Escolar Documentos
Profissional Documentos
Cultura Documentos
Bellec S., Hodouin D., Bazin C., Khalesi M. R. and Duchesne C. Modelling and simulation of gold ore leaching. World Gold Conference 2009, The Southern
African Institute of Mining and Metallurgy, 2009.
A leaching model for gold ore that takes into account the impact of cyanide and oxygen
concentrations as well as gold surface exposure on leaching performance is developed. Cyanide
consumption is estimated by an empirical model that is a function of ore particles size distribution
as well as cyanide, copper and sulphur concentrations. The proposed model has been validated by
data from laboratory leaching tests on the Sleeping Giant gold mine ore.
Keywords: gold leaching, modelling, exposure surface.
the exposure of gold surfaces to enhance dissolution through ore porosity is related to the hindered transport of
kinetic. In this type of process, the ore is crushed and cyanide to the reaction site.
ground to expose the surface of gold grains to a chemical Gold liberation (or accessibility) has an important impact
attack by cyanide in presence of oxygen. This chemical on leaching kinetics; therefore, it is interesting to be able to
reaction produces a soluble complex of gold cyanide which characterize it. Many authors have developed mathematical
is subsequently recovered by adsorption or cementation. models for the liberation of minerals such as those proposed
Others use an approach, called heap leaching, which by Wiegel and Li 12 , King 13, 14 , Barbery 15 , Gay 16 ,
consists of spraying a cyanide solution on a roughly Stamboliadis17. Unfortunately the applicability of those
crushed ore pile. In this approach, cyanide and oxygen models for estimation of gold exposure properties is
diffuse slowly through the pores and fractures of the ore to limited, since most of them were developed for mineral
reach gold surfaces. concentrations significantly larger than those observed for
Several authors studied gold dissolution kinetics. Some gold ores and with a non-preferential breakage assumption
considered pore diffusion transport in heap leaching which is not applicable due to gold grains ductility. Because
(Prosser1, Bouffard and Dixon2) while others addressed the of such particularities, only a few studies on the liberation
leaching problem by the concept of gold surface liberation. of gold are available. Lorenzen and Deventer18 modified
In those conditions, experimental results showed a Kings14 model by adding an empirical term that accounts
reduction of leaching kinetics as the chemical reaction for the fraction of leachable gold that is not directly
advances. To model that behaviour, many authors (Brittan3, exposed to cyanide. Wen et al. 19 developed a cubic
Nicol et al. 4 , Hodouin et al. 5 , Ling et al. 6 ) proposed exposure model to predict the proportion of exposed
mathematical equations based on the evolution of leachable inclusive gold. This section proposes a simple geometric
gold concentration. When it decreases, the chemical model to characterize the proportions of free gold grain,
reactions rates slow down. Changes in the kinetic exposed inclusive gold as well as inclusive leachable gold
mechanisms are also proposed to explain the slowing down through ore fractures or porosity.
effect. The initial reaction may be governed by diffusion Based on simple geometrical assumptions for gold grains
through boundary layers if pulp agitation is not sufficiently and ore particle shape size and distribution, it is possible to
vigorous (Brittan 7 ) while it is controlled by diffusion approximate the gold surface degree of exposure and its
through ore porosity in the later stages. Habashi8 presented accessibility through ore porosity. Those assumptions are:
a gold dissolution model based on an electrochemical Ore particle shape is approximated by cube of side
mechanism as well as cyanide and oxygen diffusivities dp
through a boundary layer. Rubisov et al. 9 applied that Gold grains are approximated by cubes of side dg
diffusion model to pure gold by adding a stoichiometric Gold grain orientation is the same as the ore particle
factor representing the number of leached gold moles per orientation; therefore the cubes edges are parallel
mole of consumed oxygen. Crundwell and Godorr 10 The space distribution of gold grain centres inside
proposed a model that considers the formation of a ore particles is uniform
passivation layer of materials on the surface of gold Gold is unbreakable, therefore fragmentation is
particles as they dissolved. De Andrade Lima and intergranular
Hodouin11 went one step further by including the effect of Total liberation of gold grains occurs when at least
particle size in the kinetic model. This indirectly takes into 50% of the volume of gold is out of an ore particle
account the liberation effect on leaching behaviour. Gold grains positioned farther than a threshold
This paper proposes a kinetic model that takes into distance (p) from the ore surface are not accessible
account the fast rate of dissolution of exposed gold surfaces to chemical attack by cyanide.
and slower one of gold accessible through ore porosity. The According to those assumptions, Figure 2 shows the
reduction of leaching kinetic is explained by the different zones where gold grains centres can be positioned.
disappearance of exposed gold surfaces as the reaction The centre position influences the degree of gold
evolves. This approach first requires a geometrical model to leachability, a property related to gold surface accessibility
describe the surface liberation phenomena. This model is to reagents.
presented and explained in Section 2. Section 3 introduces
the leaching model, while Section 4 proposes an empirical
cyanide consumption model. Finally, the model is
calibrated and validated on batch leaching tests.
[8]
[9]
Figure 2. Accessibility of gold in an ore particle
as a function of gold grain centre positions
[10]
From geometrical considerations, it is possible to obtain where A A and A C are the surface areas of anodic and
the proportion of inaccessible ( ) and totally liberated cathodic reaction, respectively D CN and are the
gold grains ( ): diffusivities of cyanide and oxygen in the pulp phase, and
0 is the thickness of the boundary layer, which is
[1] considered as dependant on impeller rotation speed.
The dissolution kinetic of exposed gold surfaces can then
be calculated as a function of the oxygen efficiency:
[2]
[11]
The proportion of gold grains accessible through ore
porosity ( ) is obtained by subtracting the unreachable Then, the rate of gold dissolution (in moles per unit of
volume from the unexposed one: time and gold surface available to the reaction) is given by:
[3] [12]
[19]
[20]
[21]
It is difficult to develop a rigorous cyanide consumption Batch leaching simulation equations are obtained by
model due to the impressive variety of complexes that can writing species mass balance equations. Using the kinetic
be formed with cyanide. De Andrade Lima and Hodouin11 expressions for the three classes of gold (liberated, exposed
proposed a pseudo-homogenous empirical model to and accessible through pores) in the solid phase, one
describe the cyanide consumption process. Their model obtains:
takes into account cyanide concentration and ore particle
size. This section introduces a model which takes into
[28]
account these two variables as well as the copper ( ) and
sulphur ( ) concentrations in the ore. Cyanide
consumption ( ) per unit of time and mass of liquid for
ore particles of size dp is given by: [29]
[25]
[32]
[27]
Table I
Numerical values of estimated parameters
p m
g m
g m m
1
2
m
3+1
m /min
1
2
m2/min 3
m2/min
0.1200 m 1
0.1350 m 2
0.1500 m 1
0.1650 m 2
where xdp is the mass fraction of ore in class of average considered as predictions. Figures 10 and 11 respectively
diameter dp, mCN(t) is the mass of cyanide added to the compare the results predicted by the model with the
reactor at time t and m liq is the mass of liquid in the measured liquid concentrations of gold and cyanide. The
leaching tank. agitation speed of the stirrer influences the dissolution rate
of gold but not the overall gold leachability. The large
difference in final gold liquid concentration at 1600 rpm
Model calibration and other agitation speeds is due to the different initial
The model parameters were estimated using the data concentrations of gold (see Figure 5) and their respective
obtained for batch tests as well as the above simulation and proportion of inaccessible gold grains.
liberation equations. The results of only nine tests were From those results, one can see that the calibrated model
used, keeping the others for validation: the two leaching can be use to predict the leaching behaviour of a gold ore
tests at 1500 rpm as well as the 1-hour and 4-hour tests at based on its gold grain size distribution and a geometric
the three other agitation speeds. The twenty-four liberation model.
parameters characterizing the equations are estimated by a
non-linear least square regression method minimizing the Conclusions
sum of squared differences between the gold and cyanide This paper proposed a leaching model based on gold surfaces
concentrations obtained experimentally and by simulation. accessibilities and controlled by reagent diffusivities to the
In total, 164 measurements are used to approximate the reaction site. The model allows an adequate prediction of
parameters. Figure 7 summarizes the methodology used to gold dissolution and cyanide consumption for the studied ore.
calibrate the model. The model parameter estimates are The integration of the model in a steady state simulator
presented in Table I. D is the product of two parameters (R
and D O2 ). As it can be seen in the leaching kinetic
equations, those two parameters cannot be estimated
independently. g and g are respectively the mean and
standard deviation of the beta distribution used to
approximate the gold size distribution. The variation of
sigma zero with the impeller rotation speed confirms that
external diffusion is a dominant kinetic mechanism.
Figures 8 and 9 respectively compare the gold and
cyanide concentrations in the liquid phase obtained by
simulation to the experimental data for the tests at 1200,
1350, 1500 and 1650 rpm. One can see that simulated
behaviours are reasonably fitting the measured values. The
saw tooth behaviour of the cyanide consumption model is
caused by punctual addition of cyanide in the leaching
thank.
Model validation
To validate the proposed model, the three 2-hour length
cyanidation tests at 1200, 1350 and 1650 rpm are simulated
with the calibrated parameters. Since those three tests were
not used to calibrate the model, the simulations results are Figure 7. Methodology used to approximate models parameters
16. GAY, S.L. A liberation model for comminution based leaching behaviour. International Journal of Mineral
on probability theory. Minerals Engineering. 2004, Processing. 1994, vol. 41, pp. 115.
vol. 17, pp. 525534.
19. WEN, S.B., HSIH, C.S., and KUAN, C.C. The
17. STAMBOLIADIS, E.T. The evolution of a mineral application of a mineral exposure model in a gold
liberation model by the repetition of a simple random leaching operation. International Journal of Mineral
breakage pattern. Minerals Engineering. 2008, vol. Processing. 1996, vol. 46, pp. 215230.
21, pp. 213223.
18. LORENZEN, L. and VAN DEVENTER, J.S.J. The 20. MARSDEN, J. and HOUSE, I. The chemistry of gold
interrelationship between mineral liberation and extraction. Ellis Horwood, 1992.
Steve Bellec
PhD Student, Laval University
MSc from cole Polytechnique de Montral in chemical eng.: Online data reconciliation
PhD student at Laval University in Metallurgical eng.: thesis on optimisation of gold cyanidation
process.