontinued for this year.

268 MARCH 1973

The follow ing m eetings w ere de-c id ed on:Tuesday 30th January, 1973 - Afilm s how "N ic ke l M in in g in Can ad a"and "N orth S lope A laska",Tuesday 6th March, 1973 - Anaddress by Anglo American Re-search M etallurgists on latest G oldEx tr ac tion me thods.End August, 1973- Annual Gen-eral M eeting to be addressed by theP re sid en t o f th e In stitu te .PROPOSED VISITS FOR THE EN-SU ING YEAR

It w as decided that a visit be m adeto V ierfontein Colliery and Pow erStation on or about the 11th May,1973.It w as further decided that a localvisit be held in early August, 1973to th e fo llowin g:P re sid en t S te yn - 4 Shaft - Sur-fa ce , S haft an d Und erg ro un d S ta tio n

layouts.Virginia - Task forceand Me th an e F ire d Bo ile rs . trainingNEXT COMMITTEE MEET INGIt was decided that the next Com -m ittee M eeting should be held w henCommittee Members felt it neces-sary.

GENERALI. The Chairman read a letter fromProfessor Howat in which hethanked the O .F .S. B ranch for hisp le as an t v is it to We lk om .2. It was suggested that the G.M .E.be approached to address a m eet-ing of the O.F.S. Branch - M rD . A . Smith to investigate.3. It was suggested that the O.F.S.B ra nch h old itse lf re sp on sib le fo rthe submission of at least 2

m ining pap ers durin g the year.T he C hairm an declared the m eet-ing closed at 5.45 p .m .JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

a review of real-time particle size analysers

Documents