AnalysisoAgitiation Leaching2data - Methods and Interpretation (3)

Analysis of Agitation Leaching Data Methods andInterpretationG Miller1

ABSTRACTAgitation leaching data has generally not been analysed to extract themaximum amount of information that is available. Analysis of heapleaching systems has been developing over the last five years to includethe underlying rate determining process. This has been found to bediffusion control for all copper leaching; irrespective of whether they areoxide or sulfide minerals. Extension of the diffusion model has enabledbetter scale-up protocols to be developed.

Acid consumption in copper heap leaching operations has been foundto be a first-order chemical controlled system. The analysis of the data isrelatively straightforward. However the interpretation and scale-up is notas direct as for leaching. The differences in the residence timedistributions (RTDs) are required to properly account for the scale-upbetween test columns and field.

The leaching of oxide copper in agitated systems has also been foundto be diffusion controlled. However, there are significant differences inthe control mechanisms. Initially the leaching control is the supply ofacid to the mineral surface. This is limited by the diffusion through thesurface boundary layer and is thus subject to the level of agitation. Higheragitation provides thinner boundary layers and faster diffusion rates.Similarly, higher grade material creates a thicker concentration boundarylayer and relatively slower leaching.

Acid consumption in agitated leach systems is also a first-ordersystem. However, the interpretation of the RTD is easier from batchtesting to continuous multiple CSTR series plant operations. Altering themethod of acid delivery can significantly reduce acid consumption rates.Predilution of the acid minimises the local acid concentration near theaddition point and can reduce the consumption by up to 50 per cent.

Modelling of the combined leach rate(s) and acid consumption rates inagitated systems allows comparison of various leaching strategies andselection of the most appropriate. Similarly, the model allows theidentification of economic limits to leach recovery; where the cost of acidconsumption is greater than the value of the copper recovered. From thisanalysis selection of residence times for agitation leaching can bedetermined. The modelling tools are useful in identifying optimumconditions for the leaching process.

INTRODUCTIONAgitation leaching data has not generally been analysed toextract the maximum amount of information that is available.Analysis techniques for heap leaching systems have beendeveloping over the last five years to identify and include theunderlying rate determining process (Miller, 1998, 2002, 2003and Dixon, 2003). This has been found to be diffusion control forall copper leaching; irrespective of whether oxide or sulfideminerals are being leached. Extension of the diffusion model toagitation leach systems should enable better scale-up protocols tobe developed.

Acid consumption in copper heap leaching operations hasbeen found to be a first-order chemical controlled system. Theanalysis of the data is relatively straightforward. However, theinterpretation and scale-up is not as direct as for the agitationleaching system. The differences in the residence timedistributions (RTDs) are required to properly account for thescale-up between test columns and field. In the agitation leachsystem the CSTR RTD is a well-known model that can be usedfor acid consumption prediction.

A recent Zambian leach project has undertaken preliminarytesting for agitation leaching of two ore samples. Thisinformation will form the basis of the process design andselection for the project. The ore is a typical copper belt oxideore with dominant malachite and significant mica and other platyparticles in the gangue. The Run of Mine (ROM) grade ispredicted to be 2.15 per cent ASCu (Acid Soluble Copper); butthe samples tested have proven to be of significantly higher grade at two to three times the anticipated ROM grade. As a result,interpretation of the leach testing is required in order to try andpredict the leaching characteristics of the ROM grade material.

LEACH RATE ANALYSISThe leach rate analysis is divided into two parts:1. leach rate with grind size, and2. leach rate with grade.

All the leach extractions were recalculated to reflect thecalculated head grade (based on solution and tails assays) as thisprovides the more accurate determination of the head grade. Theprocedure eliminates any anomalous +100 per cent recovery thatmay have been recorded in the primary laboratory data.

The leach recovery curves are shown in Figure 1. It wouldseem that there is no consistent pattern between the leach ratecurve and the grind size. The only (negative) information is thatthe high-grade coarse grind did not achieve a +90 per centrecovery in the six hours time used in the testing.

The leach rates shown in the simple recovery curves are not allthat amenable to analysis in this form. It has been postulated thatthe agitation leaching of copper is a diffusion controlled process;and that using this approach, further insights may be found. Theleach information has been transformed into the diffusioncontrolled, Shrinking Core model (Miller and Newton, 1999) tosee if this is an appropriate method. The recovery (R) istransformed (Tr(R)) by the Shrinking Core model:

Tr(R) = 1-2/3R-(1-R)2/3which yields a straight line when plotted against time if themodel is appropriate. The resultant leach curves are shown inFigure 2.

First Extractive Metallurgy Operators Conference Brisbane, QLD, 7 - 8 November 2005 1

1. Miller Metallurgical Services, 23 Stanfell Street, Corinda Qld 4075.Email: [email protected]

(% Cu, P55)

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Leach Time - min

3.6 -75 4.2 -150 6.4 -75 6.6 -150

Rec

%A

SC

u

FIG 1 - Leach recovery-time curves.

It is evident from this analysis that the leaching curves havebeen transformed into straight lines. As such the leaching isdiffusion controlled with a shrinking core regime. The slope ofthe line is the diffusion controlled leach rate. There are twoseparate lines in most of the leaching curves. This indicates thatthere has been a change in the underlying diffusion rate controlmechanism during the leach period.

The fast and slow leach rates have been normalised for thehead grade (a technique that arises from the shrinking core leachmodel) and are plotted as a function of grade as shown in Figure 3.

The fast leach rate has very significant inverse gradedependence to a high degree of correlation. The implicationsfrom this are: The fast leach rate is mass transfer controlled with acid

supply to the mineral surface through the boundary layer therate-determining step. This is evident from the steep gradedependency where the higher grade consumes acid at agreater rate and creates a wider concentration boundary layerthickness, which acts to reduce the observed leach rate.

The leach rate for the slow regime is particle diffusioncontrolled with only a small, acid supply, mass transfercomponent. The diffusion leach rate within the particles doesnot appear to be grind size dependent; probably due to thesurface expression of the malachite mineral and the short andwide pores into the leached particles.

The fast leach rate in the commercial plant will depend onthe agitation intensity and the availability of acid to thesurface of the ore particles. Thus the selection of the mixerswill influence the leach rate that can be predicted for theproject. As such high intensity mixers with high turnover ratewill need to be used.

The fast diffusion leach rate for a head grade of 2.15 per centASCu should not be extrapolated from this data with anydegree of confidence. However, it is required in order to

model the leaching process for ROM grade material. Fromthe available data the extrapolated Grade Normalised fastleach rate would be of the order of 0.029. This figure hasbeen reduced to 0.025 and used in the subsequent analysisreported here. Similarly a figure of 0.003 has been used forthe Grade Normalised slow leach rate portion.

The change point in leach rate from fast to slow is alsorequired for the analysis. It appears to occur at 120 minutesleaching except for the high grade -75 micron material; wherethere is no change from the fast leach rate. In this case therecovery is still quite low (70 per cent) and the leaching iscontinuing to be controlled by the supply of acid through theboundary layer. The change in leach rate for the lower gradematerial is around the 90 per cent (Tr(R) = 0.184) recovery level.This is an important observation as discussed later. The modeluses a change at the 90 per cent recovery level.

ACID CONSUMPTION RATE ANALYSISThe gangue acid consumption GAC is the nett consumption ofacid that is not returned via the solvent extraction (SX) plant.It is an economic cost and needs to be modelled closely toconfirm the economics of the process. All the GAC consumptionprofiles are shown in Figure 4.

It is evident from inspection of these curves that the GAC is alinear function of time. This means that as the leach timeincreases the total acid consumption increases directly. There isthus a relatively easy method of modelling the GAC with thislinear time related consumption.

The consumption consists of two parts. The first is the GACoor that associated with the early leach where gangue is the mainconsumer of acid. The second is the linear time related increasein GAC and is the rate part of the equation.

The GACo appears to vary with a number of parameters.Figure 5 shows variation with grind size and Figure 6 with grade.

2 Brisbane, QLD, 7 - 8 November 2005 First Extractive Metallurgy Operators Conference

G MILLER

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 100 200 300 400

Time - min

Tr(R

)

3.6/75 4.2/150 6.5/75 6.6/150

FIG 2 - Leach recovery curves using the shrinking core model.

0.000

0.005

0.010

0.015

0.020

0.025

0 1 2 3 4 5 6 7

Grade % ASCu

fast slow Linear (fast) Linear (slow)

Gra

de

No

rma

lise

dL

each

Ra

te

FIG 3 - Grade normalised diffusion leach rate grade dependence.

0

10

20

30

40

50

60

70

0 100 200 300 400

Time - min

GA

C-

kg/t

3.6/75 4.2/150 6.4/75 6.6/150

Linear (4.2/150) Linear (3.6/75) Linear (6.4/75) Linear (6.6/150)

FIG 4 - Gangue acid consumption rates.

0.0

5.0

10.0

15.0

20.0

25.0

30.0

0 20 40 60 80 100 120 140 160

55% Passing micron

GA

Co

kg

/t

FIG 5 - GACo variation with size.

The GACo dependence on grind size is expected with greaterexposure of the fast leaching copper mineral (malachite) leadingto a lower GACo for the finer material. The dependence on gradeis a little more subtle. The higher-grade copper materialconsumes a greater proportion of the available acid than thelower-grade copper material. As a result there is an inverserelationship between grade and GACo that needs to be taken intoaccount.

For a grind size selection of 55 per cent -75 microns a GACoof around 12 kg/t would be indicated. However, the gradedependency would suggest a GACo of 24 kg/t. This latter figurehas been used in the modelling.

The time linear GAC rate is shown in Figure 7 as affected bygrade, and in Figure 8 as by size.

As expected the copper grade has little effect on the GAC rateand is not included in the model. In this region once the initialGACo has been satisfied there is little or no dependence of GACrate on grade. However, there is a significant increase in GACrate with decrease in grind size. The data is still very scatteredbut it would suggest an average GAC at -75 microns of0.11 kg/t/min; which has been used in the subsequent modelling.

EFFECT OF ACID DILUTION ON GANGUE ACIDCONSUMPTION

Other recent work has shown that the addition of prediluted acidto the leach can lower the GAC (Grosse, 2004). The present workwas conducted with 8 M dilute sulfuric acid rather than with thenormal method of adding concentrated acid directly to the leachvessel. Figure 9 shows the affect of adding acid at variousconcentrations on the GAC. Figure 10 shows the mapping ofcopper recovery and GAC for the same tests.

As is evident from these results there is a significant reductionin the GAC when the acid is prediluted before addition to the oreslurry. The recovery of the copper (under constant leach timeconstraint) shows no real affect that can be demonstrated withinthe limits of the test accuracy. This has been confirmedanecdotally by Littleford (2005) for a similar operation leachingoxide zinc ores.

The response of the leaching and GAC to the acidconcentration is consistent with the modelling developed forheap leaching (Miller, 2002). Lower acid concentrations show asignificant lowering of the GAC and GAC rate for a fixed leachtime. Lower acid concentrations also lower the overall leach rate.However, for the test parameters used the total leach time wassufficient to compensate for the slower kinetics. The use ofdiluted acid will show benefits in the economics of the leachingoperation without adverse effects on the copper recovery;provided that the leach time allowance is sufficient to leach tothe economic limit.

First Extractive Metallurgy Operators Conference Brisbane, QLD, 7 - 8 November 2005 3

ANALYSIS OF AGITATION LEACHING DATA METHODS AND INTERPRETATION

0

5

10

15

20

25

30

0 1 2 3 4 5 6 7

Grade % ASCu

GA

Co

kg/t

FIG 6 - GACo variation with grade.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0 1 2 3 4 5 6 7

Grade % ASCu

GA

Ck

g/t

/min

FIG 7 - GAC rate variation with grade.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0 50 100 150 200

55% Passing microns

GA

Ck

g/t

/min

FIG 8 - GAC rate variation with size.

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

0 10 20 30 40 50 60 70 80 90 100

[Acid] %

GA

Ck

g/t

FIG 9 - GAC variation with acid dilution.

0

5

10

15

20

25

30

35

91.0 91.5 92.0 92.5 93.0 93.5 94.0 94.5

Recovery %

conc dil

GA

Ck

g/t

FIG 10 - GAC-recovery with acid dilution.

LIMIT OF ECONOMIC LEACHINGAs discussed in the literature (Miller, 1998) there is a limit toeconomic leaching when the cost of the marginal GAC consumedexceeds the nett revenue from the marginal copper leached. Thisanalysis generally results in a limiting nett acid:copper ratio ofbetween 15:1 and 25:1 depending on the relative copper and acidprices. For a net marginal copper income of $US0.80/lb (afterrecovery costs) and an acid cost of $US100/tonne, the limitingacid copper ratio is 17.3:1. Once the acid consumption exceedsthis then the leaching is no longer economic.

The comparison of the alternative leaching recipes is bestshown in a plot of the GAC as it varies with the CopperRecovery. This is shown in Figure 11 for the four test results.

It is evident that the latter part of all the leach tests showsgreatly elevated levels of GAC for modest increases in the copperrecovery. The average result is an increase in recovery offive per cent ASCu (90 per cent to 95 per cent = 5 kg/t) for anincrease in GAC of 20 kg/t (37 kg/t to 57 kg/t). Further leachingresults in higher acid:copper ratios; leading to non-economicrecovery.

LEACH PARAMETER SELECTIONThe selection of leach parameters needs to be made so as not toexceed the economic recovery. A leach model has beendeveloped to assess the relative merits of longer leach times.This is done by using the two-rate diffusion control leach modelcombined with the time-related GAC model. The differentialrecovery and GAC are calculated and plotted against the absoluterecovery. The diffusion leach model is shown in Figure 12; whilethe resultant copper recovery curve is shown in Figure 13. TheGAC model is shown in Figure 14. The overall GAC-Recoveryand acid:copper ratio (A:Cu) are shown in Figure 15.

Analysis of the model output shows clearly that as soon as theleach rate changes from the bulk mass transfer control to theparticle diffusion control the acid:copper ratio and the GAC riseexponentially. This is clear evidence that the leaching needs to beconducted in a regime that allows the copper to be extracted at asfast a rate as possible. This change point is the switch from acidtransport boundary layer diffusion to particle diffusion control.

CONCLUSIONSThe leaching of oxide copper in agitated systems has been foundto be diffusion controlled. However, there are significantdifferences in the control mechanisms. Initially the leachingcontrol is the supply of acid to the mineral surface. This islimited by the diffusion through the surface boundary layer andis thus subject to the level of agitation. Higher agitation provides

4 Brisbane, QLD, 7 - 8 November 2005 First Extractive Metallurgy Operators Conference

G MILLER

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 60 70 80 90 100

Rec % ASCu

150-6.6 75-6.3 150-4.2 75-3.6

average -500 Expon. (75-6.3) Expon. (150-6.6)

Expon. (75-3.6) Expon. (150-4.2) Expon. (-500)

GA

Ck

g/t

FIG 11 - GAC and recovery results.

0

0.05

0.1

0.15

0.2

0.25

0.3

0 100 200 300 400 500

Leach Time - min

Tr(

R)

FIG 12 - Diffusion leach model.

0

20

40

60

80

100

0 100 200 300 400 500

Leach Time - min

Reco

very

%A

SC

u

FIG 13 - Model copper recovery curve.

0

20

40

60

80

100

0 100 200 300 400 500

Leach Time - min

GA

Ck

g/t

FIG 14 - Model for GAC.

0

50

100

150

200

250

0 10 20 30 40 50 60 70 80 90 100

ASCu Recovery %

GA

Ck

g/t

0

25

50

75

100

125

Acid

:Co

pp

er

GAC A:Cu

FIG 15 - GAC and acid:Cu ratio with copper recovery.

thinner boundary layers and faster diffusion rates. Similarlyhigher grade material creates a thicker concentration boundarylayer and relatively slower leaching.

There is an economic limit to the agitation leach extraction ofthe copper in the ore. There is clear evidence that the leachingneeds to be conducted in a regime that allows the copper to beextracted at as fast a rate as possible. This change point is theswitch from acid transport boundary layer diffusion to particlediffusion control.

Acid consumption in agitated leach systems is also afirst-order system. However, the interpretation of the RTD iseasier from batch testing to continuous multiple CSTR seriesplant operations. Altering the method of acid delivery cansignificantly reduce acid consumption rates. Predilution of theacid minimises the local acid concentration near the additionpoint and can reduce the consumption by up to 50 per cent.

Modelling of the combined leach rate(s) and acid consumptionrates in agitated systems allows comparison of various leachingstrategies and selection of the most appropriate. The modelallows the identification of economic limits to leach recovery;where the cost of acid consumption is greater than the value ofthe copper recovered. From this analysis selection of residencetimes for agitation leaching can be determined. The modellingtools are useful in identifying optimum conditions for theleaching process.

The test results are limited in applicability to the plant designdue to their significantly higher head grade.

ACKNOWLEDGEMENTSThe author would like to thank the operating companies formaking the results of the testing available for analysis.

REFERENCESDixon, D G, 2003. Heap leach modelling the current state of the art, in

Proceedings 5th International Symposium Honouring Professor IanM Ritchie: Hydrometallurgy 2003, Volume 1, Leaching and SolutionPurification (eds: C Young et al), pp 289314 (TMS: Warrendale,PA).

Grosse, J A, 2004. Personal communication.Littleford, D, 2005. Personal communication.Miller, G M, 1998. Ore character effects on copper heap leach

management, in Proceedings Mine to Mill Conference, pp 121-126(The Australasian Institute of Mining and Metallurgy: Melbourne).

Miller, G M, 2002. Recent experience in copper heap leach performanceevaluation, in Proceedings ALTA Copper HydrometallurgyConference.

Miller, G M, 2003. Ore geotechnical effects on copper heap leachkinetics, in Proceedings Hydro 2003, TMS Conference onHydrometallurgy, Vancouver.

Miller, G M and Newton, T, 1999. Copper heap leach testinginterpretation and scale up, in Proceedings ALTA CopperHydrometallurgy Forum, Surfers Paradise, Queensland, October.

First Extractive Metallurgy Operators Conference Brisbane, QLD, 7 - 8 November 2005 5

ANALYSIS OF AGITATION LEACHING DATA METHODS AND INTERPRETATION

6 Brisbane, QLD, 7 - 8 November 2005 First Extractive Metallurgy Operators Conference