2005-HeapSim - Unravelling The Mathematics of Heap Bioleaching - pdf-1

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/291799359

HeapSim - unravelling the mathematics of heap bioleaching

Article · January 2005

CITATIONS READS
15 819

3 authors, including:

Jochen Petersen David G. Dixon


University of Cape Town University of British Columbia - Vancouver
114 PUBLICATIONS   2,674 CITATIONS    88 PUBLICATIONS   2,776 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

From lab to heap. View project

Flotation of oxidized ores View project

All content following this page was uploaded by Jochen Petersen on 07 February 2017.

The user has requested enhancement of the downloaded file.


HeapSim - Unravelling the Mathematics of Heap Bioleaching

Nneoma Ogbonna, Jochen Petersen


Department of Chemical Engineering
University of Cape Town
Private Bag Rondebosch 7701
South Africa
jp@chemeng.uct.ac.za

David G. Dixon
Department of Materials Engineering
University of British Columbia
6350 Stores Road
Vancouver, BC, Canada V6T 1Z4
dixon@interchange.ubc.ca

ABSTRACT
Although heap bioleaching has been recognized as an economic alterna-
tive for treating low grade mineral ores, the underlying physical, chemical
and biological processes involved are complex. A detailed investigation into
the dynamics of heap bioleaching processes has provided much insight into
the underlying complexities, and has led to the development of a sophisti-
cated modeling tool - HeapSim. This tool can be of great help in the design
and operation of heap bioleaching processes. In this paper, the mathemat-
ical model implemented in HeapSim is examined. The model takes into
account mineral kinetics, particle level effects, bacterial growth, oxidation
and adsorption, gas absorption, pore diffusion, bulk advection, gas balance
and heat conservation. Three case studies where the model has been applied
are briefly discussed.
INTRODUCTION

Heap bioleaching has been recognized as an economic alternative for treating


specific mineral ores. Its low capital and operating cost requirements make it especially
suitable for low grade ores, and it is applied mostly to copper extraction and the pre-
treatment of refractory gold ores [2]. It is also being considered for zinc and nickel [10].
However, significant drawbacks of heap bioleaching are long turnover times and large
ore inventories [16]. There is therefore a need to design efficient heaps in order to achieve
profitable operations. This requires a deep understanding of the processes taking place
in the heap.
Heap bioleaching is a complex interaction between transport processes and chem-
ical reactions that occur in the solid, liquid and gaseous phases. These processes include
liquid phase advective and diffusive transport, gas phase advection, heat generation and
transport, mineral solubilization, and microbially mediated chemical reactions. Scien-
tific investigation of heap and dump leaching has led to the development of mathe-
matical models, some of which are reviewed and compared in [5]. Several of these
models emphasize bulk scale transport phenomena over micro-scale phenomena such as
microbial growth and oxidation kinetics and diffusive transport through the stagnant
phase, and very few offer the flexibility of parameter specification that is required to
capture all the possible phenomena that may limit a given heap. This paper reviews
a rigorous model developed by Dixon and Petersen that takes into account microbial
dynamics and diffusive transport, in addition to heap hydrology and other bulk scale
processes. The model has been developed into a computer software (HeapSim), which
combines a Microsoft Excel user interface (for ease of parameter definition by the user,
and graphical presentation of results), with a FORTRAN computational engine. The
model has been validated against experimental data for chalcocite, chalcopyrite and
zinc sulfide. In this paper, the chemical reactions in the model are presented in terms
of an ore containing chalcocite, covellite and pyrite; its adaptation to other mineral
leaching situations is straightforward.
The sub-processes that occur in a heap may be classified into mineral scale
processes, particle scale processes, agglomerate scale processes and bulk scale processes
[8]. At the mineral scale, the primary consideration is mineral dissolution kinetics. At
the particle scale, emphasis is on the distribution of minerals within an ore particle,
and the change in reactive surface as leaching progresses. At the agglomerate scale -
that is a cluster of particles - modelling addresses microbial growth, propagation and
oxidation, gas (oxygen and carbon dioxide) uptake into the liquid phase, and diffusive
transport into and out of the ore particles and the surrounding stagnant solution. The
focus in bulk scale modeling is on bulk flow of solution, gas and heat. The model
reviewed here incorporates all four levels, as we shall see in the sections that follow.
CHEMICAL REACTIONS

Copper in sulfide minerals is oxidized by reaction with ferric ions. The chemical
reactions between chalcocite, covellite and pyrite contained in the ore, and ferric in the
leaching solution, are of the form
R1. Cu2 S + 1.6Fe(SO4 )1.5 → 0.8CuSO4 + 1.6FeSO4 + 0.2Cu6 S5
R2. Cu6 S5 + 12Fe(SO4 )1.5 → 6CuSO4 + 12FeSO4 + 5S0
R3. FeS2 + [2+6(2 - β)]Fe(SO4 )1.5 + 4(2 - β)H2 O → [3+6(2 - β)]FeSO4 + 4(2 -
β)H2 SO4 + β S0
β above is the pyrite abiotic elemental sulfur yield. It represents the fraction of the
second (pyritic) sulfur atom in pyrite that is oxidized only to elemental sulfur rather
than sulfate. It varies between 0.0 and 1.0 depending on the redox potential [15].
The sulfur and ferrous produced in the reactions above undergo microbially
mediated oxidation to sulfate and ferric.
R4. 2S0 + 3O2 + 2H2 O → 2H2 SO4
R5. 4FeSO4 + O2 + 2H2 SO4 → 4Fe(SO4 )1.5 + 2H2 O
Reactions R4 and R5 are pivotal as they represent the microbial regeneration of ferric
and acid, and so facilitate reactions R1-R3. In addition to the reactions above, acid is
consumed by gangue, and jarosite is produced.
R6. H2 SO4 + MO → MSO4 + H2 O (M is a divalent metal ion).
R7. 6Fe(SO4 )1.5 + 14H2 O → 2H3 OFe3 (SO4 )2 (OH)6 + 5H2 SO4
Based on the these chemical reactions, the transportable species of interest in
the leaching solution are copper sulfate [CuSO4 ], ferrous sulfate [FeSO4 ], ferric sulfate
[Fe(SO4 )1.5 ], sulfuric acid [H2 SO4 ], sulfur oxidizers [mesophiles, moderate thermophiles,
extreme thermophiles] and iron oxidizers [mesophiles, moderate thermophiles, extreme
thermophiles].
The rate of each heterogeneous leaching reaction j is given by:
1 dgj gj,0 dXj
rj = = . (1)
s dt s dt
g is the grade of the mineral, and Xj is the conversion of reaction j, or the frac-
tion of a mineral that has reacted. In Eq.(1), division by s (stagnant liquid mass
fraction [kg(stagnant liquid)/kg dry ore]) rectifies the unit of the reaction rate to
[mol/kg(stagnant liquid)/s].
The rate of mineral conversion incorporates functions for the effect of tempera-
ture, solution composition and mineral topology,
dXj
= kj (T ) fj (C) Wj (1 − X) . (2)
dt
k(T ) is the rate constant given by Arrhenius law
  
E 1 1
k(T ) = k0 exp − − (3)
R T T0
where k0 is the rate constant at the reference temperature T0 .
The concentration function f (C) takes different forms depending on the electro-
chemical nature of the reaction. The form implemented in the model is
CF e3+
f (C) = (1−m)
(4)
(kA + CF e 3+ ) (k B + CF e2+ )m
where kA is the ferric mass transfer parameter and kB is the ferric reduction parame-
ter [15]. Eq.(4) can represent any one of the three most common forms of electrochem-
ical rate control depending upon the values of the two parameters kA and kB . When
both are zero (and m ≈ 0.5), then anodic mineral decomposition is rate-controlling.
When kA = 0 and kB  CFe2+ , then ferric reduction is rate-controlling. Finally, when
kA  CFe3+ and kB  CFe2+ , then ferric mass transfer is rate-controlling.
The topological rate term W (1 − X) accounts for the change in reactive sur-
face area with time. Two classic models for fluid-particle reactions are the shrinking
core model and the shrinking sphere model [9]. Traditionally, these approaches have
been used in leach modeling, assuming relatively small, uniform and spherical particles.
However, their applicability to a coarse ore situation in a heap is limited since the par-
ticles are neither spherical nor of uniform size. Also, the topology of leaching is more
complex than these simple models allow. A generalized form is adopted,

W (1 − X) = (1 − X)φ . (5)

φ is an empirical parameter that falls between 0.5 and 2 [7].


The rates of the microbial oxidation reactions (R4 and R5) depend on micro-
bial population. The microbes may be attached to the ore or floating in the leaching
solution [12]. The rate equation for microbial oxidation reactions takes the general
form:
1 dN
r= + km · f (N ) (6)
y dt
where N is the microbial population, y is the specific cell yield (number of cells generated
per mole of oxidation product) and km is the specific maintenance rate. The oxidation
reactions proceed as energy sources for the microbes [17]. This energy is used either for
cell growth or cell maintenance. These two demands are reflected in Eq.(6).
The rate of change of the microbial population may be expressed as
dN
= N · {growth term − death term} (7)
dt
The growth term in Eq.(7) is defined by the specific growth rate kg of the microbes.
Microbial growth depends on the availability of nutrients (e.g. ferrous, sulfur, oxygen,
carbon source, acid), temperature, and the presence of growth-inhibiting factors (e.g.
toxic salts, acid).
The effect of growth-limiting nutrients on the specific growth rate is described
by the Monod model
CL
kg = kgmax fL,b = kgmax · , (8)
KL,b + CL
where kgmax is the maximum specific growth rate achievable when the concentration
of the growth limiting nutrient is not limiting, CL is the concentration of the growth-
limiting substrate, and KL,b is the saturation/Monod constant for bacterial species b.
The effect of growth-inhibition factors is given by
KI,b
fI,b = (9)
KI,b + CI

where KI,b is the growth-inhibition constant for bactrial species b and CI the concen-
tration of the growth inhibition factor. For CI  KI,b , fI,b → 1 and growth-inhibition
is low. For CI  KI,b , fI,b → 0 and growth-inhibition is high.
The death term in Eq.(7) depends on the temperature of the heap, assuming
that increasing temperature beyond a certain threshold will result in rapid microbial
death.
Acknowledging that kg is modified by the growth temperature fg (T ) (in other
words, microbial growth has an optimum temperature range, with reduced growth
outside this range), the death temperature fd (T ), and the endogenous decay factor ke ,
the following model for the change in microbial population is implemented:
dN
= N kgmax [fg (T ) (Π(1 + ke ) − ke ) − fd (T )] (10)
dt
where Y Y
Π= fL,i fI,j .
i j

Natural death of the microbes (specified by the endogenous decay factor ke ) is


modeled in Eq.(10) to take effect when Π 6= 1. This means that microbes die off when
the growth conditions are not ideal.
We define the fL,b and fI,b terms for sulfur and ferrous oxidizing microbes re-
spectively thus:
C O2 gS 0 KI,S,O2
fL,S = · ; fI,S = (11)
KL,S,O2 + CO2 KL,S,S 0 + gS 0 NS + KI,S,O2
   
CH2 SO4 CF e2+ C O2
fL,F e = 1 − exp( ) ;
KL,F e,H2 SO4 KL,F e,F e2+ + CF e2+ KL,F e,O2 + CO2
KI,F e
fI,F e = (12)
NF e + KI,F e
where the subscripts S and F e refer to sulfur and iron oxidizing microbes respectively.
Hence the rates (r4 and r5 ) of the sulfur/ferrous oxidation reactions are given by
B max
Πb kg,b
X  
r= Nb fg,b (T ) + km,b (13)
b=1
yb

where b refers to the microbial species.


Reactions R4 and R5 make use of dissolved oxygen which is absorbed from the
gas phase,

R8. O2(g) → O2(aq) .

The saturation concentration of dissolved oxygen is calculated from Henry’s law:

CO∗ 2 = k(T, C)PO2 = k ⊗ (T )σ(C)PO2 (14)

where PO2 is the partial pressure of oxygen, and k(T, C) is the solubility of dissolved
oxygen at unit partial pressure [molal/atm]. k(T, C) may be written as the product of
the solubility of oxygen in pure water k ⊗ (T ), and a salting factor σ(C). σ(C) accounts
for the decrease in oxygen solubility due to the presence of salts in solution [6], but is
not considered in the model. The solubility of oxygen in pure water takes the form:

A + BT + CT 2 + DT ln T
 

k (T ) = exp (15)
RT

where A,B,C,D are empirical constants and R is the gas constant [6].
The rate of dissolved oxygen production is modeled as

rO2 = kL a(CO∗ 2 − CO2 ) (16)

where kL a is the gas-liquid mixing coefficient. Since oxygen is sparingly soluble in


water, its accumulation in solution is negligible. Hence a steady state approximation
is made that the net production of oxygen is zero. From the stoichiometry of oxygen
consumption in R4 and R5, we may write

kL a(CO∗ 2 − CO2 ) = 3r4 + r5 . (17)

The rate of acid consumption by gangue is modeled by a first order rate law in
the concentration of sulfuric acid,

r6 = kacid Cacid . (18)

The rate of jarosite precipitation is approximated by assuming a linear equilibrium


between free acid and ferric:
Cacid
kJ (T ) = (19)
CF e3+
where kJ (T ) is the jarosite precipitation equilibrium constant.
By solving Eq.(13) and Eq.(17) iteratively till the solution converges, we obtain
the oxygen concentration, rate of ferric production and rate of acid production. Eq.(1),
Eq.(18) and Eq.(19) are evaluated to get the rates of reactions R1-R3, R6 and R7
respectively. These rates are then used in the solute transport equation (discussed in
the next section) to solve for the new species concentrations, while rO2 calculated from
Eq.(16) is used in the gas transport equation discussed later on.

SOLUTE TRANSPORT

The heap is conceptually divided into columns (Fig.1). As copper heaps are
conventionally run under dripper emitters, it is assumed that the drippers trace out the
perimeter of identical cylindrical columns, so that one can model the heap by considering
a single cylindrical column. A column is divided into bulk flow channels, and equal-
length diffusion side branches (profile side pore diffusion model with fixed pore length
[1]). Intra-particle, inter-particle and inter-agglomerate porosities are combined to form
a single pore. Pore length is approximated by half the distance between drippers.
Transport of dissolved species is by advection along bulk channels, and by diffusion
through the stagnant pore. The diffusion driving force from the flowing liquid is assumed
to be greater than the concentration gradient between adjacent pores at all times.
Hence, there is no diffusion between individual side-branches [1].
The transportable species were mentioned in the previous section. Mass balance
for a species i along a diffusion pathway (stagnant pore) is described by the continuity
equation for mass transport,
   2 
1 Si ∂Ci ∂ Ci 1 ∂Ci
1+ = Di + + Γi . (20)
s Ci ∂t ∂r2 r ∂r

S is the adsorbed microbial concentration ([cell mole/kg dry ore]). The 1/s factor
rectifies the unit of S to [cell mole/kg stagnant liquid]. The adsorbed fraction S i /Ci
is taken as zero for the chemical species, and calculated from the Langmuir adsorption
model for the microbes:
Si Si,max KL,i
= . (21)
Ci 1 + KL,i Ĉi
Si,max is the maximum number of bacteria that can adsorb per kg of ore, and KL is the
adsorption equilibrium constant.
Γi , the component generation rate of species i, is given by
X
Γi = νij rj for chemical species (see Eq.(1))
j
= Eq.(10) for bacterial species . (22)

νij is the stoichiometric coefficient of species i in reaction j.


Dripper points

Drippers


 Diffusion
Channel

   
 
 
 
 
 
 
 
Advection
Channel
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
            
                

  
 
 
 
 
 
 
 
 
 
 

                           


Ore heap                           

ζ=0

Advection
Channel
Diffusion
Channel

Bulk flow

Inter−agglomerate diffusion ξ=0 ξ=1


Inter−particle diffusion
Intra−particle diffusion ζ=1

Figure 1: Conceptual heap model

By defining an effective pore length, R̃ = τ R (where τ is the tortuosity factor)


and a dimensionless pore length ξ = r/R̃, we write Eq.(20) in dimensionless form:
    2 
1 Si ∂Ci Di ∂ Ci 1 ∂Ci
1+ = + + Γi . (23)
s Ci ∂t R̃2 ∂ξ 2 ξ ∂ξ
At ξ = 0, we have the boundary condition
∂Ci
=0 (24)
∂ξ ξ=0
that is, no flow across the boundary.
At ξ = 1, the boundary coincides with the bulk channel. Therefore we consider
bulk transport of species at this point.
Transport of species in the bulk is modeled by the advection equation (per unit
heap volume)
∂Cf,i ∂Cf,i
f = −uf +Λ . (25)
∂t ∂z
f is the flowing liquid mass fraction and uf is the flowing liquid superficial velocity.
The source term Λ is derived as follows:
Integrating Eq.(23) along the domain volume
Z 1     Z 1  
1 Si ∂Ci 2 Di 1 ∂ ∂Ci
1+ − Γi dξ = ξ dξ 2 (26)
0 s Ci ∂t R̃2 0 ξ ∂ξ ∂ξ
where dξ 2 = 2ξdξ gives
Z 1     Z 1  
1 Si ∂Ci Di ∂Ci
2 1+ − Γi ξdξ = 2 d ξ
0 s Ci ∂t R̃2 0 ∂ξ
 
Di ∂Ci

= 2 . (27)
R̃2 ∂ξ ξ=1
Rewriting the LHS as an average gives, per unit stagnant fluid volume,
    
1 Si ∂Ci Di ∂Ci
1+ − Γi = 2 (28)
s Ci ∂t R̃2 ∂ξ ξ=1
or per unit heap volume,
    
1 Si ∂Ci Di ∂Ci
s 1+ − Γi = 2 s . (29)
s Ci ∂t R̃2 ∂ξ ξ=1
This is the contribution to Λ resulting from diffusion of species into the bulk channel.
Assuming no chemical reactions take place in the bulk, Eq.(25) becomes
 
∂Cf,i ∂Cf,i Di ∂Ci
f + uf =2 s . (30)
∂t ∂z R̃2 ∂ξ ξ=1
Defining a dimensionless heap depth ζ = z/Z gives the model advection equation per
unit flowing fluid volume as
 
∂Cf,i uf ∂Cf,i Di s ∂Ci
+ =2 . (31)
∂t f Z ∂ζ R̃2 f ∂ξ ξ=1
Defining a material derivative
 
D f Z D f Z ∂ uf ∂
= = +
Dζ uf Dt uf ∂t f Z ∂ζ
and the superficial velocity as uf = Gf /ρf where Gf is the flowing liquid mass flux
[kg/m2 /s] and ρf the density of the flowing liquid, we rewrite the boundary condition
at ξ = 1 as  
DCi Di s ρf Z ∂Ci
= 2 (32)
Dζ ξ=1 R̃2 Gf ∂ξ ξ=1
where Eq.(32) has boundary condition
Ci (ζ = 0, t) = C0 . (33)
HEAT TRANSPORT

The heat transport model is important in copper sulfide leaching because the
bacteria that catalyze the oxidation reactions are sensitive to temperature. Further-
more, the rates of key reactions in the heap are temperature dependent. A complete
derivation and exhaustive discussion of the heat transport model is available in [4].
Only an overview is given in this section.
Heat transport in the heap is via
• conduction through the solids and stagnant solution.
• convection by the leaching solution flowing downwards and the gaseous phase
(including water vapor) flowing upwards.
A general enthalpy balance is written as

[rate of heat accumulation] =[net rate of heat input by conduction]+


[net rate of heat input by advection]+
[net rate of heat generated by chemical reactions]

In dimensionless form:
1 ∂2T
 
∂T Ff (T ) Fv (T ) ∂T
= − − + Q̃ . (34)
∂t τc ∂ζ 2 τf τa ∂ζ

τc , τf and τa are conduction, liquid phase advection and gaseous phase advection
timescales respectively, given as

(ρcp )h Z 2 (ρcp )h Z (ρcp )h Z


τc = ; τf = ; τa = (35)
k Gf cpf Ga cpa

where cp is heat capacity, and the overbar represents an average.


Heat generation is calculated from reactions R4 and R5. For any sulfide oxidation
reaction involving molecular oxygen, the heat generated is in the order of 100KJ/mol
electrons transferred [13]. Hence from the stoichiometry of reactions R4 and R5, the
heat generation is
Q = 1200r4 + 400r5 (36)
Eq.(36) is integrated along the radial coordinate to obtain the average heat Q generated
along a diffusion pathway. The source term in Eq.(34) Q̃ is given by

Q
Q̃ = . (37)
(ρcp )h

Defining the following parameters,


Ga kg cpv λ
η= ; φ= ; β= ; κ= , (38)
Gf Gf cpa cpa
where kg is the mass transfer coefficient and λ the latent heat of vaporization, then

Ff (T ) = 1 − φ[ψ(T ζ=0 ) − ω∞ ψ(T∞ )] + η[ψ(T ) − ψ(T ζ=0 )] (39)
dψ(T )
Fv (T ) = 1 + βψ(T ) + κ (40)
dT
Ff (T ) incorporates the evaporation of the leaching solution at the heap surface . F v (T )
represents the degree to which the presence of water vapor in the gaseous phase enhances
heat flow.
The boundary conditions are:
1 ∂T X 1
− = q̃ + [T f − T ] (41)
τc ∂ζ ζ=0 τf ζ=0

1 ∂T 1 n o
− = [1 + βωa ψ(Ta )][T ζ=1 − Ta ] + κ[ψ(T ζ=1 ) − ωa ψ(Ta )] (42)

τc ∂ζ ζ=1 τa

Eq.(41) accounts for heat flux at the surface of the heap, and heat gain due to the
application of the leaching solution. Eq.(42) accounts for heat loss due to the saturation
(with water vapor) of air applied to the base. The flux term is written
P
X q
q̃ = (43)
(ρcp )h Z

where
X
q = qconvection + qevaporation + qradiation (44)

qconvection = h[T∞ − T ζ=0 ] Newton’s law of cooling (45)

qevaporation = −kg ρa [ψ(T ζ=0 ) − ω∞ ψ(T∞ )] Newton’s law of cooling (46)
qradiation = αs G(z) − ε F1→2 σ[T 4 4

ζ=0
− Tsky ] Stefan’s law (47)

Here G(z) is intensity of solar radiation for the zenith angle z, T∞ is the ambient
temperature, and Tsky is the effective blackbody sky temperature.

GAS (OXYGEN) TRANSPORT

The transport of oxygen through the heap is included through a 1-D model for
its partial pressure. The following assumptions are made [15] :
• The overall flow of dry air is not significantly diminished by the consumption
of oxygen (Boussinesq approximation).
• The diffusion of oxygen is negligible compared to advection.
• Steady state may be safely assumed (Net rate of oxygen input by advection
equals net rate of oxygen consumed).
The steady state equation is
 
∂ G g C O2
=  s rO2 ρh , (48)
∂z ρg

where CO2 is in [mol/m3 ]. Noting the Boussinesq approximation, the gas mass flux is
written as
Gg = Ga (1 + ψ) . (49)
Also from ideal gas law , the gas density and concentration are written as
P g Mg P O2
ρg = ; C O2 = . (50)
RT RT
Mg here is the molar mass of the gas. Substituting these in Eq.(48) gives
 
Ga ∂ 1 + ψ
P O2 =  s rO2 ρh . (51)
Pg ∂z Mg

By applying Boussinesq approximation again to write


1+ψ 1 ψ
= + , (52)
Mg Ma Mv

then differentiating the LHS of Eq.(51) and rearranging, we obtain

P g  s ρh rO2 ψ 0 ∂T
− PO
∂PO2 Ga Mv ∂z 2
= . (53)
∂z 1 ψ
+
Ma Mv
z
Expressing distance in dimensionless form (ζ = )
Z
Zρh Pg s rO2 ψ 0 ∂T
− PO
∂PO2 Ga Mv ∂ζ 2
= . (54)
∂ζ 1 ψ
+
Ma Mv
The boundary condition for Eq.(54) at the heap base is
!
M v
PO2 ζ=1 = 0.21 P∞ (55)
Mv + M a ψ ζ=1
MODEL APPLICATIONS

The heap bioleaching model derived above has been applied to case studies of
chalcocite, chalcopyrite and zinc sulfide leaching. The model parameters used in the
simulations are either measured for each case study (for instance the kinetic parame-
ters), or adapted from literature (for instance some microbial parameters).
In [14, 16], column experimental results are compared with model predictions
for chalcocite leaching. The model confirms the two-stage chalcocite leaching process
observed in experiments. Simulation results show that first-stage leaching progresses
in a narrow band which moves down the column, while second-stage leaching is more
homogeneous across the entire column depth. It is further deduced from the model that
the full scale heap chalcocite leaching is limited by the rate of acid supply to reaction
sites due to an increased effective diffusion path length.
In [13, 14], thermophilic leaching of chalcopyrite is investigated using the
geocoatTM process, in which the ore particle is spray-coated with ore concentrate
in order to maximize mineral exposure. Results from the model suggest that heating a
heap to the thermophilic range and maintaining it there is unproblematic, although this
requires the presence of another reactive sulfide such as pyrite to generate the initial
heat. Furthemore, model results show that temperatures over the heap are not uniform
even at steady state conditions. It also shows that dissolved oxygen levels calculated
from Eq.(48) are low throughout, suggesting that the rate of oxygen mass transfer from
gas to liquid phase is the overall rate limiting step.
Modelling work done on zinc sulfide [11, 12] suggests that the most important
parameters to zinc extraction rate are the heap height, irrigation rate, feed acid concen-
tration, and side branch length. Their significance arises because the heap is partially
limited by the supply and diffusion of sulfuric acid. Other important parameters are
the rate of oxygen mass transfer from gas to liquid, and the feed solution temperature.

CONCLUSIONS

A detailed derivation of a mathematical model for heap bioleaching is presented


in this paper. The application of the model to actual heap bioleaching scenarios has
helped to develop a deeper understanding of the complex interactions occurring in a
heap. In this sense, the model is a valuable tool for determining design parameters
for optimized heap operation. Most validation work thus far has been at laboratory
and pilot scale. Heap scale validation thus remains an issue. Also the model assumes a
simplified vertical solution flow channel, which in real life is more complex. Agglomerate
scale transport is modeled by 1-D diffusion through equal length diffusion channels.
Again this assumption simplifies complex agglomerate scale phenomena, and the pore
length is not easily defined. Modeling work is continuing to address these issues.
REFERENCES

[1] S. C. Bouffard and D. G. Dixon, “Investigative Study into the Hydrodynamics of


Heap Leaching Processes,” Metallurgical and Materials Transactions B, Vol. 32,
2001, 763–776.
[2] J. Brierley and C. Brierley, “Present and future commercial appli-
cations of bio-hydrometallurgy,” R. Amils and A. Ballester, Eds.,
International Biohydrometallurgy Symposium IBS’99, Elsevier, Amsterdam,
1999, Vol. 81.
[3] D. Dixon and J.Petersen, “Comprehensive Modelling Study of Chalcocite Col-
umn and Heap Bioleaching,” P. Riveros, D. Dixon, D. Dreisinger and J. Menacho,
Eds., Copper 2003-Hydrometallurgy of Copper (Book 2), CIM, Montreal, Canada,
2003, Vol. 6, 493–516.
[4] D. G. Dixon, “Analysis of Heat Conservation during Copper Sulphide Heap Leach-
ing,” Hydrometallurgy, Vol. 58, 2000, 27–41.
[5] D. G. Dixon, “Heap Leach Modelling-The Current State of the Art,” C. Young,
A. Alfantazi, C. Anderson, D. Dreisinger, B. Harris and A. James, Eds.,
Hydrometallurgy 2003, TMS, Warrendale PA, 2003, Vol. 1, 289–314.
[6] D. G. Dixon and D. B. Dreisinger, “Hydrometallurgical Process Modelling for De-
sign and Analysis Part II: Leaching Kinetics and Associated Phenomena,” P. R.
Taylor, D. Chandra and R.G.Bautista, Eds., EPD Congress 2002, TMS, Pennsyl-
vania,USA, 2002, 687–708.
[7] D. G. Dixon and J. Hendrix, “Theoretical Basis for Variable Order Assumption in
the Kinetics of Leaching Discrete Grains,” AICHE J., Vol. 39, 1993, 904–907.
[8] D. G. Dixon and J. Petersen, “Modelling the Dynamics of Heap Leaching for
Process Improvement and Innovation,” , Apr. 2004, Invited key-note lecture for
Hydro-Sulfides Conference, Santiago, Chile.
[9] O. Levenspiel, Chemical Reaction Engineering, John Wiley & Sons, Inc., 1972, 2nd
Ed.
[10] H. Lizama, J. Harlamovs, S. Belanger and S. Brienne, “The Teck Cominco
HydrozinT M Process,” C. Young, A. Alfantazi, C. Anderson, D. Dreisinger, B. Har-
ris and A. James, Eds., Hydrometallurgy 2003, TMS, Warrendale PA, 2003, Vol. 2,
1503–1516.
[11] J. Petersen and D. G. Dixon, “Modelling Zinc Heap Bioleaching,” In review.
[12] J. Petersen and D. G. Dixon, “Systematic Modelling of Heap Leach Process for
Optimization and Design,” P. R. Taylor, D. Chandra and R.G.Bautista, Eds.,
EPD Congress 2002, TMS, Pennsylvania,USA, 2002, 757–771.
[13] J. Petersen and D. G. Dixon, “Thermophilic Heap Leaching of a Chalcopyrite
Concentrate,” Minerals Engineering, Vol. 15, 2002, 777–785.

[14] J. Petersen and D. G. Dixon, “A Modelling Study of the Dynamics of Sul-


phide Heap Leach Processes with a View to Improved and Novel Applications,”
XXII International Mineral Processing Congress, SAIMM, Marshalltown, South
Africa, 2003, Vol. 3, 1231–1240.
c
[15] J. Petersen and D. G. Dixon, “HeapSim Heap Bioleach Simulation Package,
Version 2.01 User Reference Manual,” , Oct. 2003.

[16] J. Petersen and D. G. Dixon, “The Dynamics of Chalcocite Heap BioLeaching,”


C. Young, A. Alfantazi, C. Anderson, D. Dreisinger, B. Harris and A. James, Eds.,
Hydrometallurgy 2003, TMS, Warrendale PA, 2003, Vol. 1, 351–264.

[17] G. Rossi, Biohydrometallurgy, McGraw-Hill, Hamburg, 1990.

NOMENCLATURE

Greek Letters
αs Heap surface solar absorptivity
 Liquid mass fraction kg water/kg dry ore
Γ Specie generation rate molal/s
λ Latent heat of vaporization kJ/kg
ν Stoichiometric factor
ω Relative humidity
ψ Saturation humidity
ρ Density kg/m3
σ Stefan-Boltzmann constant 5.67 x 10−8 W/m2 /K4
τ Tortuosity factor
ε Heap surface grey body emissivity
ξ Dimensionless diffusion pathway length
ζ Dimensionless heap height
Roman Letters
k Average heap thermal conductivity W/m/K
R̃ Effective diffusion pathway length (τ R) m
C Concentration molal
cp Heat capacity J/kg/K
D Diffusivity m2 /s
E Activation energy J/mol
F1→2 Heap-to-sky view factor
G Mass flux kg/m2 /s
g Grade mol/kg dry ore
h Heap surface heat transfer coefficient W/m2 /K
kA Ferric mass transfer parameter molal
kB Ferric reduction parameter molal
ke Endogenous decay factor
kg Mass transfer coefficient , see equation (38) kg/s
kg Microbial specific growth rate , see equation (8) s−1
KL Langmuir adsorption constant molal−1
kL a Gas-liquid mixing coefficient s−1
12
km Specific maintenance rate mol/10 cells/hr
kacid Gange-acid rate constant m3 /mol/s
kJ Jarosite precipitation equilibrium constant
KI,b Growth inhibition constant for species b cell molal
KL,b Monod constant for species b molal
m Exponent in electrochemical law , see equation (4)
N Microbial population 1012 cells/kg dry ore
P Pressure atm
Q Heat generation rate from chemical reactions W/m3
q Heat flux W/m2
R Diffusion pathway length m
R Universal gas constant 8.314 J/mol/K
r Formula rate molal/s
S Adsorbed microbial concentration 1012 cells/kg dry ore
T Temperature K or ◦C
uf Flowing liquid superficial velocity m/s
X Mineral conversion
y Specific cell yield 1012 cells/mol
Z Heap height m
Subscripts
∞ Ambient condition
a Dry air
f Flowing liquid
h Heap
s Stagnant liquid
v Water vapor

View publication stats

You might also like