Hydrometallurgy, 29 (1992) 275-295 275

Elsevier Science Publishers B.V., Amsterdam

The modelling of particulate leaching reactors--

the population balance approach

F.K. Crundwell and A.W. Bryson

Department of Chemical Engineering, Universityof the Witwatersrand, Wits, South Africa
(Revised version accepted February 12, 1992)

ABSTRACT

Crundwell, EK. and Bryson, A.W., 1992. The modelling of paniculate leaching reactors--the popu-
lation balance approach. In: W.C. Cooper and D.B. Dreisinger (Editors), Hydrometallurgy, Theory
and Practice. Proceedings of the Ernest Peters International Symposium. Hydrometallurgy: 29: 275-
295.

A mathematical model of leaching reactors using the population balance is presented. The model
incorporates the material balances for the reactants and products, the population balance which de-
scribes the change in particle-size distribution, and the energy balance. The effects of the particle
kinetics, the particle-size distribution, and the residence-time distribution are accounted for.
The approach is discussed for the design of multi-stage, continuous reactors, and examples of the
application of the model are given for various forms of the leaching reactions.
The method is then used to develop an a priori model for the pressure leaching of sphalerite, and
the results are compared with those of the Cominco's plant at Trail, British Colombia, Canada. The
model incorporates the leaching reactions, the oxidation of ferrous sulphate, the precipitation of
plumbojarosite, and dissolution of oxygen from the gas phase into the solution. The kinetic parame-
ters for each of these reactions have been obtained from independent studies. The model predicts
results that compare favourably with data from the plant operation.

INTRODUCTION

One of the first unit operations in the hydrometallurgical processing of

minerals is leaching, and a great deal of attention has focused on the deter-
mination of the kinetics of leaching reactions. The purpose of these investi-
gations has been to determine the important parameters for the design, oper-
ation and optimisation of plant performance. In order to achieve this, it is
necessary to incorporate these kinetic results into a model of the plant. How-
ever, in spite of the large number of kinetic studies, very little success has been
achieved in modelling plant performance.

Correspondence to: F.K. Crundwell, University of Witwatersrand, Department of Chemical En-

gineering, Private Bag 3, Wits 2050, South Africa.

0304-386X/92/$05.00 © 19~2 Elsevier Science Publishers B.V. All rights reserved.

276 F.K. CRUNDWELL AND A.W. BRYSON

Bartlett [ 1 ] has presented models for continuous leaching in both tank and
plug-flow reactors based on residence-time distributions. Implicit in his treat-
ment is the assumption that the particle kinetics are linear (i.e., the rate of
shrinkage is independent of concentration). A similar assumption was made
by Ruether [2 ].
The optimisation of reactor configurations for leaching was discussed by
Henein and Biegler [ 3 ]. They considered both plug-flow and continuous
stirred tank reactor (CSTR) reactors, and incorporated leaching kinetics of
monosize particles. Papangelakis et al. [ 4 ] on the other hand considered the
leaching of an arsenopyrite ore and, because the rate of reaction is controlled
by the surface reaction, they were able to use an approach similar to that of
Ruether [2 ] to model the CSTR reactor.
However, in order to model the leaching of particles of different sizes in a
reactor where the reaction is controlled by product-layer diffusion film trans-
fer, or by mixed control, it is necessary to solve the population balance equa-
tion to determine the outlet size distribution. Such an approach was adopted
by Sepulveda and Herbst [ 5 ]. They solved the population balance equation
for the various controlling mechanisms, and for various configurations of tank
reactors.
The approach of using the population balance to determine the change in
particle-size distribution during leaching is further examined here. The ef-
fects of different reaction mechanisms, various reaction orders, and particle-
size distributions on the reactor performance are examined, particularly for
the case where there is stoichiometrically sufficient reactant in the feed.
The population balance is then combined with the material and energy bal-
ances for a number of components in order to model the performance of the
zinc pressure leaching reactor at Trail, B.C. The model developed is an a priori
model, in the sense that the rate constants have all been determined indepen-
dently, so that there are no adjustable parameters in the model. A sensitivity
analysis for the various rate constants is also performed.

MODEL FOR LEACHING IN A CONTINUOUS STIRRED TANK REACTOR

Kinetics of leaching
The leaching of liberated mineral particles by a reactant in solution is a
heterogeneous reaction that frequently results in the formation of solid reac-
tion products which form on the unreacted core. This can be represented by
the generalized reaction:
aA(s)+bB(aq)~cC(s)+dD(aq) (l)
The kinetics ofleaching are often described by the unreacted shrinking-core
model. This model envisages a boundary at the surface of the unreacted core
MODELLING OF PARTICULATE LEACHING REACTORS 277

that moves towards the centre of the particle and a layer of porous solid prod-
uct that forms on the unreacted core as the reaction proceeds. The leaching
reaction involves the mass transport of reactants to and products from the
mineral surface, and involves the reaction at the unreacted surface. One or
more of these steps may be rate-controlling.
For the purposes of modelling leaching reactors it is necessary to know the
rate of shrinkage of the size of the unreacted core, and the time for complete
conversion of the particles. The rate of shrinkage, R (I,L), of the size of the
unreacted core, l, can be described by the shrinking-core model [6] and is
given by Sepulveda and Herbst [ 5 ] as:
dl
R,(I,L)=~-[ l I
PA k - ~ + 2 D a C ~
-2(a/b)
I
(l-12/L)+k-~n(l/L)2
] (2)

where:
PA= the molar density of the mineral;
ks = the surface reaction rate constant;
Da = the effective diffusion coefficient ofthe reactant B in the porous product
layer;
Ca = the reactant concentration in the leaching reactor;
L = the initial particle size;
kr= the film transfer coefficient.
The first term of the denominator on the right-hand side of eq. (2) de-
scribes the contribution to the rate of shrinkage of the unreacted core by the
rate of the surface reaction, while the second and third terms of the denomi-
nator describe the diffusion of the reactants through the product layer, and
through the film layer, respectively. If the first term of the denominator is
larger than the others, the reaction is said to be controlled by the surface re-
action, if the second dominates, the reaction is controlled by product-layer
diffusion, and if the third dominates, the reaction is controlled by film trans-
fer. Of course, any combination of these processes may also control the reac-
tion, in which case the reaction is said to be under mixed-reaction control.
It is important to note that the rate of shrinkage is dependent on both the
initial particle size and on the unreacted core size if the reaction is product-
layer diffusion controlled or film transfer controlled. This means that the rate
of shrinkage is not linear, and that in order to calculate the conversion cor-
rectly, the population balance for the system has to be solved.
The time required for a particle of original size L to dissolve to such an
extent so as to have an unreacted-core of size l is given by [ 5 ]:
p,, [,_._, ),.,_,31
T(I,L) =2(a/b)Ca -~ + 12Da -3/2+L2 + ~f-L-5_! (3)
278 F.K. CRUNDWELL AND A.W. BRYSON

Population balance model for a CSTR leach vessel

Leaching processes comprise a disperse phase, which contains a number of

particles which have distributed properties, such a size, mineralogy, reactiv-
ity, chemical composition, etc. The number balance recognizes that these par-
ticles are countable entities, and, as such, their numbers are conserved.
If n(l,L)dldL represents the number of particles per unit volume of slurry,
with unreacted core size in the range I to l+ dl and with total size in the range
L to L +dL, then the number balance at st,~ad~-sm,,, in a well-mixed contin-
uous stirred tank reactor (CSTR) can be derived from Randolph and Larson
[7]:
O(Rinl) O(Rnnl)
Qono=Q~n~ + V 01 + v OL (4)

where:
Rj = the growth rate of the unreacted-core size;
RL = the growth rate of the total particle size;
Q= the volumetric flowrate of slurry to the reactor;
V= the volume of the reactor.
The subscripts o and 1 = the inlet and outlet conditions, respectively.
If the particles retain their original size, then RL = 0, and if the volumetric
flowrate of slurry through the reactor can be regarded as being constant, then
(2o- Qt, and the population balance becomes:
O(RInt)
Ono = Qn~ + v 01 (5)

with the boundary condition that R~nt-,0 as I--,oo. The population balance
differential equation can be written in dimensionless form by using the fol-
lowing dimensionless variables:

i n* La R'{= R~ T( O,L ) "t'*= v CB

= n---£ E QT(O,L) C~-CBo
L
L* =--
L
where:
L= the average particle size of the unreacted material;
T(0,L) =the time for complete conversion of the average particle size;
CRo= the reactant concentration in the inlet.
The population balance can then be written as:
O(R?nT) R?nT n*o
(6)
MODELLINGOF PARTICULATELEACHINGREACTORS 279

with the boundary condition R'fnT ~0 as/*-~oo. This differential equation

can be integrated using the method of integrating factors and has the solution:

nT(l*,L*)=R?r, exp(- T*]v*) exp( T*/'t*)n*dl* (7)

where: T * = l/R?dl*
From eq. (2) the rate of shrinkage can be written in dimensionless form as:
-c~
( (8)
3t~r) \L*J + 12t~,LD /*-- -~-~)+tsR
* .
where:
t~n = [,/(3krfl)
t~,LD -~ E 2 / ( 1 2 D a f t )
tlR=E/(k~#)
fl=L/ ( 3kr) +L2/(12DB) +L/ks
By definition, t~o + t~,LD+ tla = I. The significance of these four parame-
ters is that they express all the information on the kinetics of leaching in di-
mensionless form. t~o is the contribution to the time for complete dissolution
of the average size particle in the feed due to external film diffusion, t~LD is
that contribution due to product-layer diffusion, and tlR is that contribution
due to surface reaction.
The distribution of unreacted-core sizes is given by:
Llla~
nT(i*) = al, f nT(l*,L*)dL* (9)

From the number density distribution given by eq. (9) the conversion can
be calculated from:

fot'~a"l*3nT(1")dl*
x=l (~0)
f ~'~ l*3n, ( l, )dl,

Material balance for a CSTR leach vessel

If Vorepresents the volume fraction of the unreacted mineral in the slurry

being fed to the reactor, and if v~represents the volume fraction of inert solids
in the slurry, then the material balance for the reactant B is given by:
280 F.K. CRUNDWELL AND A.W. BRYSON

C[ = C~o- (*X (II)

where:
¢*_ pA(b/a)Vo
(l-vo-vx)CBo
The significance of (* is that, if the amount of reactant to the reactor is
stoichiometrically exact, then ~ = 1, and CB is reduced to zero if the mineral
is completely dissolved. If the reactant is in large excess, then (*~0, and the
reactant concentration is unaffected by the extent of the mineral dissolution
reaction.
If there is more than one reaction that takes place in the leach reactor, it
may be easier to express the material balance for each component in terms of
the extent of reaction. For J chemical reactions, the molar flow rate of com-
ponent i leaving the reactor, F~, is given by:
J
F,, = Fio + ~ a,jEj ( 12 )
j=l

where:
a,j= the stoichiometric coefficient of the ith component participating in the
jth reaction;
Fro= the molar flowrate of the ith component leaving the reactor;
¢j= the extent of the jth reaction.
a~j is positive for products and negative for reactents. The molar flowrate
can also be expressed in terms of the conversion of the ith component, X,:
F,,--F,o(I-X,) (13)

Energy balancefor a CSTR leach vessel

The energy balance for a steady-state adiabatic flow process is given by:
tlo=[t, (14)
where/:/o and/:/~ are the enthalpy content per unit time of the inlet and outlet
streams, respectively, for this process. The enthalpy is given by:

where d/t~ and ~p; are the heat of formation at 298 K and the molar heat
capacity for each of the components. Assuming the molar heat capacities are
approximately constant between the temperatures of the inlet and outlet
streams, and substituting eqs. (12) and (13) for a single reaction and eq.
( 15 ) into the energy balance, eq. (14) gives:
MODELLING OF PARTICULATE LEACHING REACTORS 281

~ E o Cp,( T, - 7 o ) = ~ ( - A H R ) = F j X j ( - J / t R ) (16)
i

where:
W ,~

/~/R = Ei Ogi/~/"~"Ei Og/f

d Tref
CpidZ
AHR represents the "heat of reaction" at the temperature of the reaction.
Model solution
The process parameters for the model are the slurry feed flowrate, Q, the
reactor volume, V, the feed concentration, Cao, the solids concentration, Vo,
the inlet size distribution of the mineral, and if an energy balance is required,
the molar flow rates of the components in the feed. The kinetic parameters
are the constants ks, DB and kr. If these parameters are known, then the ex-
pected outlet size distribution may be calculated from eqs. (7) and (9) and
the conversion obtained from eq. ( l 0). The outlet reactant concentration is
given by eq. ( l I ), and the operating temperature may be obtained from eq.
(16). These equations need to be solved simultaneously, and invariably an
iterative procedure is involved. Equations (7), (9) and (10) are solved nu-
merically, using trapezoidal integration with Richardson extrapolation.

MODEL RESU LTS

Single tank
Figures 1 and 2 illustrate the results for the model for an isothermal reactor
for the limits of rate control by surface reaction (t~R = l ), by product-layer
diffusion ( IpLD
* = 1 ), and film transfer (t~o = 1 ). The feed distribution is the
gamma distribution, which in dimensionless form is given by:

n*(l*)- qq I*q-texp(-ql *) (17)

-r(q)
The dimensionless variance, o-*2, of this distribution is equal to l/q.
In the case where there is an excess of reactant (i.e.,(* =0), there is little
difference between the conversion obtained for a first-ordersurface reaction
and for product-layer diffusion, except at very low dimensionless residence
times. However, when the reactant concentration in the feed is stoichiometr-
ically exact (i.e.,~* = 1 ), the difference between these two reaction mecha-
nisms for the same dimensionless residence time becomes significant.
In both cases, when the reaction is controlled by film transfer, the model
predicts a much lower conversion for the same dimensionless residence time
than the other two mechanisms.
The effect of the reaction order on the conversion for the case where (* = I
282 F.K. CRUNDWELL AND A.W. BRYSON

1.00- -

0.85- -

C
0
. m

C_
0.70 -
>
C
0

FirsL-order reacLion
0.55--
~e= 0
SurFace reecLion
...... ProducL-lauer dirr
........... Film Lransrer

o.#o I I I I I I I
0.00 O.TS I .50 2.25 3.00 3.75 4.50 5.25
O
Res,dence Lime, T

Fig. 1. Effect of the controlling mechanism on the conversion for excess of reactant. Feed distri-
bution: Gamma, a*:=0.5.

0.80=

0.?0=
C
0

01
L
~1
> 0.60-
t-
O

0,50-
............. ..........
PPoducL
2. - I a~ler
°;o
d i FF
........... Fi Im LrensFer

040 I I I I I I I
0.00 0,75 1.50 2.25 3.00 3.75 4.50 5.25

Residence Lime,

Fig. 2. Effect of the controlling mechanism on the conversion for stoichiometric amount of
reactant. Feed distribution: Gamma, a .2 = 0.5,
MODELLING OF PARTICULATE LEACHING REACTORS 283

is shown in Fig. 3 for the case in which chemical reaction is the rate-control-
ling mechanism (t*SR= 1 ). In this case, the rate of shrinkage is given by:

R 'f-CT,m (18)

and the time for complete conversion is:

T(O,L) 2(a/b) -
(19)

As expected, the reaction order, m, has a significant effect on the conver-

sion for the same dimensionless residence time: the rate of shrinkage for a
one-half order reaction would be less dependent on the diminished reactant
concentration than that for a second-order reaction.
The influence of the spread of the feed size distribution is illustrated in Fig.
4 for the case of reaction control by product-layer diffusion where (* = I. At
low dimensionless residence times the distribution with the highest variance,
0 *2, will have the highest conversion due to higher proportion of fine material
which dissolves more rapidly. At higher dimensionless residence time, the
distribution with the highest variance will have the lowest conversion, due to
a higher proportion of coarse material which dissolves more slowly. Similar
results are obtained for an excess of reactant, and for the other rate control
mechanisms.

0.90-~
.....................--''''''''''' " ,-, .

0.80- Go w

..o o o

C
0 #.#so~'°°
, n

0.70- ## ............
01 • ... ........
t ¢ ......
# .... ..
# . ....'"
> #.# .... .... ..'
#. ....."
t- I ....'"
O 0.60-
0

0.50- / / ~" 1 F" ~r.t, t.,, - o r~ d e rt r~ x r~

.." ...... One-hal C-order r×n
........... S e c o n d - o r d e r " r,×n

0.40 I I I I I I ,'
0.00 0.75 I .SO 2.25 3.00 3.75 4.50 5.25
O
Residence time, "c

Fig. 3. Effect o f reaction order on the conversion for a stoichiometric amount o f reactant. Feed
distribution: G a m m a , a .2 = 0.5.
284 F.K. CRUNDWELL AND A.W. BRYSON

0.80-

C
0 0.75-
. _

(/)
f..
>
t- 0 . 7 0 -
0
0

0.65-

0.60 I I I I I I I
0.00 O.?S I • 50 2.25 3.00 3.75 4.50 5.25

Residence time, "r m

Fig. 4. Effect of the variance of the feed distribution (gamma) on the conversion for stoichio-
metric amount of reactant.

! .00 -

~ ank
ank
#
3
0.90
Tank 2
C
0
, u

U)
LOS0-
rank ]
>
C
0
(.)

0.70-
Pnoduct. laden
i f'Fus i on
o= ]

0.60
', I I I I I I I
0.00 0.50 ! .00 I .50 2.00 2.50 3.00 3.50 4.00
$
Residence time,

Fig. 5. Co-current operation of leach reactors for stoichiometric amount of reactant. Feed dis-
tribution: Gamma, a -=0.5.
MODELLING OF PARTICULATE LEACHING REACTORS 285

Co-current operation of tanks in series

Figure 5 illustrates the conversion obtained for the co-current operation of

equal size tanks in series. The dimensionless residence time is that for a single
tank, so the total dimensionless residence time for N co-current tanks is given
by N'r*.

MODEL APPLICATION-REACTOR MODEL FOR THE ZINC PRESSURE REACTOR AT

COMINCO'S PLANT AT TRAIL, B.C.

The zinc pressure leach plant at Trail, B.C., has been in operation since
1981 [ 8 ]. The four compartment autoclave has a working volume of 100 m 3,
and the mean residence time at the design feed rate is about 100 min.
The concentrate is fed as a slurry of between 68 and 70% solids ground to
98% less than 44/~m. The sphalerite concentrate from the Sullivan Mine has
the following composition: 48.6% Zn, 5.7% Pb, I 1.6% Fe and 32% S. The
solution to the autoclave contains 50 g/l Zn and 165 g/l H2SO4.
The autoclave is operated under a total pressure of 1300 kPa and the oxy-
gen partial pressure is 750 kPa. Oxygen is supplied by spargers to the first
three compartments, and each compartment is agitated very well. The feed
acid has a temperature of 70°C and the heat of reaction raises the tempera-
ture in the reactor to the operating temperature of about 150°C. Plumbo-
jarosite precipitates in the second, third and fourth compartments and con-
tains 2% Zn [ 9 ].
The reactions that occur in the autoclave are the leaching of sphalerite by
ferric sulphate, the precipitation of plumbojarosite, the oxidation of ferrous
sulphate to ferric sulphate by dissolved oxygen, the leaching of iron sulphide
from the concentrate, and the dissolution of gaseous oxygen. These five reac-
tions can be written as:

1. ZnS+ 2Fe 3+ ~ Z n 2+ +2Fe 2+ +S

2. Fe 3+ + 1/6 PbSO4 + 1/2 SO 2- + 2H20~
1/3 Pbo.sFe3(OH)6(SO4)2 + 2H +
3. Fe 2+ + H + + 1/4 O2(aq)-~Fe 3+ + 1/2H20
4. FeS+ 2Fe 3+-~3Fe 2+ +S
5. O2(g)-~O2 (aq)
286 F.K. CRUNDWELL AND A.W. BRYSON

Model description
Kinetics of reactions
The kinetics of the dissolution of iron-containing sphalerite concentrates in
ferric sulphate and ferric chloride solutions have received a large amount of
attention [ l 0-14 ]. The reaction mechanism is thought to be electrochemical
in nature, and the iron content of the sphalerite increases the rate constant of
the surface reaction linearly [ 13,14 ]. Although there are no reported studies
on the rate of reaction of the Sullivan concentrate, it is assumed that this
concentrate will react at a similar rate to other concentrates with similar im-
purity levels of iron.
The concentrate from the Gamsberg Mine in South Africa contains 9.08%
iron, while that of the Sullivan Mine contains 11.6% iron. The Gamsberg
sphalerite concentrate has been studied extensively [ l l - 13 ] at various ferric
and ferrous sulphate concentrations and at temperatures between 60°C and
90 ° C. The kinetics of dissolution of this concentrate can be described by a
surface reaction of order one-half (i.e., m = ½), with a rate constant of
k's=6190 e x p ( - 5 5 3 3 / T ) (mol/ma) ~/2 m/rain and an effective diffusion
coefficient for the diffusion of ferric ion in the product layer of
Da= 5.46× l0 -7 e x p ( - 2 8 8 7 / T ) m2/min. The expression for the rate of re-
action when the surface reaction is controlling has been linearised by minim-
ising the integral of the sum of squared errors in the range of interest. In this
case the linearisation has been performed in the region of 3-10 g/l Fe 3+, which
results in a linear rate constant of ks= 542.3 exp ( - 5533/T) m/min.
The rate of shrinkage is then given by:
R~'- -C~[7.9X 10~Sexp(2887/T)+ 1.778X 10-3exp(5533/T)] (20)

4.74X lO-4exp(2887/T)(l*-I*"~
L*] + 1.778X 10-3exp(5533/T)
where C[ =CB/128.9. The time for complete conversion of the average par-
ticle size (22.76× l0 -6 m) at a ferric ion concentration of 128.9 mol/m 3 is:
i

T(0,L) =0.0258 exp(2887/T) + 1.38× l0 -5 exp(5533/T) (21)

In applying these kinetics it is assumed that the addition of a surface-active
agent to the leach results in a porous product layer on the sphalerite surface,
rather than a thick coating of molten sulphur. Further experimental work is
recommended in order to test this assumption.
The precipitation of lead jarosite is assumed to be sufficiently rapid to be
at equilibrium. The equilibrium constant can be estimated from the reactor
outlet conditions. The outlet from the leach autoclave contains 5 g/l Fe and
30 g/l H.,SO4 [ 8 ]. Assuming that most of the iron is in the ferric state, and
that the sulphate and lead concentrations remain constant, then the equilib-
MODELLING OF PARTICULATE LEACHING REACTORS 287

rium concentration quotient for the precipitation ofjarosite at 150°C is given

by:

K= 10--36°8 [ Fe3 + ]
- [H +]2 m3/m°l (22)

where the concentrations refer to the total concentration of that particular

ion, that is the pseudo-equilibrium constant has been written so as to include
the dissociation of the various ionic species in solution.
The kinetics of the oxidation of ferrous sulphate by oxygen has received a
lot of attention. A recent report by Dreisinger and Peters [ 15 ] has clearly
identified the factors affecting the rate of this reaction under zinc pressure-
leaching conditions. The rate of this reaction is second order with respect to
the ferrous iron concentration, first order with respect to the oxygen concen-
tration, and is affected by the copper concentration in solution. No data were
provided by Martin and Jankola [ 8 ] on tbe copper present in the Sullivan
concentrate. The Gamsberg concentrate contains 0.14% Cu, while another
South African concentrate, that from Black Mountain, contains 0.54% Cu.
Assuming that the Sullivan concentrate contains about the same amount of
copper, it is not unreasonable to expect a copper concentration of about 5
mol/m 3 Cu, at which concentration, a rate constant of about 2.5 X 10 -3 m6/
(mol 2 min) at 150 °C could be expected. The activation energy for this reac-
tion was measured by Verbaan and Crundwell [11] to be 68.8 kJ/molK,
Therefore the rate of ferrous sulphate oxidation may be written as:
rF~2+ =ka[Fe '+ ]2[02] mol/m 3 min (23)
where: ka= 7.4 × 105 exp( - 8 2 5 1 / T ) m6/(mol2 min)
The dissnlution of the FeS from the sphalerite concentrate is assumed to
occur at the same rate as the spha!erite dissolution reaction. The rate of dis-
solution of the oxygen from the gaseous phase into the leaching solution is
described by:
ro2=ks([O2]sat- [02]) mol/m a min (24)
For a well-agitated system, k5 was assumed to be high, and a value of 20
min-~ was estimated from the results of Dreisinger and Peters [ 16 ]. It will
be shown, however, the operation of the plant under normal operating con-
ditions is insensitive to this parameter.

Process parameters
The molar flow rates of the feed were obtained from Martin and Jankola
[8 ]. Dreisinger and Peters [ 16 ] provided data on the feed size distribution
(on a mass basis). This distribution is described by a Rosin-Rammler distri-
bution with a mean size of 22.76 × 10 -6 m, and a slope parameter of 1.563.
288 F.K. CRUNDWELL AND A.W. BRYSON

Population. material and energy balances

The material balances for each component and for each compartment were
written in terms of the extent of reaction. The material balances for each com-
ponent are given in Table 1.
The conversion of reaction 1 was obtained by solving the population bal-
ance equation by numerical integration of eq. (7) and from that obtaining
the conversion from eq. (10). The extent of reaction in a compartment is
related to the conversion of sphalerite by:
~., =F, oXz,s (25)
where Fro is the molar flow rate of ZnS in the inlet to a compart-'ent and X
is the conversion achieved in that compartment. The extent of reaction 4 is
related to the conversion of sphalerite by the molar ratio of FeS to ZnS in the
concentrate. Therefore:
~.4=0.28FtoXzns (26)
The extent of reaction 2 is obtained from eq. (19) and can be calculated from:
3.608
F6o-2el -e~ +e3-2c4 = (F4o-2~2 - E3)210-
Q~o~
,,,

(27)

where Qsol is the volumetric flowrate of solution. The extent of reaction 3 can
be calculated from the material balance for the ferrous ion species in solution.
The extent of reaction 3 can be evaluated from the equation:
O=e~_ksV(F7o_2Ci-¢3-3¢4)2(F,o-l/4e~+es)Q~o~ (28)
The extent of reaction 5 may be calculated from the material balance for
oxygen:

TABLE I

Molar flow rates and extents for each compartment

Compartment Molar flowrate and extent

I ZnS Ft = F t o - e t
2 0 : (aq) /:2 = F , o - !/4e~ + es
3 H20 F 3 - F3o- 2e2 + ½e3
4 H+ I"4=/'~o + 2e2- ~3
5 SO~- F5 =/:so- ~e2
6 Fe -~+ F6= I % o - 2 e l - c 2 + % - 2e4
7 Fe 2+ FT~-FTo+2el -c3+3e4
8 Jarosite Fa= Fso+ !/3~2
9S lq~= lq~o+ct +~4
10 Zn "+ Fio= FIoo+el- 0.058e2
! I FeS FI i = Fi io- ~4
MODELLING OF PARTICULATE LEACHING REACTORS 289

0 = e s - k s V{ [O2]sat--(F2o-1/4e3 + es)/Qsoi} (29)

Since the precipitated plumbojarosite contains about 2% Zn the conversion
of zinc to solution is given by:

x_Fio - Floo _ el --0.058E2

(30)
Flo Flo

The operating temperature of the autoclave may be obtained from the en-
ergy balance. Since a temperature profile between the different compartments
is not reported, it is assumed that, within the reactor, there is significant heat
transfer between the compartments but that the reactor a- 'Jhole is an adi-
abatic reactor. The adiabatic energy balance is given by:
FtotCp ( T - To) = e! ( - A/~°! ) + E3( - - zJI'~IO3 ) "~"E4( - - ZJ/'~rO4 ) (31)
where:
Ftot = the total molar flowrate to the autoclave;
T= the temperature within the reactor (equivalent in each compartment);
-AHR! =the heat of reaction for reaction 1 ( - 29.8 kJ/mol);
-zJHR3=the heat of reaction for reaction 3 ( - 102 kJ/mol);
-ZfHR4=the heat of reaction for reaction 4 ( -80.6 kJ/mol).
The specific heat capacity, Cp is average value for the inlet and, since the
molar flowrate is dominated by water, this is taken to be that of water.
The left-hand side of eq. (31 ) represents the rate of energy removal from
the reactor, and will be denoted R (T). R (T) is a straight line as a function of
the reactor temperature. The right-hand side of eq. (31 ) represents the rate
of energy generation due to reaction, and will be denoted G (T).

Model solution
Equations (25)-(31 ) express the material balances for each of the com-
ponents entering and leaving each compartment in the reactor and the overall
energy balance. The rate processes for each of these reactions have been de-
termined independently of this investigation and, therefore, this is an a priori
model of the reactor, that is, there are no adjustable parameters in the model.
The solution of eq. (25) involves solving the population balance to obtain
the conversion. Under the conditions of the Trail operation, it is expected
that the dissolution mechanism is controlled by both surface reaction and
product-layer diffusion, with product-layer diffusion dominating, that is,
t~,LD = 0.78.
The extents for reactions 1-5 may then be obtained at a particular temper-
ature from the conversion of the sphalerite particles, and the material balance
and equilibrium equations represented by eqs. (25)-( 30 ).
290 F.K. CRUNDWELL AND A.W. BRYSON

Model results

Martin and Jankola [ 8 ] provided operating data for the leach autoclave at
two different feed flow rates, and also noted that the autoclave could operate
at up 200% above its design capacity. Figure 6 represents their data, and illus-
trates the prediction by the model for a temperature of 150°C for various
flowrates of the feed slurry to the reactor on the reactor performance.
This figure indicates that the model is a good representation of plant data
that are available. The model predicts a lower conversion than that obtained
on the plant for the first two compartments, and a higher conversion than that
obtained in the last two compartments. Clearly, further data need to be ob-
tained in order to thoroughly test the model.
The solution of the energy balance, eq. (31), is illustrated in Fig. 7. The
model predicts an operating temperature of 140 ° C. The plant operates at be-
tween 145 ° C and 155 oC.
The details of the molar flow rates within the reactor are given in Table 2.
The discharge from the fourth compartment can be compared with that ob-
tained in the plant. The model results are in good agreement with the plant
data. It is interesting to note that the model predicts that there is no precipi-
tation ofjarosite in the first compartment of the reactor, which is observed in
practice. In addition, the model predicts that most of the consumption of oxy-
gen (es) occurs in the first compartment of the reactor. Very little oxygen is
consumed in the final compartment, and for this reason oxygen is not sparged
1 ,OO -'~
CompaPLmenL
A 4

C 0,90- 2
0
(/)
t.
(P
>
E
0
o 0.80

Model
Plane

0,70 -I I I I I
0.50 o.:,s 1.oo 1.2s 1.so 1.75

Slurr, U rlowrate, m3/min

Fig. 6. The effect of the slurry flowrate to the Trail autoclave.

MODELLING OF PARTICULATE LEACHING REACTORS 291
,,e.
!

C) RCT3
,e--,

32.0~
X

C G(T)
E 24.0 1
--)

t-~ 1 6 . 0 ~
I--

f_
o 8.00--

rv" 0.00 I I I I I I I I
90.0 100 1I0 120 130 1#0 150 160 170
T e m p e r , at. or, e, °C

Fig. 7. The solution of the energy balance for the Trail autoclave.

TABLE 2

Molar flow rates for each component at 105% of design capacity

Component Feed Molar flow rate (mol/min) in compartment Actual

Autoclave
1 2 3 4 Discharge

ZnS 913.5 180.5 34.5 14.0 6.01 4.9

H÷ 3093.4 856.4 758.7 737.8 729.9 562.4
Fe '~+ 0 173.8 154.5 146.2 143.0
J 82.3
Fe :+ 0 52.2 19.6 10.5 6.4
Jarosite 0 0 24.2 32.0 35.1 76.7
Zn 2+ 701.4 1507.8 1577.6 1596.8 1604.2 1615.4
FeS 255.5 29.47 8.71 2.95 0.71 2.8
02(aq) 0 7.92 8.61 8.70 8.72

02 consumption 564.3 63.2 17.6 7.0 944.8 ~

in compartment

IThis refers to the total plant consumption given by Martin and Jankola [ 8 ].

into the last compartment. The actual and calculated conversions for plant
operation at 105% of design capacity are given in Table 3.
Although this is an a priori model, it is instructive to examine the effect of
the kinetic parameters on the reactor performance. The kinetics of the leach
reaction are incorporated in the time for complete conversion of the average
particle size. Figure 8 shows that, as expected, this parameter has a strong
influence on the reactor performance. The kinetics of the leaching reaction
292 F.K. C R U N D W E L L A N D A.W. B R Y S O N

TABLE 3

Cumulative conversions obtained in each compartment at 105% of design capacity

Compartment

1 2 3 4

Calculated (%) 88.12 95.76 97.86 98.67

Actual plant 88.7 96.5 97.9 98.4
data' (%)

'Martin and Jankola [8 ].

% .00 - CompaPLmenL
4
3
0.9S--

r"
0

O9 0 , 9 0 -
C.

C
0 0,8S-

0,80-
Model
$ Plant
O. ?S Li |* - - ] i i
~o.o 2o.o 30.0 40.0 so.o
T¢O,Lav el, min

Fig. 8. The effect of the time for complete conversion of the average particle size on the conver-
sion for the Trail autoclave.

that have been used in the model have been extrapolated from data obtained
at temperatures of between 60°C and 90°C. Since this is such an influential
parameter, it is suggested that the kinetics of this reaction be examined in
detail, especially in the high conversion ( > 95%) region which is applicable
to the fourth compartment.
The effect of the rate constant for the ferrous sulphate oxidation reaction is
shown in Fig. 9. This figure illustrates that at a rate constant greater than
0.5 X l0 -3 m6/(moF min) this parameter has very little influence on the re-
actor performance. Such low rate constants are not expected.
The effect of the rate constant for oxygen dissolution is illustrated in Fig.
10, which shows that this parameter can have a strong influence on the reac-
tor performance if it is below 5 min-~. This would mean that the compart-
 MODELLING OF PARTICULATE LEACHING REACTORS 293

Compartment
1 .00- .----" 4
v
...-.3
0.90- "" 2

£ 0.80-
0

¢. 0 . 7 0 -

>
r" 0 . 6 0 -
0
0

0.50-

0.40-- * P I ant
Model
0.30 I I I I I I I I
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00

1000 x k 3, m6/Cmol 2 min)

Fig. 9. The effect of the rate constant for ferrous sulphate oxidation on the conversion for the
Trail autoclave.
Compartment

I.00 --
, .

'"" 2

t- ........... 1
O
'" 0.80-
(n
£.

>
r"
0
(..1
0.60 -

Model
P I ant.

0.40 I I I I I I I I
0.00 5.00 ,0.0 ,s.o 20.0 2s.o 30.0 35.0 40.0
-1
k S , min

Fig. 10. The effect of the rate constant for oxygen dissolution from the gas phase to the solution
on the conversion for the Trail autoclave.

ments were very poorly agitated, which is not the case, or that the gas sparging
into the contents of the compartment was very inefficient. Since this has such
a strong influence on reactor performance and is relatively easy to rectify, it
is unlikely that the reactor would be allowed to operate under these condi-
294 F.K. CRUNDWELL AND A.W. BRYSON

tions. Therefore, the plant performance is insensitive to k5 under normal plant

operating conditions.

CONCLUSIONS

The population balance has been used to describe the change in particle size
distribution for the leaching of a mineral in a continuous stirred tank reactor.
The effect of various operating parameters, such as the feed size distribution
and the reactant concentration, and various kinetic parameters, such as the
reaction order, and the reaction mechanism have been examined.
This model has been applied to the modelling of the zinc pressure leach
reactor at Trail, British Colombia, using independently determined kinetic
parameters. The model and the plant data are in good agreement for conver-
sion of sphalerite, for the consumption of oxygen and for the production of
jarosite. In order to produce a model that is numerically more accurate, it is
important to have a better understanding of the co-precipitation of zinc with
the plumbojarosite, and to have experimental data for the kinetics ofleaching
at the conditions of the reactor, especially in the very high conversion region.

NOMENCLATURE

a,b Stoichiometric coefficients

G~ Concentration of species B in solution, mol/m ~
Molar heat capacity, kJ/molK
DB Effective diffusion coefficient orb in the product layer, m2/min
F Molar flow rate, reel/rain
li Enthalpy content per unit time, kJ/rain
lq, External film transfer coefficient, m/rain
k, Surface reaction rate constant, m/rain
K Concentration quotient tbr jarosite precipitation m'Vmol
I Characteristic size of unreacted core, m
L Initial size of unreacted particle, m
n(I,L ) Number of particles ofcore size I and initial size L per unit volume, I/m 5
q Parameter in the gamma distribution (see eq. ( 17 ) )
Q Volumetric flow rate of slurry, mS/rain
r Rate of reaction, mol/m 3 rain
R,(I,L) Rate of shrinkage of the dimension I of the particle, m/rain
t Time, rain
T Temperature, K
T(I,L) Reaction time for a particle of size L to shrink to size I, min
l:I Volume fraction inerts
t,o Volume fraction unreacted mineral value
V Volume of reactor, m 3
X Conversion

Stoichiometric coefficient (see eq. ( 12 )

E Extent of reaction, mol/min
PA Molar density of A., mol/m 3
MODELLING OF PARTICULATE LEACHING REACTORS 295

0* 2 Dimensionless variance
T Residence time in reactor, min
Stoichiometric ratio (see eqn. ( 11 ) )

REFERENCES

1 Bartlett, R.W., Conversion and extraction efficiencies for ground particles in heteroge-
neous process reactors. Metall. Trans. B, 2B (1971): 2999-3006.
2 Ruether, J.A., Reaction in a cascade of continuous stirred tank reactors of particles follow-
ing the shrinking core model. Can. J. Chem. Eng., 57 (1979): 242-245.
3 Henein, H. and Biegler, L.T., Optimisation of reactor volumes for topochemical reactions.
Trans. Instn. Min. Metall., 97 (1988): C215-C223.
4 Papangelakis, V.G., Berk, D. and Demopoulos, G.P., 1990. Mathematical modelling of an
exothermic leaching reaction system: Pressure oxidation of wide size arsenopyrite partic-
ulates. Metall Trans B, 21B (1990): 827-837.
5 Sepulveda, J.E. and Herbst, J.A., A population balance approach to the modelling of mul-
tistage continuous leaching systems. In: T.W. Chapman, L.L. Tavlarides, G.L. Hubred and
R.M. Wellek (Editors), Fundamental Aspects of Hydrometallurgical Processes. AIChE
Syrup. Ser., 57 ( 1978): 41-55.
6 Levenspiel, O., Chemical Reaction Engineering. Wiley, New York (1972), Chapt. 12.
7 Randolph, A.D. and Larson, M.A., Theory of Particulate Processes, Analysis and Tech-
niques of Continuous Crystallization. Academic Press, New York ( 1971 ), 260 pp.
8 Martin, M.T. and Jankola, W.A., Cominco's Trial zinc pressure leach operation. CIM Bull.,
78(876) (1985): 77-81.
9 Au-Yeung, S.C.F. and Boiton, G.L., Iron control in processes developed at Sherritt Gordon
Mines. In: J.E. Dutrizac and A.J. Monhemius (Editors), Iron Control in Hydrometallurgy.
Ellis Horwood, Chicester, UK ( 1986 ), pp. 131-151.
l0 Bobeck, G.E. and Su, H., The kinetics of dissolution of sphalerite in ferric chloride solu-
tion. Metall. Trans. B, 16B ( 1985): 413-424.
i I Verbaan, B. and Crundweli, F.K., An electrochemical model for the leaching ofa sphalerite
concentrate. Hydrometallurgy, 16 ( 1986): 345-359.
12 Crundwell, F.K., Kinetics and mechanism of the oxidative dissolution of a zinc sulphide
concentrate in ferric sulphate solutions. Hydrometallurgy, 19 ( 1987): 227-242.
13 Crundwell, F.K., Effect of iron impurity in zinc sulphide concentrates on the rate of disso-
lution. AIChE J., 34 ( 1988 ), 1128-1134.
14 Palencia Perez, I. and Dutrizac, J.E., The effect of the iron content ofsphalerite on its rate
of dissolution in ferric sulphate and ferric chloride media. Hydrometallurgy, 26 ( 1991 ):
211-232.
15 Dreisinger, D.B. and Peters, E., The oxidation of ferrous sulphate by molecular oxygen
under zinc pressure-leach conditions. Hydrometallurgy, 22 ( 1989): 101-119.
16 Dreisinger, D.B. and Peters, E., The mathematical modelling of the zinc pressure leach. In:
J. Szekely, L.B. Hales and H. Henein (Editors), Mathematical Modelling of Materials Pro-
cessing Operations. TMS-AIME, Warrendale, Pa. ( 1987 ), pp. 347-369.