ISBN : 81-87053-53-4Processing of Fines (2)

Eds.: P. Bhcatacharv}a. R. Singh and N. G. GoswcuniO NML Jcurn.chedpur-831 007, India, 2000. ply. 373-387

Computer aided design and optimization ofmineral processing plants by a state of the

art simulatorPRADIP, S. RAHA and P. C. KAPUR

Tata Research Development and Design Centre, Pane 411013, India

ABSTRACT

Tata Research Development and Design Centre (TRDDC) has developeda state of the art mineral processing simulator called SimL8. Jr performsmodelling, simulation and optimisation functions and provides viablestrategies for enhancement of the performance of mineral processingplants. A number of case studies on plant diagnostics, grinding,classification, flotation and pressure filtration are taken up todemonstrate the utility of modelling and simulation on SimL8 platform.Keywords : Modeling, Simulation, Plant audit, Diagnostics, Data reconciliation,SimL8. Grinding, Classification, Flotation, Circuit design, Filtration.

INTRODUCTION

The efficiency and viability of a mineral processing plant operation is deter-mined by its productivity, yield, product quality, cost, energy consumption, en-vironmental impact and so on. These i ndices, in turn , are impacted by the chem-istry of the process, as controlled by chemicals, reagents, surfactants, polymers,pH etc., and by the physical unit operations employed. The advances made inmolecular structure, properties and behaviour now permit us to select chemicalswith much greater reliance on our understanding of the fundamentals than onempirical knowledge. Molecular modeling and reagent design is an emergingfield, which holds considerable promise for mineral processing industry. Recentadvances in reagent design'' "I have shown that it is possible to identify moreselective reagents, tailor - made for a specific separation problem at hand. De-tailed discussion on these topics. however, is outside the purview of this paper.The reader is referred to several excellent monograph and reviews.

373

I^r. Hl IIY ii u, ttiII iI4 IiY IId^IM^n^I1lrlaktilY11W^INN lYlrOirlrm^lui Wl Ville ^^ I L^^ ^I^^^OIYY^IM^,^ w

PRADIP, .5'. RAHA wt( ( P. C. KAPCR

Improvement in the process chemistry, in isolation, is by no means adequatefor optimisation of the mineral processing plant. It is concurrently necessary toanalyse in detail the various unit operations that comprise the plant as well as themanner in which these are interconnected, that is, the circuit configuration. Forthis purpose, one has to carry out computer aided process design, simulation,optimisation and circuit synthesis. SirnL8, the state of the art mineral processingsimulator developed by TRDDC. perform these tasks with relative ease in a userfriendly environment.

As shown in Fig- I, the first step in an exercise directed at plant optimisationusually calls for collection of plant data, followed by data reconciliation, plantaudit and diagnostics. Next step entails acquiring or building mathematical modelsof the various unit operations comprising the plant. such as size reduction. clas-sification, beneficiation, liquid-solid separation etc. The models can he cmpiri-ca], semi-empirical or more realistic and detailed phenomenological models inthe particle population balance paradigm. The last approach requires a soundunderstanding of the actual physical phenomena prevailing in a unit operation.Once the mathematical models are validated with plant / laboratory data, itshould be possible to simulate the performance of individual units and the circuitas a whole, and come up with strategies for pertor-mance enhancement. Thesecomplex and computational intensive tasks are best carried out with help of atailor-made, dedicated and user-friendly modeling-simulation tool such as SimLB.

Plant audit & diagnostics

Design / selection ofappropriate reagentcombination (dosageoptimization)

Size Reduction& classification

Unit operationsOptimization &circuit design

Separationof desiredcomponent

Solid - Liquidseparation

I i,,. I Sclrenrcrric diogmiir sliosrif? , rirriotts steles- in I^lcurt njrrirni.z a tiorr

MODEL ING-SIMULATION TOOLS

Very few cornntercial modelling-simulation tools for mineral processing plantsare currently available. These are USIM PAC lrom BRCM-T. France: JKSimFloatfrom JKMRC, Australia: and SimL8 from TRDDC, Pone. Based on inc rc than

374

PRAI)fP S. RAI1 t and P C. KAPL)R

fifteen years of TRDDC experience in modelling, simulation and optimization ofmineral processinunit operations and circuits. SintL8 was developed in col-C7

with the software-engineering group of Tata Consultancy Services andHindustan Zinc Limited. It incorporates the latest Simulator Development Envi-ronment (SDF), a proprietary technology of Tata Consultancy Services, whichenables seamless additions of new unit Operations, process models, solvers oroptimisers. SiniLB has 3 modes of operation. namely, data reconciliation'parameter estimation and simulation. It has a user friendly graphical user inter-face, which interacts with the hackend solver, optimiser and a large hackendliterary of process models. The architecture ol' SiiuL8 is shown in Fig 2.

Data ReconciliationParameter Estimation

Simulation

Plant flow sheetEquipment

specificationst Process specific

variablesPlant data

Solvers Optimizers

GrindingClassificationFlotationMGSFiltration

I-ditahle reportGraph

Display valueRun info for output

CAST? STUDII?S

Fig. 2 : Archilecriuce (it .SirrrLti

Design andoptimization of

gnndmg circuits

Design andoptmization of

flotation circuits

Design of mineralprocessing

circuits

A Iew case studies carried out by us are presented in the following sectionsto illustrate the utility and power of SiniLK.

Plant Audit and Diagnosticsi15-171

A process audit is a methodical and systematical exploration of a system forrluantifying its performance. Process auditing of it mineral processing plant atregular intervals can provide valuable information to the plant engineers regard-ing the plant health and drift in the plant performance with time. It is alsopossible to identify the prohlci ultic process units Iron the plant audit. This casestud) k'~as done on a Pb circuit. shown in big. 3. of a Ph-Zn plant. The Ph circuit

37

^ ^r1Y^^w^r iiir^ ia m*-f ` 'W`iriY 1 11 YYl^1^^ -x11id"I I ^MYYI^ w^wud **Iwi0A

PR AMP 1'. k.lILt ^i id I'. (' K1I'l li

comprises of rom,'1ci , Rol. sea' emger ( Sc 1. cleaner I ((' I I ). cleaner' t(-I?) andcleaner) ( ('l ) lxiik,.

Based on reconciled data. the bar-chary in H'-tore 4 show (he variation in ilicPh grade and Ph reco\er^ acro,s \i:rious tuns.

-mounts of ,graphite carbon and insoluble across the runs are "howl illFig. 5.

The suriations in the performance of different process units across the datasets are shoNsn in I i_'. 6(a-d). In an ideal separation the distribution of the desiredcomponent IPhi i, unite and unde,irahlc, i(ir. C. Insoluble eeru.

t : Ph (irrnrl

60

4 40

20

sett Set 2 sets

Pb Grade

13 'b Rerovar)

Data Set

l'h iti ed anal re~^^rr1v h nett' r l01 1

376

PRADIP S. RAHA and P. C. KAPUR

15

10

5

0

Data Set

Fig. 5 : Impurities in Pb concentrate

a

zData Sol

Fig. 6(a): Rougher performance in terms of separationefficiency of Pb from Gr . C and insolubtes

Fig 6(c ) Cleaned performance intarms of separationefficiency of Pb from Gr C and insolubles

a 'Yetns

'rC

Dale SetFig. 6(b): Scavenger performance in terms of separation

efficiency of Pb from Gr. C and insolubles

3Data Set

Fig 6(d ) Cleaner3 performance in terms of separationefficiency of Pb from Gr C and insolubles

Figlra,

M^I^NY^Wi a^.. ^. ,a^..^^W^Yun^r^ ^+sr^r^...... ..Il .W N^u^^ee^. ^ib,^.. r^anM

/'R;11)//' S. RAHA anal P C. KA1'f'R

It will he seen from these plots that there are considerable variations in theperformance of the process units in different runs- On the whole. Rougher andScavenger are working reasonably efficiently as compared to the cleaner banksin rejectin* insolubles. For separating Ph. Cleaner] is not working ^4cll: how-ever, cleaner 3 has better separation performance for Ph , and in sci it i,performing quite well.

Information of this kind based on plant data but appr^priateI reconciled formass balances using a tool like SiniLS, if collected on a regular basis and docu-mented. again with the help of a data processing tool like^Siml.8 having thosefunctionalities, can provide valuable means of optimir.ing plant performance, ifthe trends of this kind established on the basis of plant data are correlated withactual mineralogical or operating conditions . one can make suitable chant*c, toenhance plant pcrforrnarrce.

{Grinding and Classification"'2111

Case study /

This illustration is for a typical chronic ore heneticiation plant. The originalchrome ore grinding section is shown in Fig*. 7. The problem in the plant wvasrelated to low throughput with relatively high generation of slime (-37 micron).SimL8 was employed to mass balance, model and simulate the grinding circuit.

It was found from the plant audit studies (consisting of sampling canipaig*nsBillowed by data reconciliation and performance assessment of each individualunit) that the sieve bend was overloaded and was not performing well. Based onextensive simulations. a circuit modification was proposed in which the finerfraction of screw classifier is not reclassified in the sieve bend but directly takenout as product. Moreover, the sieve bend overflow is sent hack to screw classi-fier, unlike the original circuit in which the sieve bend overflow is recycled tothe ball mill. These changes led to a significant increase in [he overall circuitthroughput and reduction in proportion of slimes fraction. The tziodified circuitis shown in Fig. 8.

As the separation efficiency of sieve bend was poor. the throu-hput of themodified circuit could be improved by bypassing the screw classifier finer frac-tion. The sieve bend perforniace before and after the Modification is shown inFig. 9. The sieve bend \,%a,; overloaded and reduction in the sieve bend loadingcan improve its separation efficiency.

37N

PRADIP S. RAHA and P. C. KAPUR

t0

0a

04 -

Fig. 7 : Chrome ore grinding circuit

Fig. 8 : Modified chrome ore grinding circuit

-+- wftrnp tw a CM)-t- orasl etrsncy {r 4 I Mt)

0,0 :^ * r .100 700 900 400

Purtic)N S. (micron)600

Fig, 9 : Sieve bendeflicienc v

379

mioidd wo I,. I i 1iai144 mkok rk"114d it i n . Y 1i IhFtr , rYiq^IiriYM1^M!YW^M^ Jim Ft^ frI 1+64^r^IM is* igloo I J i N. i ilr

PRAD!P S. RAHA orrc! P C. KAP(,R

A comparison of the performance of the two circuits is shown in Table 1.

J ibfe / Comparison of per/or-ntance of tiro circuits

Circuit Feed (t/hr) Feed (t/hr) Ball mill Product. Product(-3 mm) (-15+3 mm) throughput wt`4 (-37 w1c4

(t/hr) micron) (-500+37micron)

Normal 81.46 25.70 69.70 34.88 62.95Modified ckt 151.39 47.76 112.18 34.13 65.45

A tool like SimL8 can thus he very gainfully utilised in plant practice in notonly assessing the unit performance quantitatively but also to find suitable rem-edies to those operating sub-optimally.

Case stctdV 2

The primary and secondary hydrocyclones in a classifier circuit can he com-bined in various ways as shown in Fig. 10.

Fig. 10 . Clussifiers crrrrrrr, er! in r

PRA1)1P S. RAHA and P. C. KAPUR

Cumuiativ Size Distribution in Final Product

e a lB p ro.m

Partlcl^ skza Emicrer. )

Fig. 11 : Product size disiiibrittori in various hvdrocvclone urrangemenis

Flotation circuit[2 3

The flotation circuit will be described with a case study of the Ph circuitalready shown in Fig. 3.

Case study 1 :

Some interesting simulation results were obtained when the final Ph concen-trate stream is fed back to various banks i.e., rougher, scavenger, cleaner], cleaner2and cleaner 3. Fig. 12 illustrates the effect of recycling path (the results ofincreasing proportion of recycling of final concentrate back to the flotation cir-cuit) on the rejection of graphite carbon content and Ph recovery. It would seemthat recycling of final concentrate to scavenger has the most disastrous effect onlead recovery with only a small improvement in graphite carbon rejection. Thisis expected since material fed to scavenger bank will go out of final tails, leadingto high loss of valuable material. When recycled to rougher hank, even thoughthe recovery drops, it is accompanied with a better performance in graphiterejection. Recycling in cleaner 1, cleaner 2 and cleaner 3 shows improved per-formance with better rejection of graphite carbon and lower drop in recovery inthat order. The simulation results suggest that cleaner 3 concentrate can he re-cycled back for improving the quality of concentrate.

Case study 2 :Based on empirical experience, plant engineers often shift cells from one

hank to another. The effect of shifting some cells from rougher to scavenger hankwas simulated with SimL8. Fig. 13 shows the effect ail' increasing the number ofrougher cells, while keeping total number of rougher and scavenger cells con-

381

I ^IrYrN II' 11 Mw+.^ IAYa Ir y1 bid "^ ^YMf^a Y^II ! 1 r^^la1W+ Yy^ Y11r'

PRA MP S. RAHA curd 1'. U KAPUR

stint, on Pb grade and recovery. With an increase in numher of' rougher ccl]s_ Phgrade drops. Ph recovery increases slightly as number ol, rougher cell is in-creased from 2 to 3 and then it begins to drop again with further increase innumber of rougher cells.

% graphite in final concentrate

f-i,g. 12 : f-jfec: of ,reveling of filial cr',l tnerate at rurir , rrs /nicks

0

Number of rougher calls

1--i". /3 ( !^JJ.S /ruin uc'crirRI Leer to curt Ju r

0-th level s imrrfctiion :

7

SiuiL;i has malty detailed and realistic nrathenuatical models for erindirrc!.classification and Ilotation. Folr other unit operations. "'hick have not hccnmodelled yet. a facility of O-ill level siltrulutiolli ha, been prrv idCd. In t)-ill ICv Cl

3x^

PRAI) IP S. RAfiA turd P C. KAPORsimulation, it is assumed that separation efficiency of a process unit, which canhe calculated in SimLS, remains approximately invariant of any reasonablemodification in the circuit configuration. Thus, one set of separation efficienciescan he employed to simulate the modified circuits. This is demonstrated withreference to the . Pb circuit in Fig. 3. It is known that a multi gravity separator(MGS) can he quite effective in rejecting graphite carbon from the Pb circuit. Adecision had to he taken regarding the position of MGS in the Ph circuit. Withthis objective in mind. several circuit configurations were tested by simulationin order to obtain the best possible circuit configuration. Fig. 14 shows theoptimal circuit configuration -obtained. There is a significant improvement in Pbgrade with only marginal drop in recovery when MGS feed comes from cleanerI concentrate and cleaner 2 tails. Table 2. shows the expected performance ofdifferent circuit modification.

Fig. 14 : Pb circuit with MGS

Table 2 : Effect of MGS on Performance of Pb CircuitCircuit Configuration Grade (%) Recovery (%)

Pb Gr.C PbWithout MGS 48.12 9.58 62,26

Feed Conc. TailScavenger Cleaner 1 Rougher 46.17 8.74 66.67

ConcCleaner 1 Conc Cleaner 2 Cleaner 1 71.87 0.86 50.88

With Rougher Conc Cleaner 1 Rougher 71.23 0.81 46.29MGS Cleaner 1 Conc Cleaner 2 Cleaner 1 68.41 0 .75 57.19

and Cleaner 2tail

Cleaner 1 Conc Cleaner 2 Rougher 73.15 0.67 41.66

383

t.^^.f .t Yrtrf^rlMM ^i r it b iH lO ^rv^^1^^MIY i x xwi ngMdl. , ^^1J L 1iL1 trl t^/ 1 ^rtYi1W rMY^n1 r ^Yl rlil^ i

PR;1i)1/' S. I?.II/,A uml I' (. &.1 P1 11 ,

Solid Liquid ticparationtu..'IAll nnncral pro:essing plant, have sold liquid separation units. like thi'k-

ener. pressure filtration etc. The various solid liquid separation devices arc listedin Table 3.

Table 3 : Deu ate rin,i,' e!('t,ieesPres>urc ikPat Deice

O Gravity belt filter< 1(1 Thii ^cnrr

PRADIP S. RANA and P. C. KAPUR

Fig. 16 : Effect of initial +lrrri \ height on overall throughput

ACKNOWLEDGEMENT

Development of SimL8 has involved hundreds of man-months effort andcontribution of many of our colleagues in TRDDC, TCS and HZL. We aregreatful to all of them for their inputs. We also greatfully acknowledge the helpand support of the management of TCS and HZL.

REFERENCES

1. Ravishankar, S.A., Pradip , Khosla, N.K., (1995), " Selective flocculation of ironoxide from its synthetic mixtures with clays: A comparison of polyacrylic acid andstarch polymers ", Int. J. Mineral Processing , 43, p. 235.

2. Pradip , Das, K . K., Singh , R., (1995 ), "Flotation of complex sulphide ores, In:selected topics in mineral processing ", (Eds.) Pradip and Kumar , R., Wiley Eastern,New Delhi, p. 118.

3. Pradip , Kumar , R., (Eds. ), ( 1995), "Selected topics in mineral processing ", New AgePub., New Delhi , p. 300.

4. Das, K. K., Pradip , Suresh , B., (1995), "Role of molecular architecture and chainlength in the flotation separation of oxidised copper - lead-zinc minerals usingsalicylaldoxime derivatives ", Proceedings , XIX International Mineral Processing Con-gress, San Francisco, SME-AIME , USA, vol . 3, p. 245.

5. Mulukulta, P.S, (1994), (Ed.,) ""Reagents for better metallurgy", SME-AIME, Colo-rado, p. 364.

6. Pradip, (1994), "Benefieiation of alumina-rich [ndian iron ore slimes", Metals, Ma-terials & Processes, 6(3), p. 170.

7. Pradip, (1994), "Reagents design and molecular recognition at mineral surfaces".Proceedings, Symposium on Reagents for Better Metallurgy, (Ed.) Mulukutla, P.S.,SME-AIME, Chap. 24, p. 245.

385

wbwedi"rore' xM r^ wrYr ,^^.. , wrrlYNam4 l r^ YW^.r $ ^$air4I rM Mrrt^.Y . ^a

PRA DIP S. RAHA until P C. KAPI 'R

8. Pradip. ( 1992), "Design of crystal -structure specific surfactants based on molecularrecoeznition at mineral surfaces", Current Science, 63(4), p. 180-

9. Pradip. Fuerstenau. D.W., ( 1991 ), "The role of inorganic and organic reagents in theflotation separation of rare-earth ores", Intl. J. Mineral Processing. 32. p. I.

10. Pradip, Singh, R., Das, K. K., Sankar. T. A. P., Sunilkumar, T. S., Narsimhan, D..Rao, N. K., (1991), "Studies to improve the flotation recovery in the heneficiationof a complex mixed sulphide-oxide ore from Malanjkhand. India", ProceedingsXVII Intl. Mineral Processing Congress, Dresden. Germany IV, p. 151.

1 I . Pradip, ( 1988). "Applications of chelating agents in mineral processing", Mineralsand Metallurgical Processing, 5(2), p. 80.

12. Pradip, ( 1988), "Testing and evaluation of reagents for mineral flotation". Mineralsand Metallurgical Processing, 5(3), p. 114.

13. Sornasundaran, E. Moudgil, B.M.. (Eds..) ( 1988). "Rcazents in Mineral Technol-ogy", Marcel Dekker. p. 755.

14. Jones, M.J., Oblast, R., (Eds..) (1984), ''Reagents in mineral industry', 1MM. Lon-don, p. 294.

15. Sabne, M. P. et. al.. (1995), "Improvements in mineral processing plant performancethrough computer-aided plant auditing and diagnostics", In : Mineral ProcessingRecent Advances and Future Trends, Ed.: S.P.Mehrotra and R.Shekhar, Allird Pub-lishers, New Delhi.

16. Kapur, P.C., Monika Agrawal and Fuerstenau, D.W., ( 1993), "A two-tier approachto the mass balance of multi-level mineral processing data", Part I: Performanceindex and local balance around a node. Int. Journal of Min. Process., 39. p. 209.

17. Kapur, PC.,. Monika Agrawal and Fuerstenau.D.W., (1993), "A two-tier approachto the mass balance of multi-level Mineral processing data". Part 2: Splicing ofstream flows and global balance over a circuit, Int. Journal of Min. Process.. 39. p.225.

18. Sivakumar S. et. at.. (1998). "Modelling and simulation of chrome ore grindingcircuit at Sukinda", Tata Search. p. 77.

19. Austin, L.G., Klimpel, R.R., and Luckic, P.T., (1984), "Process engineering of sizereduction : Ball milling". Society of Mining Engineers.

20. Prasher, C.L., (Ed.), (1987), "Crushing and Grinding Process Handbook, John Wiley& Sons., p. 259.

21. Kapur. P.C., "Modelling of tumbling mill hatch processes". Chapter 9 in 1231.22. Lynch, A.J., (1977), "Mineral crushing and grinding circuits: Their simulation, op-

timization and control", Elsevier Scientific Publication Co., p. 87-

23- Fintoft, B. C., Plitt, L. R. and Turak, A. A.. 1987, "Cyclone modelling - A reviewof present technology"', CIM Bulletin. Vol 80, p. 39.

24. Klimpel, R.R., (1980), "Estimation of weight ratios given component make-up anal" sisof streams", Trans SMEfAIME, Vol 266, p. 1882.

386

PRADIP S. RAHA and P C. KAPUR

25. Kapur. P.C., (1970), "Analysis of the bond grindability test". Trans IMM, Vol 79,p C 103

26. Kapur. P.C.. (1982), "An improved method for estimating the feed-size brekakagedistribution functions", Powder Tech, Vol 33, p_ 269-

27. Nageswararao. K., (1995), "A generalised model for hydrocyclone classifiers". AusIMM, 300(2), 21 (Dec).

28. Napier-Munn, T. J.. (Ed.), Mineral Comminution Circuits-Their operation and opti-mization, JKMRC, p. 273.

29. Pradip, Sabne, M. P., and Sivakumar, S., (1996), Computer-aided optimal design ofmineral separation circuits, Transport Phenomena in Metallurgical Processes, NML,Jamshedpur, p. 192.

30. Sabne, M.P., Pradip, Vora, S.B., and Kapur. P.C., (1995), Modeling, Simulation and41Optimization of a complex multi-stage multi-component flotation separation circuit,in Mineral Processing Recent Advances and Future Trends, Ed.: S.P.Mehrotra andR,Shekhar, Allied Publishers, New Delhi,.

31. Pradip, Rao, R. and Sivakumar, S., (1995), "Role of computer aided modelling andplant optimization in mineral processing", Quantitative Approaches in Process Met-allurgy, Eds: A.K.Tripathi, PDutta and H.S.Ray, Allied Publishers, New Delhi.

32. Dey A, Kapur, P.C., Mehrotra S.P., (1989), "A search strategy for optimization offlotation circuits", Int. J. of Min. Process., 26, p. 73.

33. Kapur, P.C., Dey A, and Mehrotra, S.P., (1989), "Identification of feed and simula-tion of industrial flotation circuits", Int. J. of Min. Process, 26, p. 73.

34. Landman, K.A. and L.R. White, (1997), "Predicting filtration time and maximizingthroughput in a pressure filter", AIChE J., 43, 3147.

35. Landman, K.A., L.R. White and M. Eberl, (1995), Pressure filtration of flocculatedsuspensions", AIChE J., 41,1687.

36. Buscall, R. and L.R. white, (1987), "Consolidation of concentrated suspensions I:The theory of sedimentation". J.Chem. soc. Faraday Trans., 83, 873.

397

page 1page 2page 3page 4page 5page 6page 7page 8page 9page 10page 11page 12page 13page 14page 15