SIMS 49

MODELING OF AN INDUSTRIAL COPPER LEACHING

AND ELECTROWINNING PROCESS, WITH VALIDATION
AGAINST EXPERIMENTAL DATA
Bernt Lie1 , Tor Anders Hauge2
1
Telemark University College, P.O. Box 203, 3901 Porsgrunn, Norway
2
Xstrata Nikkelverk, Kristiansand, Norway
[email protected], [email protected]

Abstract The raw material comes from a granulated matte pro-

duced by Xstrata's Sudbury smelter in Canada, BCL's
The Cu process at Xstrata Nikkelverk, Kristiansand, smelter in Botswana, as well as custom feed from other
Norway is considered. This process is an important part sources. The granulated matte is crushed to a very ne
of the nickel re nery. With the current operation, the powder (Calcine) and then transported to the chlorine
level of variation in measured quality variables is larger leaching plant. The Calcine contains mostly CuO, but
then desired, and it is of interest to reduce this variation also impurities of Ni, Co, Fe and other metals; this Cal-
through improved control. The process is characterized cine is leached in a sulfuric acid mixture, the mixture is
by time delays, multivariable behavior, and nonlinear- then pumped through lters where nondissolved metal
ities, and it is thus a candidate for model based con- oxides are removed, and then through scrap columns
trol. A new dynamic model with 39 states is formulated where other impurities (Ag, Te, and B) are cemented
based on steady state total mass balances and dynamic and removed. Next, the slurry, or electrolyte, is pumped
specie balances, and includes reaction kinetics in the to an electrowinning section where electrolysis is used
leaching sections and use of an electrochemical descrip- to produce Cu metal at the cathode and H2 SO4 and O2
tion of the electrowinning section. Utilizing the simpli- at the anode. The resulting anolyte from the electrowin-
fying assumptions and available measurements, steady ning is then pumped back to the Cu leaching section.
states as well as the currency ef ciency of the electrol-
Important quality measurements throughout the Cu
ysis process are computed by solving linear equations
process are the concentrationS of Cu and H2 SO4 (free
for the electrowinning section. For the leaching sec-
and so-called available H2 SO4 ). The most important
tions, steady state balances with a second order reac-
inputs to manipulate these quantities are the mass ow
tion rate constitute overdetermined multivariable poly-
of Calcine to the leaching section, the volumetric ow
nomial equations; solving a corresponding least squares
of fresh H2 SO4 to the leaching section, and the volu-
problem yields the steady states and the reaction rate
metric ow of recirculated anolyte to the leaching sec-
constant with a small error norm. Additional measure-
tion. The quality measurements represent a challenge to
ments from the process validate the steady states of the
control due to time delays, strong couplings, and non-
model. Using the steady states as initial values, steps
linearities. In order to reduce the variation of the qual-
in the 3 inputs indicate that the model is nonlinear and
ity measurements, it is of interest to consider model
strongly coupled. The simulations con rm suggestions
based multivariable controllers such as Model predic-
for signs of steady state gains as well as typical poles.
tive control (MPC). As the model is intended for con-
Keywords: Copper production, Leaching, Electrowin- trol purposes, this will be re ected in the structure and
ning, Dynamic model choice of simplifying assumptions. A model intended
for model based control, should be validated against
process knowledge and measurements to assess its qual-
ity.
1 Introduction

1.1 Background 1.2 Previous work

In this paper, we consider a chlorine leaching and elec- The process and its operation is described in [1] – [4].
trowinning process which is an important part of the Based on these, a M.Sc. project [5] was carried out with
nickel re nery of Xstrata Nikkelverk in Kristiansand, some initial work on dynamic models of the system. In
Norway. [6], an attempt is made to develop a linear empirical
www.scansims.org Proceedings of SIMS 2008 131
Oslo, Norway, October 7-8, 2008
SIMS 49

model of the system, and some useful ideas of expected analyzed three times per week and thus may represent a
signs in the transfer matrix are given. Still, no thor- signi cant unknown disturbance between the samples.
ough dynamic mechanistic model of the system exists. In the process, all tanks are assumed to be perfectly
General methods for formulating balance equations are mixed. The liquid content in all tanks is assumed to
given in e.g. [7] and [8], systematic methods to describe be constant, and the liquid density is assumed to be
systems with chemical reactions is given in e.g. [8], ba- the same everywhere. The scrap columns are assumed
sic reaction kinetics is discussed in e.g. [9], while e.g. to perfectly cement unwanted metals, and lters are as-
[10] gives a thorough description of modeling of elec- sumed to perfectly remove unwanted material in the pu-
trochemical systems. ri cation section.
The Calcine needs to be slurri ed in order to ease
1.3 Structure of paper the transport and enable the leaching of copper. This
is done by mixing the Calcine with recycled liquid
In section 2, a description of the process is given, in- from the electrowinning section, so-called anolyte: the
cluding a discussion of the ow sheet, how the process anolyte is assumed to consist of copper sulfate CuSO4
operates, and a discussion of inputs and outputs and and sulfuric acid H2 SO4 dissolved in water. The Cal-
(1)
their expected correlations. In section 3, a mechanis- cine is slurri ed in two tanks, with volumes Vs and
(2)
tic model is developed. In section 4 , the model is an- Vs , respectively, where j 2 fCuO; CuSO4 ; H2 SO4 g.
alyzed both from a steady state perspective and from a With perfect mixing, the mass concentrations of the
(k) (k) (k)
dynamic perspective. Finally, in section 5 a discussion species j are s;j = ms;j =Vs . Assuming constant
is given and some conclusions are drawn, together with density in the liquid, the volumetric ow out of both
(k)
indications for future work. of the volumes Vs are V_ s2` = V_ e2s .
The leaching section consists of 5 tanks in series, where
(1)
the rst tank with constant liquid volume V` receives
2 Process description the slurri ed mixture in ow V_ s2` and an acid feed V_ a
with a mass concentration a;H2 SO4 of sulfuric acid. In
2.1 Flow sheet the second tank in the leaching section, another acid
feed containing hydrogen chloride (HCl) is added (not
The Xstrata copper production plant consists of 4 sec- shown in g. 1); the purpose is to enable the removal of
tions, g. 1 (all gures and tables are given after the additional metals in the slurry. Since we do not consider
References section): (i) the slurri cation section where these other metals (perfectly removed in the puri cation
the particulate raw material Calcine containing metal section), and since this ow is negligible, we simply ne-
oxides such as CuO is slurri ed by mixing it with re- glect this second acid feed in our description. In the fth
(5)
cycled anolyte, (ii) the leaching section where sulfuric leaching tank, V` , spilled water is added with a rate
acid H2 SO4 is added to the slurry in order to leach the V_ w2` ; the purpose is to return water/liquid that is spilled
solid copper oxide into copper sulfate CuSO4 ; copper throughout the process, and thus to enable keeping the
sulfate is a salt of the sulfuric acid, (iii) the puri cationliquid levels constant .
section where various components are removed from In the slurri cation and leaching tanks, copper oxide
the slurry: remaining copper oxide in the rst lter, undergoes the following irreversible stoichiometric re-
other metals are cemented in scrap columns and ltered action
out in the second lter, and (iv) the electrowinning sec-
k
tion where copper sulfate is electrolyzed to release solid CuO + H2 SO4 ! CuSO4 + H2 O;
copper at the cathode and oxygen gas at the anode.
since the temperature of the slurry is more or less con-
stant, k is constant. With specie vector given by
2.2 Process operation and assumptions T
= CuO H2 SO4 CuSO4 H2 O ,
2.2.1 Slurri cation and leaching sections the stoichiometric matrix (vector) of the reaction is
The raw material Calcine is transported into the slurri - = 1 1 1 1
cation tanks using several screw conveyors. The veloc-
ity of the rst conveyor determines the mass ow rate and the reaction can be written as
m_ c . The mass ow rate is calculated at a point close to
0! .
the slurri cation tanks using measured weight and con-
veyor velocity. Thus, there is a substantial time delay The reaction is assumed to be elementary, and the re-
between the point where the mass ow rate is deter- sulting molar reaction kinetics is thus
mined and the point where it is calculated. The mass
fraction xc;Cu of copper in the form of copper oxide, is = kcCuO cH2 SO4 .
www.scansims.org Proceedings of SIMS 2008 132
Oslo, Norway, October 7-8, 2008
SIMS 49

Since the remaining of the description will be mass tanks is known as anolyte. The outlets from the elec-
based, we use that trolysis tanks are combined in a manifold, and is then
sent to a mixing tank with volume Vem before recycling
1
cj = the anolyte at rate V_ e2s to the slurri cation unit. There
Mj j is also a certain bleed V_ em2bl from the mixing tank to
avoid the build-up of inerts/impurities. A ow V_ w2em
where Mj is the molar mass and j is the mass concen-
of make-up water is added to the mixing tank to replace
tration to arrive at
spilled water. Because the uid out of the puri cation
k section and into the electrowinning section has an un-
= . (1)
MCuO MH2 SO4 CuO H2 SO4 suitably high concentration of copper sulfate, this uid
is diluted by a recycling V_ em2d of anolyte from the post-
The vector of reaction rates for the species, r~ (mol electrolysis mixing tank of volume Vem into the dilution
based), is related to by tank of volume Ved .
T The electrolysis can be described as follows: although
r~ = .
the content of the liquid has been described as copper
We don't really care about the water content, so let us sulfate CuSO4 and sulfuric acid H2 SO4 , these mole-
denote the rst three elements of by . Similarly, let cules will in fact exist partially disassociated, i.e. in
denote the rst three elements of and r~ denote the some equilibrium
three rst elements of r~ . Then
CuSO4 Cu2+ + SO24
T
r~ = v . H2 SO4 2H+ + SO24 .
To get rates in mass basis, r^ , we need to multiply the
rate of each specie by the molar mass Mj . Let In addition, water is partially disassociated:

M = diag MCuO MH2 SO4 MCuSO4 . H2 O H+ + OH .

Then
This mixture is an electrical conductor, and by injecting
r^ = M v T . (2)
an electric currency I through the mixture from the an-
ode to the cathode, the following happens: (i) electrons
2.2.2 Puri cation section e ow into the mixture at the cathode; electric cur-
rency is de ned as positive in the opposite direction of
In the puri cation section, all remaining copper oxide is the actual direction of electron ow, (ii) at the cathode,
assumed to be removed in the rst lter ( lter presses). copper is neutralized:
Thus, after these lter presses, there is no more cop-
per oxide, and the leaching reaction is nonexistent for Cu2+ + 2e ! Cu(s) ,
the remaining of the puri cation and electrowinning
processes. Thus, the leaching reaction only takes place
(k) (iii) at the anode, a similar amount of electrons are
in the slurri cation volumes Vs and the leaching vol-
(k) pulled out to uphold the currency I:
umes V` : these are the only volumes where the con-
centration of copper oxide is nonzero. The remaining 1 (g)
puri cation section consists of 3 buffer tanks with vol- SO24 + H2 O ! H2 SO4 + O2 + 2e .
(k) 2
umes Vpb before and after a group of scrap columns
and lters which are used to remove the remaining un- Since the number of copper molecules produced in a
desired metals. The scrap columns are essentially lled time interval dt, NCu , is equal to half the number of
with copper that helps in cementing these unwanted electrons involved, Ne , we have NCu = 1 Ne
z
metals for removal in a subsequent lter bank (indi- where zCu = +2 is the valency of copper. The Cu produc-
cated by m _ MeO;o in the ow sheet). Although the scrap tion rate of copper (molecules produced per time unit)
columns have considerable volume, the void fractions thus satis es N_ Cu;o = 1 N_ e . The ow of electrons
(k) zCu
"ps are tiny so that these columns have fast dynamics.
N_ e can be converted to molar ow of electrons n_ e
by multiplying with Avogadro's number NA , and mo-
2.2.3 Electrowinning section lar ow of electrons can be converted to charge ow Q_
with units Coulomb/second by multiplying with Fara-
In the electrowinning section, the main unit is the elec- day's constant C, Q_ = C n_ e ; clearly Q_ = I, i.e. the
trolysis/electrowinning tank of volume Vew ; in reality, currency in Ampere. Finally, we nd the mass based
this consists of a large number of small electrolysis production rate of copper, i.e. the amount of copper
tanks in parallel. The uid at the exit of the electrolysis pulled out of the electrowinning tank, by using the mo-
www.scansims.org Proceedings of SIMS 2008 133
Oslo, Norway, October 7-8, 2008
SIMS 49

lar mass MCu of copper: 2.3 System inputs and outputs

m
_ Cu;o = MCu n_ Cu;o = MCu NA N_ Cu;o
The purpose of the process is to produce copper. The
1 _ most direct way to do this, is to control the currency I
= MCu NA N
zCu e in the electrolysis process; this is what is done. How-
1 1 MCu ever, electrolyzing copper in the tanks of volume Vew
= MCu CNA N_ = I.
zCu C | {z e } zCu C will also remove copper sulfate; if no action is taken,
_
Q=I the content of the system may end up in imbalance. The
control problem thus is to ensure a suitable chemical
Here it should be noted that if I is in units A = C= s,
composition in the system. Three measurements are
then m
_ Cu;o is in units g= s. With the more convenient
used to measure the chemical composition of the sys-
time unit of hours, then a unit conversion must be in-
tem: y1 is the mass concentration H2 SO4 of sulfuric
cluded.
acid into the electrowinning section (so-called free sul-
In practice, the injected currency I will have some leak- furic acid), y2 is the mass concentration CujCuSO4 of
age/stray currency, i.e. not all of it will go through the copper in the form of copper sulfate at the outlet of the
mixture. It is thus in practice necessary to include a electrolysis tanks, while y3 is the mass concentration of
currency ef ciency factor to get the nal expression so-called available sulfuric acid at the outlet of the elec-
MCu trolysis tanks; available sulfuric acid is the combined
m
_ Cu;o = I. concentration of sulfates, assumed to be sulfuric acid,
zCu C
denoted H2 SO4 .
By comparing the anode reaction, we see that there is For y1 we thus have
a simultaneous production of H2 SO4 in the electrowin-
ning tank of (3)
y1 = pb;H2 SO4 . (5)
MH2 SO4
m
_ H2 SO4 ;g = _ Cu;o .
m
MCu y2 and y3 takes measurements from the anolyte after the
manifold at the outlet from the electrolysis tanks. The
In reality, the electrolysis takes place in 8 groups of measurement technique used to measure y2 measures
electrolysis units, where the units in each group are in the copper part ew;CujCuSO of the copper sulfate con-
4
series wrt. the currency (but in parallel wrt. electrolyte tent ew;CuSO , hence
4
ow). Thus
8
X 8 nj MCu
MCu MCu X X k k y2 = ew;CujCuSO4 = ew;CuSO4 . (6)
m
_ Cu;o = j Ij = j Ij . MCuSO4
j=1
zCu C zCu C j=1
k=1
Finally, y3 measures the sulfate content and assumes
In groups 1–3, the currency is the same, Ie , in groups
that it is pure sulfuric acid, ew;H2 SO4 . Thus
4–6 the currency is the same, Iw , while in group 7 the
currency is Iw +I7 and in group 8 the currency is Ie +I8 .
The number of tanks in each group is nj . The ef cien- y3 = ew;H2 SO4 = ew;H2 SO4 + ew;H2 SO4 jCuSO4
cies j are not known in detail, and it makes more sense
to operate with an overall currency ef ciency or
nj
8 X
X MH2 SO4
ew;CuSO4 .
k k y3 = = +
j Ij = I ew;H2 SO4 ew;H2 SO4
MCuSO4
j=1 k=1
(7)
which means that To control these 3 outputs, the following three control
MCu variables are used: u1 is the mass feed rate of Calcine,
m
_ Cu;o = I, (3)
zCu C
_ c,
u1 = m (8)
where

I = (n1 + n2 + n3 + n8 ) Ie u2 is the volumetric recirculation of anolyte,

+ (n4 + n5 + n6 + n7 ) Iw
u2 = V_ e2s , (9)
+n7 I7 + n8 I8 . (4)

The overall ef ciency is less than unity, not only due while u3 is the volumetric feed rate of acid,
to leakage currency, but also due to that at all times,
some electrolysis tanks are out of production. u3 = V_ a . (10)

www.scansims.org Proceedings of SIMS 2008 134

Oslo, Norway, October 7-8, 2008
SIMS 49

2.4 Operational experience mass of specie j, Vi is the volume of tank i, and r^i;j is
the reaction rate in tank i of specie j.
Typical values for manipulated inputs u, disturbances The result is 21 ordinary differential equation, one for
w, and measurements y are given in Tab. 1. each of the 3 species in the 7 tanks.
From operation of the system, some additional values
known from experience are listen in Tab. 2.
3.3 Puri cation section
Less systematic experience is available for the dynam-
ics of the system, but Tab. 3 indicates the sign of the The specie balances for the 6 volumes in the puri ca-
transfer matrix, as well as whether the system is as- tion section (3 buffer tanks and 3 scrap columns) are
sumed to have integrators using the notation s j T similar to the balances for the slurri cation and leach-
where s is the steady state sign and T is the T indicates ing sections, except that in this section we have only two
the time constant. species, j 2 fCuSO4 ; H2 SO4 g and there is no reaction
rate, r^i;j 0.
This adds 12 ordinary differential equations to the total
3 Process model model, one for each of the 2 species in the 6 tanks.

The following dynamic models can now be formulated:

3.4 Electrowinning section
3.1 Overall mass balances The specie balances for the dilution tank and the mix-
ing tank of the electrowinning section is similar to the
Using the total mass balances under the assumption of balances for the puri cation section. For the electroly-
constant liquid density and constant liquid level in each sis/electrowinning tank, we have
volume, leads to the following relationships:
d ew;H2 SO4 1
_ _ = V_ ed2w
Ve2s = Vs2` dt Vew ed;H2 SO4
(1)
V_ s2` + V_ a = V_ `;o _
ew;H SO Vew2m + m _ ew;H SO ;g 4
2 4 2
(1)
V_ `;o + V_ w2` = V_ `2p d ew;CuSO4 1
= _
ed;CuSO4 Ved2w
V_ `2p = V_ p2e dt Vew
V_ p2e + V_ em2d = V_ ed2w + V_ ed2m _
ew;CuSO4 Vew2m _ ew;CuSO4 ;o .
m
V_ ed2m + V_ ew2m + V_ w2em = V_ em2d + V_ em2bl + V_ e2s
This adds 6 ordinary differential equations to the total
V_ ed2w = V_ ew2m + V_ vap .
model, one for each of the 2 species in the 3 tanks.
We thus have 14 variable volumetric ow rates and 7
steady state mass balances to relate them, which means
that we must specify 7 of the variables. Clearly, two 4 Simulation and validation
of these are the manipulatable volumetric ow rates
u2 = V_ e2s and u3 = V_ a . In addition, we will specify the 4.1 Steady state model
following 5 ow rates: V_ w2` , V_ ed2m , V_ em2d , V_ vap , and
V_ em2bl ; these can be considered as disturbances. The re- A steady state analysis will give useful information
maining 7 volumetric ows can then be computed from about the process when combined with known operat-
the steady state mass balances. ing conditions. It turns out that steady state analysis will
give suf cient information both to nd the unknown
electrolysis ef ciency and the leaching reaction con-
3.2 Slurri cation and leaching sections stant k, as well as the steady states of the system. In
addition to the operating conditions of Tab. 1, we need
These sections involve the leaching reaction. A typical the parameters in Tab. 4.
specie balance has the form
By formulating the steady state specie balances for the
d i;j 1 electrowinning section as well as the steady state total
= _ i;j;i V_ i;o i;j + Mj r^i;j Vi
m mass balances, and combining these with the operating
dt Vi
conditions and parameters, we get a number of linear
where i;j is the mass concentration of specie j 2 equations which enables us to nd the steady state mass
fCuO; CuSO4 ; H2 SO4 g in each of the 7 tanks of the concentrations for the electrowinning system, as well as
slurri cation and leaching sections. m _ i;j;i is the total the mass ow rate of copper produced. Combining the
input mass ow of specie j to tank i, while V_ i;o is the mass ow rate of copper with the known currency en-
total volumetric ow out of tank i. Mj is the molar ables us to nd the currency ef ciency = 0:77779.
www.scansims.org Proceedings of SIMS 2008 135
Oslo, Norway, October 7-8, 2008
SIMS 49

For the slurri cation and leaching sections, the ele- In conclusion, the responses of the developed model
ment balances (sulphur and copper) are reaction invari- are quite similar to what was expected, and the model
ant. This can be used to compute the exit concen- is a valuable rst step in developing a control relevant
(5)
tration of copper oxide, `;CuO and the feed concen- model for the copper production process. No similar
tration of sulfuric acid, a;SjH2 SO4 . Next, formulat- model is known from the literature.
ing the 21 steady state specie balances yields 21 equa-Planned future work with the modeling of the copper
tions in 19 unknowns, the least squares method is used production plant includes further validation based on
to determine the remaining 19 unknowns uniquely, in- transient measurement, investigation of parameter sen-
cluding the reaction rate constant which is found to besitivity (e.g. the void fractions of the scrap columns are
3
k = 2:9637 moll g h . uncertain), and a study on the possibility to reduce the
The resulting steady states are given in Tab. 5. These order of the model. Next, implementation of the model
steady state values should be compared to the experi- with a state estimator is of interest, perhaps with online
ence values in Tab. 2, which indicates that the steady parameter estimation. Finally, it is of interest to study
state model is relatively accurate. how the model can be used in model based controllers.

4.2 Dynamic model 6 References

In addition to the steady states of Tab. 5 and parameters
of Tab. 4, we need the parameters of Tab. 6 (the void [1] T.A. Hauge, Cu process introduction, Internal re-
fractions are “guesstimates”). Fig. 2 show the results of port, Xstrata Nickel, Kristiansand, Norway, 2007
a 50% increase (solid line) and a 50% decrease (dash- [2] T.A. Hauge, Process description: Modeling and
dot line) in u1 (m _ c ) after 5 h, as well as the results of a control of a Cu plant, Internal report, Xstrata
constant input u1 (dotted). Fig. 3 show similar results Nickel, Kristiansand, Norway, 2007.
for u2 (V_ e2s ), while Fig. 4 show similar results for u3
(V_ a ): [3] E.O. Stenholt, H. Zachariasen, and J.H. Lund, “Fal-
conbridge chlorine leach process”, Transactions
of the Institution of Mining and Metallurgy, Sec-
5 Discussion and conclusions tion C, 5, p. C10–16.

[4] E.O. Stenholt, O.M. Dotterud, E.E. Henriksen, P.O.

A dynamic model of the Xstrata copper production
Ramsdal, F. Stålesen, and E. Thune, “Develop-
process has been developed, based on specie balances
ment and plant practice of the Falconbridge chlo-
and total mass balances. Emphasis has been put on
rine leach process”, CIM Bulletin, vol. 94, no.
making the model consistent so that there is no arti -
1051, p. 101–104, 2001.
cial accumulation of mass in the system, and so that
only available measurements/knowledge is used. Avail- [5] Y. Li, T. Xu, and A. Bedelbayev, Modeling and con-
able measurements are combined with the steady state trol of a Cu plant, Project report, Telemark Uni-
model, and from this the unknown parameters (currency versity College, Norway, 2007.
ef ciency and reaction rate constant k) as well as the
steady state values are computed in a least squares fash- [6] M. Komperød, Topics in Data Preprocessing and
ion; the computed steady state values are useful as ini- Model Realization, with Real-Life Applications
tial values in simulations. from the Process Industry, M.Sc. thesis, Tele-
mark University College, Norway, 2008.
The additionally measured values of Tab. 2 when com-
pared to the computed steady states of Tab. 5 indicate [7] R.B. Bird, W.E. Stewart, E.N. Lightfoot, Trans-
that the model is relatively good in steady state. That port Phenomena, second ed., John Wiley & Sons,
the steady states reported in Tab. 5 are the true steady New York, 2002.
states, is validated by the steadiness of the dotted lines
in Figs. 2 – 4. [8] R. Aris, Mathematical Modeling. A Chemical En-
Also, the suggested signs in Tab. 3 are con rmed by the gineer's Perspective, Academic Press, San Diego,
dynamic simulations. Furthermore, by comparing the 1999.
responses of positive and negative input steps, the sim- [9] O. Levenspiel, Chemical Reaction Engineering,
ulations show that the model is nonlinear. Finally, al- Second ed., John Wiley & Sons, New York, 1972.
though the dynamic simulations does not show true in-
tegration of the measured concentrations, the plots sug- [10] J. Newman and K.E. Thomas-Alyea, Electro-
gests some extremely slow modes which are approach- chemical Systems, third ed., John Wiley & Sons,
ing integration. Hoboken, NJ, 2004.

www.scansims.org Proceedings of SIMS 2008 136

Oslo, Norway, October 7-8, 2008
SIMS 49

Table 1: Known operating conditions for measurements yj , manipulated inputs uj , and disturbances wj .
Variable Value
(3)
y1 = pb;H2 SO4 26 g= l
y2 = ew;CujCuSO4 60 g= l
y3 = ew;H2 SO4 183 g= l
u1 = m _c 733 104 g= h
u2 = V_ e2s 82 103 l= h
u3 = V_ a 500 l= h
_
w1 = Vw2` 1700 l= h
w2 = V_ ed2m 20 103 l= h
_
w3 = Vem2d 330 103 l= h
w4 = V_ vap 500 l= h
w5 = V_ em2bl 5100 l= h
w6 = xc;Cu 0:63 gCu= gmass
w7 = Ie 9:34 103 A
w8 = Iw 8:82 103 A
w9 = I7 7:74 103 A
w10 = I8 3:38 103 A

Table 2: Additional values of process variables, known from experience.

Variable Value
a;H2 SO4 1834 g= l
(3)
pb;CujCuSO4 100 g= l
ed;CujCuSO4 70 g= l
em;H2 SO4 90:4 g= l

Table 3: Dynamic characteristics of transfer matrix, given as “sign of steady state gain | dynamics”. Dynamics
indicated by “I” denotes whether the system is expected to have an “integrator” (or possibly: slow time constant).
A question mark indicates uncertainty. A zero indicates zero gain.
u1 u2 u3
y1 j I + j? + j I
y2 + j I 0 0
y3 0 0 +jI

www.scansims.org Proceedings of SIMS 2008 137

Oslo, Norway, October 7-8, 2008
SIMS 49

Leaching

Purification
Slurrification

Electrowinning

Figure 1: Detailed principle ow sheet for the industrial copper leaching and electrowinning process.

www.scansims.org Proceedings of SIMS 2008 138

Oslo, Norway, October 7-8, 2008
SIMS 49

Table 4: Constants and parameters for the model.

Variable Value
n = (n1 ; : : : ; n8 ) (76; 76; 52; 84; 84; 30; 8; 10)
MH 1 g= mol
MO 16 g= mol
MCu 63:5 g= mol
MS 32 g= mol
zCu 2
C 9:65 104 C mol 1

Table 5: Steady states computed from steady state models.

(1) (1) (1)
s;CuO = 58:6158 g= l s;CuSO4 = 174:4265 g= l s;H2 SO4 = 74:3260 g= l
(2) (2) (2)
s;CuO = 53:2891 g= l s;CuSO4 = 185:1135 g= l s;H2 SO4 = 67:7597 g= l
(1) (1) (1)
`;CuO = 34:0256 g= l `;CuSO4 = 221:9917 g= l `;H2 SO4 = 55:2208 g= l
(2) (2) (2)
`;CuO = 23:8008 g= l `;CuSO4 = 242:5056 g= l `;H2 SO4 = 42:6167 g= l
(3) (3) (3)
`;CuO = 17:5975 g= l `;CuSO4 = 254:9513 g= l `;H2 SO4 = 34:9698 g= l
(4) (4) (4)
`;CuO = 13:5176 g= l `;CuSO4 = 263:1366 g= l `;H2 SO4 = 29:9405 g= l
(5) (5) (5)
`;CuO = 10:538 g= l `;CuSO4 = 263:25 g= l `;H2 SO4 = 26 g= l
(1) (1) (1)
pb;CuO = 0 g= l pb;CuSO4 = 263:25 g= l pb;H2 SO4 = 26 g= l
.. .. ..
. . .
(3) (3) (3)
pb;CuO = 0 g= l pb;CuSO4 = 263:25 g= l pb;H2 SO4 = 26 g= l
ed;CuO = 0 g= l ed;CuSO4 = 173:48 g= l ed;H2 SO4 = 76:179 g= l
ew;CuO = 0 g= l ew;CuSO4 = 150:71 g= l ew;H2 SO4 = 90:402 g= l
em;CuO = 0 g= l em;CuSO4 = 150:57 g= l em;H2 SO4 = 88:9827 g= l

Table 6: More model parameters.

Variable Value Variable Value
(1) (1)
Vs 7:4 m3 "ps 0:1
(2) 3 (2)
Vs 4m "ps 0:1
(k) (3)
V` 27:5 m3 "ps 0:1
(1) 3
Vpb 15 m Ved 10 m3
(2)
Vpb 15 m3 Vew 430 5 m3
(3) 3
Vpb 45:3 m Vem 75 m3
(1) l3
Vps 35 m3 k 6:3267 104 mol g h
(2)
Vps 40 m3 0:77779
(3)
Vps 42 m3 mix 1 kg= l

www.scansims.org Proceedings of SIMS 2008 139

Oslo, Norway, October 7-8, 2008
 1
SIMS 49

1
100

50

y1
0
0 5 10 15 20 25 30 35 40 45 50
80

60

y2
40

20
0 5 10 15 20 25 30 35 40 45 50
183

183
y3

1
183

183
0 5 10 15 20 25 30 35 40 45 50
time t [hr]

(3)

1
Figure 2: Responses in outputs y1 = pb;H2 SO4 , y2 = ew;CujCuSO4 and y3 = ew;H2 SO4 upon a 50% step increase
(solid line) and decrease (dash-dot line) in u1 = m
_ c after 5 h.

1
60

40
y1

20

0
0 5 10 15 20 25 30 35 40 45 50
70

60
y2

50

40
0 5 10 15 20 25 30 35 40 45 50
183.5

183
y3

1
182.5

182
0 5 10 15 20 25 30 35 40 45 50
time t [hr]

(3)
Figure 3: Responses in outputs y1 = pb;H2 SO4 ,
y2 = ew;CujCuSO4 and y3 = upon a 50% step increase
1
ew;H2 SO4
(solid line) and decrease (dash-dot line) in u2 = V_ e2s after 5 h. 1

35
1
30
y1

25

20
0 5 10 15 20 25 30 35 40 45 50
65

60
y2

55
0 5 10 15 20 25 30 35 40 45 50
200

190
y3

180

170
0 5 10 15 20 25 30 35 40 45 50
time t [hr]

(3)
Figure 4: Responses in outputs y1 = pb;H2 SO4 ,
y2 = ew;CujCuSO4 and y3 = ew;H2 SO4 upon a 50% step increase
(solid line) and decrease (dash-dot line) in u3 = V_ a after 5 h.

www.scansims.org Proceedings of SIMS 2008 140

Oslo, Norway, October 7-8, 2008 1