Zinc Beneficiation Simulator

CHE445Z
CHEMICAL ENGINEERING PROJECT
Development of a spreadsheet-based simulator for a

zinc beneficiation practical
Mineral Processing Research Unit at the University of

Cape Town
PREPARED BY: MICHAEL FLETCHER (FLTMIC007)

NEIL ROBINSON (RBNNEI002)
PREPARED FOR: DR. D.A. DEGLON

MR. P. LEKOMA
13 November, 2002
© Copyright 2002
Development of a Spreadsheet Based Zinc Simulator Synopsis
Synopsis
The hydrometallurgical zinc beneficiation process is studied by third year chemical
engineering students at the Western Cape Mineral Processing Facility in Stellenbosch. Currently,
students learn by first researching the zinc beneficiation process and giving a presentation to
demonstrate theoretical knowledge gained. They then perform experiments which demonstrate
the principles behind the industrial process. It is felt that the theoretical and practical knowledge
gained from these tasks is not adequately reconciled.
To combat this problem the process (from milling through to electrowinning) has been
successfully modelled using the Microsoft Excel spreadsheet add-on LIMN.
It is proposed that the simulator should be used as an educational tool for the third year
students. It is hoped that exposure to the simulator will help students to bridge the gap between
their theoretical and practical knowledge.
The simulator successfully shows trends in the process with changes of variables, but does
not, and is not intended to predict accurate process values.
The major assumption made for the entire process was that the ore fed to the system
consisted of pure ZnS. i.e. no impurities.
The models chosen for each unit operation are:
Milling: Standard population balance model, incorporating JKMRC Massive Sulphide
ore appearance function. Cyclone cut based on reduced recovery curve
Flotation: 1st order kinetics model, with rate constant estimated using the Jameson
equation
Roasting: Mass transfer limited shrinking core model in a fluidised bed
Leaching: Thermodynamic equilibrium model
Electrowinning: CSTR operating in a semi-batch environment
Several student learning assignments are proposed. They are:
A Fresh Feed Exercise
An Optimisation Exercise
Expected outcomes from the learning assignments are:
Understanding of knock-on effects in a system
Appreciation of the sensitivity of unit operations
An appreciation of the impact of economic and production factors on optimisation
i
Development of a Spreadsheet Based Zinc Simulator Table of Contents
Table of Contents
Synopsis ----------------------------------------------------------------------------------------------------i
Table of Contents---------------------------------------------------------------------------------------- ii
List of Figures ------------------------------------------------------------------------------------------ iv
List of Tables -------------------------------------------------------------------------------------------- iv
Nomenclature -------------------------------------------------------------------------------------------- v
1 Introduction ---------------------------------------------------------------------------------------- 1
1.1 Background -------------------------------------------------------------------------------------------- 2
1.2 Practical Procedure ---------------------------------------------------------------------------------- 2
1.3 LIMN – The Flowsheet Processor----------------------------------------------------------------- 3
1.4 Aims and Objectives --------------------------------------------------------------------------------- 4
2 Review of the Hydrometallurgical Route ------------------------------------------------------ 5
2.1 Communition and Flotation ------------------------------------------------------------------------ 6
2.2 RLE (Roasting, Leaching, Electrowinning) Process Review -------------------------------- 6
2.2.1 Roasting of Zinc Bearing Ore -------------------------------------------------------------------------------6
2.2.2 Leaching -------------------------------------------------------------------------------------------------------7
2.2.3 Electrolysis ----------------------------------------------------------------------------------------------------8
3 Review of Mathematical Models ---------------------------------------------------------------- 9
3.1 Modelling of the Ball Mill------------------------------------------------------------------------- 10
3.1.1 Population Balance Model -------------------------------------------------------------------------------- 10
3.1.2 Modelling the rate of breakage---------------------------------------------------------------------------- 11
3.1.3 A simplified Ball Mill Model ----------------------------------------------------------------------------- 12
3.2 Hydrocyclone Modelling -------------------------------------------------------------------------- 12
3.3 Flotation circuit Modelling ----------------------------------------------------------------------- 15
3.3.1 First order Flotation Kinetics------------------------------------------------------------------------------ 15
3.3.2 Estimation of the First order rate Constant -------------------------------------------------------------- 15
3.3.3 Second order Flotation Kinetics -------------------------------------------------------------------------- 16
3.4 Roasting Models – Kinetic and Thermodynamic -------------------------------------------- 16
3.4.1 Roasting Thermodynamics -------------------------------------------------------------------------------- 16
3.4.2 Bubbling Bed Model --------------------------------------------------------------------------------------- 17
3.4.3 Shrinking Core Model ------------------------------------------------------------------------------------- 18
3.5 Zinc Leaching Models ----------------------------------------------------------------------------- 18
3.5.1 Thermodynamics of Leaching ---------------------------------------------------------------------------- 18
3.5.2 Leaching Kinetics – Mass-Transfer Limited Model --------------------------------------------------- 19
3.5.3 Thermodynamic Equilibrium Model --------------------------------------------------------------------- 20
3.6 Electrowinning Models ---------------------------------------------------------------------------- 20
3.6.1 Single pass reactors----------------------------------------------------------------------------------------- 20
3.6.2 Semi-batch reactors----------------------------------------------------------------------------------------- 21
4 Models Used and Major Assumptions Made -------------------------------------------------22
4.1 The Simulator Appearance ----------------------------------------------------------------------- 23
4.2 Ball Mill Model-------------------------------------------------------------------------------------- 23
ii
Development of a Spreadsheet Based Zinc Simulator Table of Contents
4.2.1 Model Selection--------------------------------------------------------------------------------------------- 23

4.2.2 Assumptions Made ----------------------------------------------------------------------------------------- 24
4.2.3 User Input Variables --------------------------------------------------------------------------------------- 24
4.3 Hydrocyclone model ------------------------------------------------------------------------------- 25
4.3.1 Model Selection--------------------------------------------------------------------------------------------- 25
4.3.2 Assumption Made ------------------------------------------------------------------------------------------ 25
4.4 Flotation Model ------------------------------------------------------------------------------------- 25
4.4.1 Model Selection--------------------------------------------------------------------------------------------- 25
4.5 Roasting Models ------------------------------------------------------------------------------------ 26
4.5.1 Thermodynamic Model Selection ------------------------------------------------------------------------ 26
4.5.2 Kinetic and Mass Transfer Model Selection ------------------------------------------------------------ 27
4.5.3 Assumptions------------------------------------------------------------------------------------------------- 27
4.5.4 Roasting Unit User Inputs --------------------------------------------------------------------------------- 27
4.6 Leaching Models------------------------------------------------------------------------------------ 28
4.6.1 Mass Transfer Limited Model ---------------------------------------------------------------------------- 28
4.6.2 Thermodynamic Equilibrium Model --------------------------------------------------------------------- 29
4.6.3 Assumptions------------------------------------------------------------------------------------------------- 29
4.6.4 User Inputs -------------------------------------------------------------------------------------------------- 29
4.7 Electrowinning Models ---------------------------------------------------------------------------- 29
4.7.2 User Inputs -------------------------------------------------------------------------------------------------- 30
5 Results and Discussions -------------------------------------------------------------------------31
5.1 Individual Unit Trends ---------------------------------------------------------------------------- 32
5.1.1 Changing the Ball Mill speed ----------------------------------------------------------------------------- 32
5.1.2 Changing the superficial gas velocity during flotation------------------------------------------------- 33
5.1.3 Changing the acid flowrate through the leaching tank ------------------------------------------------- 34
5.1.4 Changing the current used in the Electrowinning cells ------------------------------------------------ 35
5.2 Circuiting Effects ----------------------------------------------------------------------------------- 36
5.2.1 Effect of changing the Fresh Feed to the system ------------------------------------------------------- 36
5.2.2 Effect of changing the Fresh Feed size distribution ---------------------------------------------------- 37
6 Concluding recommendations------------------------------------------------------------------38
6.1 Student Exercises ----------------------------------------------------------------------------------- 39
6.1.1 Assignment 1 – Fresh Feed Exercise--------------------------------------------------------------------- 39
6.1.2 Assignment 2 – Optimisation Exercise ------------------------------------------------------------------ 40
6.2 Possible improvements to the Simulator------------------------------------------------------- 40
6.3 Acknowledgements --------------------------------------------------------------------------------- 41
7 References -----------------------------------------------------------------------------------------42
8 Appendices-----------------------------------------------------------------------------------------45
Supplementary Calculations ------------------------------------------------------------------------------- 45
Method Used to Vary Size Distribution------------------------------------------------------------------ 46
Simulator Spreadsheets ------------------------------------------------------------------------------------- 47
iii
Development of a Spreadsheet Based Zinc Simulator List of Figures
List of Figures
Figure 1.1 Schematic diagram showing student learning----------------------------------------------- 3

Figure 2.1 Example of a typical fluidised bed roaster (From Reuter and Lans, 2001)------------- 6
Figure 2.2 Flow sheet of the leaching process at IMMSA, excluding advanced refining.
(Adapted from information and diagrams by Alfar and Castro, 1998, and Reuter and
Lans, 2001) --------------------------------------------------------------------------------------- 7
Figure 3.1 Ball Mill in closed circuit with hydrocyclone----------------------------------------------- 12
Figure 3.2 Reduced efficiency curve based on partition curve --------------------------------------- 13
Figure 3.3 Partially reacted ZnS particle, illustrating gaseous diffusion in and out of the system.
(Adapted from Fogler, 1999) ----------------------------------------------------------------- 18
Figure 3.4 Simple semi-batch reactor system with continuous or intermittent addition of
reactant (Adapted from Pletcher and Walsh, 1993)--------------------------------------- 21
Figure 4.1 Schematic layout of the Unit operations used in Zinc Beneficiation ------------------ 23
Figure 5.1 Graph showing the effect of mill speed on mill product size --------------------------- 32
Figure 5.2 Graph showing the effect of volumetric gas flowrate on flotation recovery --------- 33
Figure 5.3 Graph showing the effects seen when changing the mass flow rate of acid ---------- 34
Figure 5.4 Graph showing the effect when the current in the electrowinnig cell is altered ----- 35
Figure 5.5 Overall process efficiency as a function of fresh feed------------------------------------ 36
Figure 5.6 Overall process efficiency as a function of the fresh feed size distribution----------- 37
Figure 8.1 Chart illustrating effect of shifting particle size distribution to the right -------------- 46
List of Tables
Table 3.1 Table showing the Breakage Appearance Distribution matrix ------------------------- 10
Table 3.2 Table showing Rajamani and Herbst (1991) variables in hydrocyclone modelling- 14
Table 4.1 JKMRC Massive Sulphide ore appearance function ------------------------------------- 24
iv
Development of a Spreadsheet Based Zinc Simulator Nomenclature
Nomenclature
Symbol Description Units

α Constant parameter representing efficiency
τ residence time s
ρb Density of grinding material kg/m3
τb Flotation residence time in the bank hr
τT residence time in mixing tank s
A electrode area m2
Ai cross-sectional of particle of size fraction i m2
As cross-sectional area of the particle m2
bij mass fraction of particle of size that appears in size i after breakage
φc Volume fraction of ZnS fraction
c(IN, 0) initial concentration of reactant mol.m-3
c(IN, t) concentration of reactant at time t mol.m-3
C1 Fraction of cyclone feed above critical size reporting to underflow
C2 Fraction of cyclone feed below critical size reporting to underflow
CA0 Concentration of gas mol.m-3
cb concentration in the bulk of the solution mol.dm-3,
mol.m-3
cs concentration at the solid surface mol.dm-3
d25 Particle diameter where 25% of feed reports to underflow µm
d50 Cut size µm
d50(c) Corrected Cut Size µm
d75 Particle diameter where 75% of feed reports to underflow µm
db Bubble diameter mm
Dc cyclone diameter cm
De Diffusivity m2.s-1
di Diameter of particles in size fraction i mm
di Discharge rate from mill hr-1
Di inlet diameter cm
di,j rate of discharge of particle size i or j hr-1
DM mill diameter m
Do vortex finder diameter cm
dp Particle diameter mm
Du apex diameter cm
F Faraday’s constant A.s.mol-1
fi feed of size fraction i t/hr
Finsol mass flow rate of insoluble material kg.s-1
v

-1
Fsol mass flow rate of soluble material kg.s
h distance from bottom of vortex finder to top of underflow orifice cm
H mass hold-up in mill (subscript M) tonnes
iL mass transport controlled limiting current A
JB fractional mill filling with balls
Jg Superficial gas velocity m/s
k first order rate constant hr-1
kcat specific reaction rate (determined experimentally)
kf kinetic rate constant for fast floating fraction hr-1
ki First order reaction constant for flotation of size fraction i hr-1
kL mass transport coefficient m.s-1
km mass transfer coefficient dm.s-1
ko specific rate of breakage constant hr-1
KR a function of mass transfer coefficients, volume of solid particle in
bubbles, clouds and emulsion, and the specific reaction rate
ks kinetic rate constant for slow floating fraction hr-1
L mill length m
M solids feed rate of fresh feed (FF) and underflow (UF) ton/hr
M Mass of mineral floated ton/hr
M0 Mass of mineral originally in the cell ton/hr
M0,f mass of fast floating mineral initially in cell ton/hr
M0,s mass of slow floating mineral initially in cell ton/hr
Mi Mass flow of size fraction i t/hr
n Number of cells in the bank
n number of electrons dimensionless
n molar flux mol.dm-2
N rotational rate of the mill (fraction of critical)
ni mass flux of Zn from size fraction i kg.m-2
P Net power draw kW
Pa Probability of attachment
Pc Probability of collision
Pcoll Probability of collection
Pd Probability of detachment
pi,j product flow of size fraction i or j t/hr
Q Volumetric flow rate of feed slurry m3/hr
Q volumetric flowrate m3.s-1
Qpulp volumetric flowrate of slurry m3/hr
R the fraction of material above critical size in mill (subscript M), fresh
feed (subscript FF) and underflow (subscript UF)
R constant ratio of mass of solvent per mass of insoluble solids kg/kg
R Radius of core at time t m
R0 Initial radius m
vi

r Ei specific selction function hr-1
Ri Fractional Recovery of size fraction i

ri,j breakage rate of particle size i or j hr-1
S mass flow of pure solvent kg.s-1
S, L density of solids and liquids respectively g/cm3
Si breakage rate of particle size i hr-1
t time s, hr
ub velocity of bubble rise m.s-1
V volumetric percentage of solids in the feed %
V0 initial volume of particle m3
Vb volume of bulk solution m3
Vcore volume of particle core after time t m3
Vpulp slurry hold-up m3
VR reactor volume m3
x Normalised particle size (d/d50)
X conversion, where A denotes species A, single denotes single pass fraction
Xout mass soluble material per solute free solvent in underflow kg/kg
Y Fraction of Feed appearing in the underflow
Yout mass soluble material per mass solute free solvent in overflow kg/kg
δ fraction of total bed occupied by bubbles fraction
εmf porosity at minimum fluidisation velocity fraction
ρs density of solid kg.m-3
vii
Development of a Spreadsheet Based Zinc Simulator Introduction
Introduction and Background
1 Introduction
1
1.1 Background
The Departments of Chemical Engineering of the University of Cape Town, the University of
Stellenbosch and Cape Technikon jointly own and operate the Western Cape Mineral Processing
Facility (WCMPF) in Stellenbosch. In 2001, the practical course that 3rd year mineral processing
students undertake was reorganised to concentrate on the zinc beneficiation route, more
precisely the hydrodynamic route. The course sets out five practicals that the students are to
complete and are listed below:
Operation of a Continuous Ball Mill in closed circuit with a Screen Classifier

Kinetics of Continuous Flotation [of Zinc Sulphide]
Roasting
Leaching [and iron removal]
Introduction to electrometallurgy principles
These experiments are designed to illustrate the principles of the RLE process (Roast-Leach-
Electrowinning) to students, which is the most common process to produce zinc (Reuter and
Lans, 2001). The other main route is the pyrometallurgical route which involves pre-treatment
(e.g. sintering), concentration (e.g. Waelz kiln process) and production (eg. Imperial Smelting
Process). This route of zinc beneficiation is however of little interest in this modelling exercise
as it is not studied at the WCMPF. Milling and flotation are common to both routes as they are
the accepted means of size reduction and ore concentration in zinc treatment.
Of the five practicals that were designed, only four are currently physically conducted by the
students. The roasting practical was deemed exceedingly “boring” by students and
demonstrators alike in 2001, and has therefore become a paper exercise based only on theory.
1.2 Practical Procedure

At the beginning of the second semester, students who are completing the mineral processing
practical course are required to give a presentation on zinc beneficiation. A literature pack is
distributed for students to obtain references from and the following topics are to be covered in
their presentations.
1. Environmental impact, legislation and reasons for recycling

2. Mineralogical origin, properties, applications, production and market performance
3. Hydrometallurgical Zn-Extraction processes
4. Pyrometallurgical Zn-Extraction processes and some new recycling developments
After the presentations, the students are then required to perform all four practicals in no
particular order. If one was to schematically represent the way in which students learn, Figure
1.1 would best show the two areas of interest in this thesis. As can be see in Figure 1.1, a
students understanding of zinc beneficiation will be derived from theory (Presentations) and
2
practice (practicals).
This project sets out to simulate the processes that

are mimicked by the experiments using purely
theoretical models. The use of these models will
Theoretical
allow students to experiment with variables that Knowledge
are impossible to change during the allocated
practical time.
Student
Knowledge
The final model that describes the whole system is
to be used as an education tool for the students. It Practical
will allow them to see how changing key factors in Knowledge
certain unit operations affects units downstream,

and what the impact on the final product is. This
will hopefully enhance the learning of students by
Figure 1.1 Schematic diagram showing
creating a stronger link between theoretical student learning
knowledge and the lessons learnt by completing
the practicals. Students will therefore complete their presentations as before, but then before
going on to do the practicals, they will be exposed to the simulator. This will enable them to get
familiar with how the units are described in theory and what trends they can expect from making
changes to input variables. Students will then go onto complete the practicals, hopefully with a
better understanding of why certain effects take place.
1.3 LIMN – The Flowsheet Processor

LIMN is a spreadsheet based flowsheet processor that is an add-on application for Microsoft®
Windows™ Excel software. LIMN has the ability to draw flowsheets quickly and easily with a
large selection of built-in icons, as well as the ability to draw one’s own icons. The flowsheet is
then coupled with LIMN’s general purpose flowsheet solution engine to make it simple to
complete mass balances.
Utilising individual sheets to describe unit operations, one can use standard spreadsheet
functions to set up process models and solve for the product streams from a particular unit.
Macros embedded in the flowsheet enable the units to have inputs and outputs specified (linked
to the unit ops sheet) and allow the overall mass balance to be solved iteratively – a task which is
cumbersome without LIMN.
LIMN is however not a dynamic simulator and only solves for steady state, but still gives a good
indication of how certain parameters affect the overall product. LIMN is an additive system, i.e.
streams need to be specified in a measurement that can be added and subtracted directly (such as
mass). When certain unit ops use non additive measurements (e.g. concentration), one will need
to convert the product streams into an additive measurement before they can be used in the
flowsheet.
3
1.4 Aims and Objectives

Taking all the above into account, the following main points can be used to describe the
expected outcomes of this project:
1. The simulator is to be an educational tool that enhances student learning

2. The simulator is designed to predict trends, rather than accurate process values
3. Students using the simulator must be exposed to the effects of changing variables on
individual unit operations as well as the knock-on effect that changes in one unit can
have down stream
4. From the results of several simulation runs, students must be able to interpret results
from a theoretical and practical viewpoint
5. The simulator inputs and outputs must be clear and easy to use
4
Development of a Spreadsheet Based Zinc Simulator Review of the Hydrometallurgical Route
Review of the Hydrometallurgical Route
2 Review of the Hydrometallurgical Route
The hydrometallurgical route for the beneficiation of zinc begins with

communition and flotation. The concentrate from flotation usually proceeds to
a refinery for further processing. This processing involves roasting, leaching
and electrowinning.
5
2.1 Communition and Flotation

The SME Mineral Processing Handbook (1985) gives a concise overview of various Pb-Zn-Cu
and Pb-Zn concentrators around the world. Each operation varies slightly in approach,
depending on the precise mineralogy of their ore, but the general process can be gleaned from
the information easily enough.
Initially ore is crushed, often in underground crushers, such as the Brunswick and Buick
concentrators. Crushed ore is ground in milling circuits, consisting of primary rod-mills and
secondary ball-mills. Generally, the mills are operated in closed circuit with hydrocyclones as the
classifiers of choice.
Ore that has exited the size reduction stages is floated. Two flotation circuits exist at
concentrators – firstly the lead float, tailings of which proceed to the zinc float. In some cases
(e.g. Buick) regrinding is employed in the flotation circuits.
2.2 RLE (Roasting, Leaching, Electrowinning) Process Review

The RLE process is not the only option available to refine zinc, but it is the most popular
(Reuter and Lans, 2001). This section will describe the process in more detail incorporating
practices of the zinc refinery of IMMSA in Mexico. This refinery has used the RLE process for
over 20 years, and treats the floated concentrate from six mines (Alfaro and Castro, 1998). Thus
it is a good case study of an operation in the industry.
2.2.1 Roasting of Zinc Bearing Ore

The purpose of roasting is to convert the water
insoluble zinc sulphide into soluble zinc oxide.
Simultaneously ferrous content in the ore is
oxidised. Some of the zinc oxide reacts with iron
oxide to form zinc ferrite. Zinc ferrite is not soluble
in water and so complicates the leaching stage
(Reuter and Lans, 2001). The chemical reactions
that occur in the roaster are explicitly stated in
Section 3.4.1, on page 16.
The IMMSA refinery uses a Lurgi fluidised bed
roaster to process an average of 600t/d of zinc
concentrate. Roasting temperatures are maintained
at around 920°C using water sprays.
Energy from the hot flue gas leaving the roaster is
recovered by a boiler. This recovered energy is used
to drive the water spray pumps and other utilities. Figure 2.1 Example of a typical fluidised
bed roaster (From Reuter and Lans, 2001)
After further gas cleaning, the gas (consisting of 10%
to 12% SO2) proceeds to the Sulphuric Acid Plant
(Alfar and Castro, 1998).
6
An alternative to conventional fluidised-bed roasting is flash roasting, using the TORBED

reactor. This technology is best applied to fine powders of a maximum size less than 50 microns.
Residence time is in the order of milliseconds (Dodson et al., 1999)
2.2.2 Leaching
There are three distinct leaching stages of the calcine product from the roaster. They are Neutral
Leaching, Weak Acid Leaching and Hot Acid Leaching. A flow sheet of the process is illustrated
in Figure 2.2. The IMMSA has further refining steps where valuable trace metals such as
cadmium are recovered (Alfar and Castro, 1998). These refining steps are of no interest for the
purposes of this project.
The following description of the leaching process is elaborated on by Reuter and Lans (2001),
and Alfar and Castro (1998).
Spent Electrolyte Neutral Calcine
Leach
To electrolysis
Separation
Weak Acid
leach
To iron removal
Separation
Spent electrolyte and

Hot Acid Concentrated H2SO4
Leach
Separation
Solid residue
Figure 2.2 Flow sheet of the leaching process at IMMSA, excluding advanced
refining. (Adapted from information and diagrams by Alfar and Castro, 1998, and Reuter
and Lans, 2001)
7
Calcine from roasting and spent electrolyte is added to a cascade of neutral leaching tanks. These
tanks are mixed and sparged with oxygen. The streams are pumped to thickeners. Overflow
from the thickeners proceeds to electrowinning, while the underflow goes to the weak acid leach.
The weak acid leaching stage is fed by the underflow from the neutral leach stage and the
overflow from the hot acid leach. The overflow goes to the iron removal stage where either the
jarosite or goethite process is employed to precipitate the iron out of the solution. The overflow
from that process then proceeds to electrolysis.
Hot acid leaching allows the zinc entrained in the zinc ferrite to be leached. The hot acid leach
tanks are fed concentrated sulphuric acid and spent electrolyte. The tanks and agitators require
special coatings to prevent corrosion.
2.2.3 Electrolysis
At the IMMSA, purified solution from leaching is received in two 1000m3 tanks. It is pumped
continuously to forced convection cooling towers. This is mixed with spent electrolyte in the
main distribution launder. This combined solution is further cooled in more cooling towers
before being distributed amongst the cells. The cell house has 384 concrete cells with paraliners.
The cell cleaning cycle is approximately 30 days (Alfar and Castro, 1998).
Cathode surface areas for cells range from 1 to 4.5m2, and current density from 280 to 640A/m2
(James et. al., 2000).
8
Development of a Spreadsheet Based Zinc Simulator Review of Mathematical Models
Review of Mathematical Models
3 Review of Mathematical Models
This section will cover most of the Mathematical models available to simulate
the unit operations. Some are adapted from other applications whilst others
are very well documented and commonly used.
9
3.1 Modelling of the Ball Mill

There are three approaches to describe the main phenomenon that takes place in a ball mill
(Yovanovic and Moura, 1993).
The process is represented either chemically or physically using Microphenomenological

models;
Statistical models based on regression of a series of experiments that describe only the
results and not the process itself;
Models which describe certain macroscopic, physical aspects of the process
(macrophenomenological)
3.1.1 Population Balance Model

The most widely used models are those of the macrophenomenological type and they can be
subdivided into two categories: energetic and kinetic. The more common of these two is the
kinetic approach. Here the mill is assumed to be a reactor where the larger particles are reacting
to form smaller particles. This model can be described by first order reaction kinetics as seen in
Equation 3.1:
dM i (t )
= − Si ⋅ M i (t ) Equation 3.1
dt
where Si is the selection function and represents the rate of breakage. This corresponds to the
kinetic constant in first-order chemical reactions (Yovanovic and Moura, 1993). When using this
model one also needs to take into account the breakage function (bij) which represents the
fraction of larger particles that are reduced and can be found in the next (smaller) size fraction.
The breakage function corresponds to the stoichiometric coefficient in a chemical reaction
(Yovanovic and Moura, 1993).
The Breakage function is commonly found in the form of a normalised matrix and a simple
example can be seen in Table 3.1.
Table 3.1 Table showing the Breakage Appearance Distribution matrix
bij Breakage from size 'j'

1 2 3 4 5
1 0 0 0 0 0
Fraction ending
2 0.5 0 0 0 0
3 0.3 0.5 0 0 0
4 0.2 0.3 0.5 0 0
in 'i'
5 0 0.2 0.5 1 0
10
A simple way to describe the meaning of Table 3.1 is to look at the behaviour of one size
fraction. Consider size fraction 1 (j), A one looks down the column one can see that of the
particles of this size fraction that are broken, 50% appear in size fraction 2, 30% in size fraction
3, 20% in size fraction 4 and none in size fraction 5. Similarly one can view the breakage
distribution of each size fraction. One can look across a row (i) to discover the appearance of
material. Therefore, looking at size fraction 4, we can tell that during breakage 20% comes from
size fraction 1, 30% from size fraction 2 and 50% from size fraction 3.
If it is assumed that the mill is operating at steady state and that there is perfect mixing taking
place inside the mill, then Equation 3.2 can be used to solve the mass balance over a particular
size fraction (Morrell and Man, 1997).
i rj ri
0 = f i − pi + ∑ bij pj − pi Equation 3.2
j =1 dj di
Using this equation, one can solve for the product flow of all the size fractions individually,
provided that the rate of breakages of each size fraction is known as well as the breakage
function. The discharge of a particular size fraction can be calculated as follows in Equation 3.3:
Q pulp
di = Equation 3.3
V pulp
If there is no segregation in the mill then all solid particles behave like water and thus di will be a
constant for all size fractions (Morrell and Man, 1997). Using Equation 3.3, Equation 3.2 and
Table 3.1, one is able to solve for the product flowrate of all size fractions.
3.1.2 Modelling the rate of breakage

Morrell and Man (1997) also found that the specific rate of breakage of a size fraction can be
related to certain mill properties as described below.
J B DM2 LN
ri ∝ Equation 3.4
V pulp
If the above equation is combined with Equation 3.3 it can be found that the rate of breakage is
directly proportional to the net mill power divided by the hold-up in the mill as described in the
following equations.
P
ri ∝ Equation 3.5
V pulp
11
P ∝ DM3 Lρ b J B Equation 3.6
Equation 3.5 above has been varified by Herbst et al. (1977) who used milling tests in a 25-, a 38-
and a 76-cm ball mill and found the selection functions to be directly proportional to the specific
power draw of the mill. Namely;
 P 
ri = ri E   Equation 3.7
V 
 pulp 
where rEi is the specific selction function and can be calculated using a simple equation that was
formulated by Rajamani and Herbst (1991) and is written below.
1.427
 dd 
ri ∝  i i −1 
E
Equation 3.8
 dd 
 1 2 
3.1.3 A simplified Ball Mill Model

A more simplified model is put forward by Rajamani and Herbst (1991) to enable slower on-line
computers (namely the HP2100) to predict results to a reasonable level of accuracy. The
following equation was proposed to predict the closed circuit operation of a ball mill.
dRM
HM = M FF RFF + M UF RUF − k o H M RM − (M FF + M UF )RM Equation 3.9
dt
The simplification that is made here is the combinations of size fractions. This model only looks
at a specific size and then looks at what is above it and what is below it. This critical size would
in most instances be the same, or very close to, the cut size defined in the cyclone or screen
controlling the recycle to the mill. The value of the breakage constant would have to be solved
experimentally.
3.2 Hydrocyclone Modelling

Most mills run today are done so in a closed
circuit (i.e. they have a recycle system) as in
Figure 3.1. These recycle streams are first
classified before being fed back to the mill. The
most common classifier used is a hydrocyclone
due to its greater efficiency (Wills, 1988). In
order to model the cyclone, one first needs to Figure 3.1 Ball Mill in closed circuit with
define the cut size (d50). The cut size, as defined hydrocyclone
by Wills (1988, pp 358), is the size of a particle
12
that has an equal chance of going to either the overflow or underflow. Most mathematical
models use a corrected cut size (d50(c)) which accounts for the short-circuiting that occurs in the
cyclone.
Modelling of hydrocyclones is very complex as all parameters involved are interrelated. Plitt
(1976) developed an empirical model that enables cyclone performance to be calculated with
reasonable accuracy without the need for experimental data (Wills, 1988).
14.8Dc0.46 Di0.6 Do1.21e 0.063V

d 50(c ) = Equation 3.10
Du0.71h 0.38Q 0.45 (S − L )
0.5
These models are however only useful if the cut size needs to be determined from a specific
cyclone size. Another approach to the modelling of a cyclone is to specify a specific cut size and
then determine how efficient the cyclone will perform. This is most commonly achieved
through the use of a partition curve as seen in Figure 3.2 (Wills, 1988). However it must be
noted that Figure 3.2 is in fact a reduced efficiency curve but the curve shape is identical to that
of a partition curve. The efficiency of the cyclone can be described by what is called the
imperfection (Wills, 1988) is defined by Equation 3.11. This imperfection represents the slope
of the partition curve. The closer the slope is to vertical, the more efficient the hydrocyclone
(Wills, 1988).
1.000
Ideal
Fraction of feed appearing in underflow
Real
0.500
0.000
0.000 1.000 2.000
(d/d50)
Figure 3.2 Reduced efficiency curve based on partition curve
13
d 75 − d 25
I= Equation 3.11
2d 50
The curve seen in Figure 3.2 can be described mathematically by Equation 3.12 (Tarr, 1985).
This equation however only takes into account the solids fraction recovery and doesn’t take into
account any short-circuiting that may take place in the cyclone. Short-circuiting is the term used
to describe the effect of solids that are entrenched in the flow and pass through the cyclone
unclassified.
eαx − 1
Y = αx Equation 3.12
e + eα − 2
The constant α in Equation 3.12 is a representation of the efficiency and is dependent of the
feed material. Equation 3.12 was first used by Yoshioka and Hotta in 1955 (Tarr, 1985) who
determined that α falls in the range of 3 to 4 for single stage processes in metallurgical plants.
When some washing (short-circuiting) is possible then α may reach 5 or 6. The larger the value
of α, the more efficient the cyclone will be. It can be found that α will normally range between
1 and 10 (Tarr, 1985).
The approach taken by Rajamani and Herbst (1991) to simplify the ballmill model results in its
own accompanying model for the attached hydrocyclone. As mentioned before, the model
proposed combines size fractions so that there are only two parts, those above the critical size
and those below it. Rajamani and Herbst therefore propose that the classification action be
described by two constants, C1 and C2. the dependancy of C1 and C2 on the cut size was foundto
be linear and can be described by Equation 3.13 and Equation 3.14
C1 = a11 d 50 + a12 Equation 3.13
C 2 = a 21 d 50 + a 22 Equation 3.14
The model parameters used by Rajamani and Herbst (1991) were calculated at steady-state
experimentation using linear regression and their results can be seen in Table 3.2.
Table 3.2 Table showing Rajamani and Herbst (1991)

variables in hydrocyclone modelling
a11 0.0056
a12 1.181
a21 0.0480
a22 -2.029
14
3.3 Flotation circuit Modelling
3.3.1 First order Flotation Kinetics

The flotation process is used to concentrate the sphalerite (ZnS) present in the ore. The
modelling of a flotation process is very complex, as it is not yet fully understood. Due to this
there are a wide variety of models available. The simplest model available is the first order
kinetics model (Fichera and Chudacek, 1992) as first proposed by Zuniga, 1935.
(
M = M 0 1 − e − kt ) Equation 3.15
Flotation is governed by first order kinetics, and the most common models found in literature
are based on this assumption. In industrial practice, cells are commonly found in banks which
contain any number of cells in series. Using first order kinetics one can find an expression to
describe the recovery obtained at the end of the bank as seen in Equation 3.16, this is a size by
size recovery.
1
Ri = 1 − n
 τ  Equation 3.16
1 + k i b 
 n
The rate constant would in most instances be solved from pilot plant scale batch testing and then
scale-up techniques used to design the actual flotation plant. However, for modelling purposes,
the rate constant can be determined from theoretical estimates. To do this, microkinetics needs
to be investigated and combined with some practical macrokinetics.
3.3.2 Estimation of the First order rate Constant

One of the most common methods of describing the first order rate constant is through the
Jameson equation (Jameson et al, 1977) and can be seen in Equation 3.17.
3 J g Pcoll
k= Equation 3.17
2 db
Pcoll is referred to as the probability of collection and takes into account three independent
probabilities, namely the probability of collision (Pc), attachment (Pa) and detachment (Pd). The
relationship can be seen in Equation 3.18.
Pcoll = Pc Pa (1 − Pd ) Equation 3.18
Many attempts have been made at modelling the probability of collision (Pc) using the bases of
hydrodynamic forces as the predominant force. Flint and Howard (1971) solved the Navier
Stokes equations numerically to obtain the probability of collision. Using this approach, Reay
15
and Ratcliff (1973) modified the modelling method and obtained an expression for the
probability of collection of the form seen in Equation 3.19.
m
 dp 
Pcoll ∝   Equation 3.19
 db 
The power (m) in Equation 3.19 is dependent on particle to fluid density and can range between
1 and 2.5. Equation 3.19 can again be modified to give Equation 3.20.
n
dp
Pcoll = m
Equation 3.20
db
Here it must be noted that n ≠ m
3.3.3 Second order Flotation Kinetics

There are also many other models that are not first order, the JKMRC (Julius Kruttschnitt
Mineral Research Centre) has found that a double distributed parameter model better describes
the flotation of sulphides (Thorne et al, 1976). This model uses the assumption that there are
two different rate constants, namely slow and fast floaters. This model was first proposed by
Kelsall (1961) and then later presented by Frew and Trahar (1981) as Equation 3.21:
(
M = M 0, f 1 − e
−k f t
)+ M (1 − e )
0,s
− k st
Equation 3.21
This model will however require that the feed composition, in terms of fast and slow floaters, be
known. It will also require knowing the rate at which fast and slow material floats and will in
most instances be solved using experimentation.
3.4 Roasting Models – Kinetic and Thermodynamic

This section is an overview of potential models that could be applied to the simulation of the
fluidised-bed roaster unit operation on a zinc refinery.
3.4.1 Roasting Thermodynamics

The desired roasting reaction in the RLE (Roast-Leach-Electrowin) process is
ZnS + 3/2O2 → ZnO + SO2 (Reuter and Lans, 2001) Equation 3.22
This reaction is achieved in the temperature range of 900°C to 950°C for a fluidised bed reactor
(Graf, 1996). The reaction predominance is also a function of partial pressure of oxygen and
sulphur dioxide.
The iron content in the ore leads to some undesired side-reactions, namely
16
2FeS + 7/2O2 → Fe2O3 + 2SO2 Equation 3.23
2FeS2 + 11/2O2 → Fe2O3 + 4SO2 Equation 3.24
the oxide products of which combine with ZnO to form zinc ferrite (ZnO.Fe2O3) which creates
problems in the leaching operation as the zinc cannot be liberated under normal leaching
conditions. (Reuter and Lans, 2001).
Which reaction products are formed, and in what quantities is controlled by temperature and
partial gas pressure.
3.4.2 Bubbling Bed Model

The most comprehensive and detailed fluidised-bed reactor model developed so far is the
bubbling bed model of Kunii and Levenspiel (Fogler, 1999). This model is based on the
following premise. The reactant gas (air in this case) enters the reactor at the bottom and bubbles
up through the bed. Mass transfer occurs as the reactant gas (O2) moves out of the bubble,
contacting with the solid particles (ZnS), forming the reactant product (ZnO). Product gas (SO2)
moves back into the bubble, and then leaves the reactor via the top of the bed.
The major factors affecting conversion for this model are:
1. Rate of mass transfer of products and reactants in and out of the bubble
2. Time for the bubble to pass through the bed
3. Rate of reaction at the particle surface
Unknown variables in this model are plentiful. Following the procedure laid out by Fogler (1999)
one calculates fluid mechanics parameters, namely: porosity at minimum fluidisation velocity,
minimum fluidisation velocity, velocity of bubble rise and bubble size. These parameters are then
used to calculate mass transport coefficients.
Armed with these values, reaction rate parameters can be determined by first calculating fraction of
bed occupied by bubbles, fraction of bed consisting of wakes, and volumes of solid reactant in
the bubbles, clouds, and emulsion.
Finally the mass of solid reactant (ZnS) required for a chosen conversion of oxygen to SO2 is
defined by Equation 3.25.
ρ s As ub (1 − ε mf )(1 − δ ) 1
W= ln Equation 3.25
k sol K R 1− X
It is important to note that Kunii and Levenspiel derive their model for the purposes of a
catalytic fluidised-bed, and not directly for ore roasting. The model as described by Fogler (1999)
uses Equation 3.25 to solve for mass of catalyst required for a chosen gas conversion. The model
has been directly adapted to roasting purposes, and not derived from scratch. It does not solve
for solids conversion which is the factor that is of interest in this case.
17
3.4.3 Shrinking Core Model

O2
Blair (1985) states that complete roasting of
sulphides to oxides happens very rapidly and
proceeds effectively to completion under fluid bed SO2
roasting conditions. He goes on to state that this is R
because of the porous nature of the calcines formed
which allow easy penetration of oxygen to the core,
and easy diffusion of SO2 from it.
Because of the turbulent nature of fluidised beds, it R0
is reasonable to assume that transport of gaseous
ZnO layer
reactant to the suspended particle surface is not
limiting. ZnS layer
In turn, this information leads to an assumption
that the kinetics of the reaction can be modelled Figure 3.3 Partially reacted ZnS particle,
illustrating gaseous diffusion in and out of the
using a shrinking core model. system. (Adapted from Fogler, 1999)
The shrinking core model applied to ZnS roasting assumes all particles are approximately
spherical, and that a layer of ZnO forms around the core of ZnS (that shrinks as it reacts with
oxygen). Oxygen diffuses in through the ZnO layer, and SO2 diffuses out. This is more clearly
illustrated in Figure 3.3.
A derivation of the shrinking core model yields time as a function of the unreacted (core) radius.
(Equation 3.26)
ρ c R02φc  
2 3
 R  R
t= 1 − 3  + 2   Equation 3.26
6 De C A0   R0   R0  
In the case of this educational tool, roasting time is a known parameter, and thus the cubic
equation would be solved for R.
3.5 Zinc Leaching Models

Calcine from roasting contains predominantly zinc oxide, about 4 wt% zinc sulphate and several
percent zinc ferrite (Gupta and Mukherjee, 1990).
No explicit models exist in the literature for leaching of zinc. Thus, one must look to general
thermodynamic principles and transport kinetics for possible correlations.
Thermodynamics gives one an indication of whether or not the reaction will occur under certain
conditions, while kinetics provides information on the rate of the reaction.
3.5.1 Thermodynamics of Leaching

The leaching process takes place in two steps. The cold neutral leaching reaction is:
18
ZnO (s) + 2H+(aq) ⇔ Zn2+(aq) + H2O (Hayes, 1985) Equation 3.27
Then to liberate zinc locked up in zinc ferrites, a hot acid leach is used:
ZnO.Fe2O3 + 8H+ ⇔ Zn2+ + 2Fe3+ + 4H2O Equation 3.28
The thermodynamics of Equation 3.27 and that of the zinc-water system is discussed by both
Hayes (1985) and Jackson (1986). Regions in which dissolution of zinc takes place are best
illustrated by a potential/pH diagram. In the modelled process the region of interest is below a
pH of 7 and at a potential of greater than -0.85V (at 25°C, and assuming an activity of 10-3
mol.kg-1).
Equation 3.28 is discussed in Zinc College Course Notes (Reuter and Lans, 2001), along with the
Goethite process (used to precipitate the iron from solution).
It is known that the reaction takes place under these conditions, and assuming the system
operates under these conditions, dissolution will take place. Thermodynamics do not give an
indication of the rate of reaction.
3.5.2 Leaching Kinetics – Mass-Transfer Limited Model

Jackson (1986) states that there are 4 stages in leaching when a gaseous reactant is involved. They
are:
1. Transfer of gaseous reactant from gaseous phase into liquid solution
2. Transport of reactant through solution to the solid-liquid interface
3. Reaction at the interface
4. Transport of products from the interface back into the bulk solution
Generally the rate limiting step is transport to the solid-liquid interface. For the purposes of the
leaching model this would be assumed to be the case.
A leach tank would be well mixed, with fluid velocities high enough to cause the turbulent flow
regime. Such conditions eliminate diffusivity as a significant factor in mass transfer, and the
traditional understanding of a concentration gradient becomes useless. The introduction of a
mass transfer coefficient compensates for these problems (Jackson, 1986).
The rate of mass transfer per unit area of surface area is expressed in Equation 3.29.
dn
= k m ( cb − c s ) Equation 3.29
dt
km can be estimated using the standard correlations involving the Sherwood number (Sh).
Assuming Zn2+ concentration at the particle surface and the mass transfer coefficient remain
constant with time, the mass of zinc leached into solution after a chosen time can be determined.
19
3.5.3 Thermodynamic Equilibrium Model

Although thermodynamic equilibrium is never achieved in practice, an equilibrium model can be
implemented, assuming ideal conditions.
Seader and Henley (1998) consider an N-stage countercurrent leaching process. Solids entering
the process are classified as soluble and insoluble. Pure liquid solvent fed to the system is
assumed to completely dissolve the soluble material, and to be inert with respect to the insoluble
material.
This model can be adapted for a single tank. The material balance of soluble material around this
single tank is:
Fsol = Yout(S – RFinsol) + XoutRFinsol Equation 3.30
At equilibrium, the ratio of mass of soluble material per mass of solute free solvent is equal in
the underflow and overflow. i.e.
Xout=Yout Equation 3.31
Substituting Equation 3.31 into Equation 3.30 yields Equation 3.32
Fsol
X out = Equation 3.32
S
Percentage recovery of soluble material can be expressed by
Yout ( S − RFinsol )
%Recovery = Equation 3.33
Fsol
3.6 Electrowinning Models

In modelling the electrowinning process, the electrowinning cells are treated as reactors. There
are three classical reactor configurations: simple batch reactor, plug flow reactor and constant stir
tank reactor (Pletcher and Walsh, 1993).
The simple batch reactor is labour intensive and best suited for small operations (Pletcher and
Walsh, 1993). This is not appropriate for this modelling exercise and so will not be further
investigated.
3.6.1 Single pass reactors

Considering a single pass reactor with a steady volumetric flow rate of electrolyte passing
through, the single pass conversion for a plug-flow reactor is expressed as:
20
 k A 
X Asingle = 1 − exp − L τ  Equation 3.34
 VR 
And for a constant stir tank reactor:
 
 1 
X single
= 1−  Equation 3.35
A
k A 
1 + L Q 
 
Conversion is limited by mass transport and electrode area, and so it is important to optimise
these variables. Pletcher and Walsh (1993) show that the product of kLA can be calculated as:
iL
kL A = Equation 3.36
nFcb
Equation 3.34 and Equation 3.35 are derived by Pletcher and Walsh (1993).
3.6.2 Semi-batch reactors

Typical tankhouse configurations are semi-
batch (Reuter and Lans, 2001), a schematic of
which is shown in Figure 3.4.
Reactor Mixing Tank
The equation describing the overall fractional

conversion of reactant in a semi-batch reactor
at time t, derived from the mass balance on
the system, is (Pletcher and Walsh, 1993):
Figure 3.4 Simple semi-batch reactor system
with continuous or intermittent addition of
reactant (Adapted from Pletcher and Walsh, 1993)
c( IN , t )  t 
X A =1− = 1 − exp − ⋅ X Asingle  Equation 3.37
c( IN , 0)  τT 
Substitute Equation 3.34 or Equation 3.35 into XAsingle, depending on whether a PFR or CSTR
type model is desired.
21
Development of a Spreadsheet Based Zinc Simulator Models Used and Major Assumptions Made
Models Used and Assumptions Made
4 Models Used and Major Assumptions Made
The following section will cover the exact way in which each unit was modelled
with reference to Section 3. The major assumptions that were made will be
listed along with the strengths and weakness of each particular model. All user
inputs that can be varied in the simulator will also be mentioned. The
Simulator layout can be seen in Figure 4.1 following. Please see the Appendix
for print outs of all the spreadsheets used.
22
4.1 The Simulator Appearance
Figure 4.1 Schematic layout of the Unit operations used in Zinc Beneficiation
4.2 Ball Mill Model
4.2.1 Model Selection

The model that was used for the ball mill was the standard population balance, taking into
account appearance and breakage (Equation 3.2). The values for the appearance function were
taken from the JKRMC’s ball mill model for Massive Sulphide containing lead, zinc and copper.
The first column of the appearance function can be seen below in Table 4.1. The additional
columns for the matrix were derived from Table 4.1 by moving it down one row per column
across. The last size fraction was normalised with respect to 1, to ensure that mass was
conserved.
23
Table 4.1 JKMRC Massive Sulphide ore appearance function
Size Fraction Appearance function

1 0
2 0.1081
3 0.1442
4 0.1472
5 0.1253
6 0.1006
7 0.0805
8 0.06444
9 0.05076
10 0.03958
11 0.03103
Below 11 0.10829
The selection function was based on the power draw associated with the mill. It was calculated
using a combination of the approach taken by Morell and Man (1997), Equation 3.5 and
Equation 3.6, Herbst et al (1977), Equation 3.7, and Rajamani and Herbst (1991), Equation 3.8.
These four equations were combined to solve for the selection function at any given mill size.
The Simplified model approach taken by Rajamani and Herbst (1991) seen in Section 3.1.3, was
not used due to the fact that it was over simplified for the computing power available when
using LIMN.
4.2.2 Assumptions Made

It was assumed that the appearance function for the mill would remain constant at all
times
The constants required to solve the selection function were chosen so that the selection
function was of a similar magnitude to that presented in LIMN (Wiseman, 1999)
75% solids assumed to be fed to mill
4.2.3 User Input Variables

Mill fresh feed (will effect downstream units)
Fresh feed size distribution
Mill dimensions (length and diameter)
Mill speed (defined as the fraction of the critical speed)
Fractional filling of the mill with balls
24
4.3 Hydrocyclone model

The model chosen for the hydrocyclone was that of a reduced efficiency curve as seen in Figure
3.2 and modelled using Equation 3.12. Due to the fact that there are many size fractions, which
tell us the maximum and minimum size the particles will lay between, it was decided that an
average particle size will be used when solving for d/d50. The average particle size was solved
using Equation 4.1 where n represents the size fraction in question.
d n = d n +1 d n Equation 4.1
This model was chosen because of its simplicity and general use in mineral processing
applications. The major weakness of this model is that it tells us nothing about the actual
cyclone (ie. dimensions). This however is not of major importance as in the practicals the mill is
actually in closed circuit with a vibratory screen. This would mean that the most common thing
changeable in the practical would be to change the screen size. Changing the screen size is
effectively the same as changing the cut size in the cyclone, therefore making this model more
appropriate to the students.
The value of α in Equation 3.12 was left as a user defined constant. α can possibly range
between 1 and 10, but for mineral processing plants, it is best defined between 3 and 6 (Tarr,
1985). The simplified model of Rajamani and Herbst (1991) was not used due to its application
being with the simplified ball mill model only.
4.3.2 Assumption Made

There is no short-circuiting taken into account
Separation is on a dry mass into and out the cyclone
Operates perfectly with respect to the partition curve

Cut size that cyclone operates at (d50)
Sharpness of the cut (efficiency)
4.4 Flotation Model

The model chosen for the flotation circuit was the first order kinetics model most commonly
used. The second order kinetics was not used due the fact that in order to use them, one would
need to know the composition of the ore with respect to fast and slow floaters. In practice this
is normally done through a large amount of test work, or from other plants in the area using
similar ore. One of the major assumptions made throughout this project is that the ore fed to
the mill, and the system, is pure ZnS. Therefore the grade of the concentrate will not be
25
considered.
The practical that is carried out by the students requires them to solve for the kinetic rate
constant for four continuous cells in series as a function of particle size. This gave more
justification to the use of first order kinetics as it would give students the opportunity to
compare the results they obtain in the practical with the results they generate using the simulator.
To solve for the rate constant, the Jameson equation (Equation 3.17) was combined with
Equation 3.20 (where n=1.5 and m=2) to form Equation 4.2.
1.5
3 Jgdp
k= Equation 4.2
2 d b3
Using Equation 4.2 and Equation 3.16 the mass recoveries of all the size fractions can be
obtained and hence the flowrate of concentrate and tailings determined.

Pure ZnS is feed to the system therefore grade not required
First order kinetics based on Jameson equation
Fixed bubble size
Square cells
Average particle diameter assumed similar to that of Section 4.3.1
Fixed percentage solids in feed to bank

Number of cells in bank
Dimensions of one cell
Volumetric air flowrate
Bubble size
Fraction solids in feed
4.5 Roasting Models
4.5.1 Thermodynamic Model Selection

The first consideration to take in setting up the roasting model is a thermodynamic
consideration. How much detail is required for the purposes of this educational tool?
In industry it is highly unlikely that the roaster operator will run the roaster at conditions that will
allow any product other than ZnO to dominate. Unforeseen occurrences could alter the
conditions such that undesired products dominate for a short period of time, but the operator
will correct this and so these conditions will never be allowed to reach steady state.
It is thus safe to discount thermodynamic effects for the purposes of this educational tool, as
long as temperature and partial pressure input to the simulator is kept within acceptable
26
thermodynamic bounds (i.e. conditions in which ZnO predominates as a product).
4.5.2 Kinetic and Mass Transfer Model Selection

The bubbling bed model (described in Section 3.4.2) would be the best model to use for a
fluidised-bed. Unfortunately the complexity of the model is inhibitive for the purposes of this
project. A further problem is that is derived to determine the weight of catalyst required for a
certain gas conversion. In the case of the roaster, conversion of gas is irrelevant. The mass
conversion of ZnS to ZnO is what is important, and so this model seems inappropriate in its
current form.
The shrinking core model is a superior choice to the bubbling bed model in these circumstances.
A particle size distribution in the roaster is known (the concentrate from flotation), and roasting
time can be chosen. Knowing R0 for each size fraction, R at time t can be calculated for each size
fraction. The volume of the particles is known, and so mass conversion can be determined. Note
that density is constant, so mass conversion can be calculated using volume.
3
V − Vcore R03 − R 3  R
X= 0 = 3
= 1 −   Equation 4.3
V0 R0  R0 
The obvious drawback of the shrinking core model is the large number of simplifying
assumptions that are required – as listed in Section 4.5.3.
4.5.3 Assumptions
The following assumptions and conditions have been applied in order to make the model
functional and useful as an educational tool in the short time available.
Shrinking core model is applicable

All particles are approximately spherical
Reaction is limited by oxygen transport through the outer shell of the particle
No mass transfer limitations in the bed of the reactor
Ideal gas law is reasonable to determine oxygen concentration at the particle surface
Nitrogen partial pressure has no impact on the reaction
Sulphur dioxide concentration remains constant at all times
Thermodynamic predominance is neglected, as long as partial pressure of oxygen, and
temperatures remain in acceptable ranges
Residence times in the millisecond range are acceptable – see Section 2.2.1.
4.5.4 Roasting Unit User Inputs

Oxygen partial pressure (does not take nitrogen content of air into account)
Reactor temperature
Roast retention time
27
4.6 Leaching Models

Due to the general assumption that the ore fed to the process consists of pure sphalerite (ZnS),
the complications that industry experience with zinc ferrite are avoided. This is a dramatic
simplification of the leaching system, eliminating the need for the hot acid leach stage and
jarosite precipitation. Thus the unit operation of the leach simulator will consist only of the
weak acid stage.
4.6.1 Mass Transfer Limited Model

The most detailed model to use is the mass transfer limited model discussed in Section 3.5.2. At
first glance, this model seems to be the best choice. It is not to complex to implement, while the
only other option – that of thermodynamic equilibrium – is perhaps an oversimplification,
especially since industrial leaching operations will never have great enough residence times to
reach equilibrium.
Unfortunately two complications present themselves when attempting to model the system on a
size-by-size basis (the basis on which all previously discussed units have been modelled).
Firstly, after roasting the particles are now no longer pure sphalerite. The particle core consists of
sphalerite, but the shell is ZnO. Only the ZnO is soluble in water. This problem can be resolved
by assuming that the whole particle is ZnO for the purposes of mass transport calculations, and
then balancing the mass of ZnS and ZnO at the end of the simulation.
The second problem is far more severe and more difficult to resolve.
Consider Equation 3.29, the rate of mass flux of the leached component (zinc in this case) into
solution. It is only applicable to a single size fraction. To incorporate multiple size fractions, the
following procedure is followed:
The concentration of zinc in the bulk solution is a sum of the fluxes from each size fraction. i.e.
∑ ni Ai
Equation 4.4
cb = i =1
Vb
Thus, for size fraction i, assuming concentration at the surface is a constant
 n 
 ∑ ni Ai 
dni 
= km i =1
− cs  Equation 4.5
dt  Vb 
 
 
This model has up to twelve size fractions. This function is impossible to integrate analytically,
and so can only be solved numerically.
Recall that LIMN operates in the Microsoft Excel spreadsheet environment. Excel does not have
any built-in capabilities to solve multiply differential equations. It would be necessary to write a
28
macro in the programming language VBA. Neither Fletcher nor Robinson have any knowledge
of VBA and unfortunately time allocated for the project was too short for them to attempt to
learn it.
As a result, this most promising approach to the leaching model was discarded.
4.6.2 Thermodynamic Equilibrium Model

The implemented option was the thermodynamic equilibrium model. Such a model would never
be used in totally predictive simulator, but at the conceptual stage of the project, with the mass
transfer model seemingly too complex, it appeared to be the only option. All mass fractions were
grouped together and a total mass independent of particle size was calculated to enter the
leaching unit. Conversion was then calculated following the equations set out in Section 3.5.3.
In hindsight, Fletcher and Robinson feel that it might not have been necessary to completely
discard the mass transfer model. All size fractions could have been grouped together as in the
thermodynamic equilibrium model, and the total mass of the size fractions summed. This
assumption would effectively reduce Equation 4.5 back to Equation 3.29. At the conceptual
phase of the project Fletcher and Robinson were focussed on delivering a simulator that would
operate on a size-by-size basis in every unit operation. Once this model was shown to be
unusable in this application, it was completely discarded. Once the decision was made to group
size fractions, it did not occur to them to re-examine the possibility that the mass transfer model
could be used in this way.
4.6.3 Assumptions
Thermodynamic equilibrium is achieved – i.e. ratio of mass of soluble material per mass
of solute free solvent is equal in the underflow and overflow
A single leaching tank is used
Leaching occurs independently of particle size
Zinc will only be leached from ZnO, while ZnS will pass through the system
Only concentrated H2SO4 is added to the leach – there is no spent electrolyte
recirculation
4.6.4 User Inputs

Acid flowrate
4.7 Electrowinning Models

The only consideration left with respect to the choice of electrowinning model is whether to
treat the reactor as a plug flow or constant stir tank reactor. All other considerations are
discussed in Sections 3.6 and 2.2.3.
The advantages and disadvantages of PFRs and CSTRs in industry are well documented
elsewhere. The choice simply came down to what most would enhance the students’ learning
experiences. In the practicals that the students perform, well mixed batch reactors are used. In
industry, it is more likely that a plug flow regime exists within the semi-batch set up since most
cathodes are flat plates. Thus the two were combined, and the choice of the model incorporates
29
a CSTR model and the semi-batch set up.

Operation runs in semi-batch mode, with CSTR model for the cells
Bulk zinc concentration of the electrolyte remains constant with time
Circulating flow rate in the electrowinning circuit is equal to the exiting flow rate from
leaching
No impurities in the electrolyte – thus only zinc plates the cathodes
Circuit operates independently of pH
4.7.2 User Inputs

Recirculating tank volume
Batch operation time
Limiting current
30
Development of a Spreadsheet Based Zinc Simulator Results and Discussions
Results and Discussion
5 Results and Discussions
The following section will cover some basic examples of the type of trends that
can be generated using the simulator. The examples were chosen based on
which variables are easiest to change during the practicals run at the WCMPF.
31
5.1 Individual Unit Trends
5.1.1 Changing the Ball Mill speed
70
68
66
% passing 0.150mm
64
62
60
58
0.55 0.6 0.65 0.7 0.75 0.8 0.85
Fraction of critical speed
Figure 5.1 Graph showing the effect of mill speed on mill product size
The trend seen in Figure 5.1 is that which is expected theoretically as well as physically. By
increasing the mill speed, the impact velocity and frequency of contact is increased. This results
in better breakage and thus smaller particles.
For a mill of this size, the optimum speed is reported to be around 72%. As can be seen in
Figure 5.1, there is still an increase in the amount of fines being produced above this optimum.
The optimum running of a mill is based on the operating costs and efficiency (defined as energy
required per mass feed). Increasing the speed above the optimum has very little or no effect on
the efficiency (Wills, 1988).
Figure 5.1 shows no sign of levelling off at much higher speeds (namely above 1). This however
will not happen in reality. The critical speed of the mill will cause all the balls to adhere the sides
of the mill and no breakage will take place. The reason for the constant increase seen in the
simulator would be due to the directly proportional relationship between the rate of breakage
and the mill speed.
32
5.1.2 Changing the superficial gas velocity during flotation
70.00
68.00
66.00
64.00
Recovery (%)
62.00
60.00
58.00
56.00
54.00
52.00
50.00
1 1.5 2 2.5 3 3.5 4
3
Air Flowrate (m /s)
Figure 5.2 Graph showing the effect of volumetric gas flowrate on flotation recovery
Figure 5.2 shows how the volumetric flow rate of air affects the flotation performance. The
general increase in recovery as the flowrate increases is expected and will be seen by the students
during their completion of the practical. It must however be noted that this model in the
simulator is based on pure sphalerite being feed to the system and so grade of recovery is not an
issue. If it were, then one would need to remember that by increasing the flowrate, the froth
stability will be altered and the grade may be severely affected.
When looking from a purely physical point of view, the changing of the flowrate of air will have
a knock-on effect in the flotation cells. It will change the mixing in the cells and will also have
an affect in the bubble size. Here it was assumed that bubble size remains at a constant and thus
the result seen in Figure 5.2 is a purely theoretical result. It also implies that if you were to
continually increase the flowrate of air, the recovery would continually increase to some
maximum. However, at some point the air flowrate would be so great that it would physically
cause the flotation medium to “spew” out of the top.
33
5.1.3 Changing the acid flowrate through the leaching tank
100.00
95.00
90.00
85.00
Recovery (%)
80.00
75.00
70.00
65.00
60.00
0 2.5 5 7.5 10 12.5 15 17.5 20 22.5
Flowrate of Acid (ton/hr)
Figure 5.3 Graph showing the effects seen when changing the mass flow rate of acid
Referring to Figure 5.3 above, it is obvious that the zinc recovery increases steeply at first.
Recovery shoots from less than 66% to just under 90% with an increase of acid flowrate of 5tph.
The steep curve then levels off and it takes a further increase of 12tph of acid to shift recovery
up to 95%.
This behaviour is an expected result as recovery can never reach 100%, but will instead tend
towards that unachievable figure.
Average zinc recoveries in industry are in the region of 88.5% (James et al., 2000). Thus it is clear
that the model is performing satisfactorily in terms of recovery results. A recovery of 88.5% is
about optimum for this scenario, achieving the best recovery with minimum acid usage.
Whether the acid flowrates to the leaching section are reasonable or not is unclear. James and his
co-workers (2000) do not report mass flowrates, but rather acid concentration. Also, the effect
of air sparging is not accounted for.
34
5.1.4 Changing the current used in the Electrowinning cells
100
90
80
Recovery (%)
70
60
50
40
500 1000 1500 2000 2500 3000
Current (A)
Figure 5.4 Graph showing the effect when the current in the electrowinnig cell is altered
Current applied to the electrowinning circuit is a major variable. It is the source of greatest
expense to the unit operation, and ideally should be kept as low as possible. The other important
factor is the length of time that the batch runs for. Keeping batch time constant, and varying
current, Figure 5.4 was generated using LIMN. Low current resulting in low operating costs also
results in unacceptably poor recovery. Extremely high currents achieve acceptable recovery.
Although not illustrated here, the longer it takes to run the batch, the better the recovery will be
– but this also means that a lower current will be applied for a longer period of time, also
escalating cost.
Physically, what occurs in the cells when current density is increased is more electrons are made
available to the electrolyte. Thus a greater number of Zn2+ ions can receive electrons during the
batch, and thus more zinc is deposited on the cathode.
35
5.2 Circuiting Effects

Following is a description of how the system performs when two of the main overall system
inputs are changed. On a mineral processing plant, the most obvious and most possible input
change will be in the fresh feed to the plant. The fresh feed is most likely to change in two areas,
namely its feed rate and the size distribution of the feed.
5.2.1 Effect of changing the Fresh Feed to the system
80
70
Elemental Zinc recovery (%)
60
50
40
30
20
0 10 20 30 40 50 60 70 80
Fresh Feed (tons/hr)
Figure 5.5 Overall process efficiency as a function of fresh feed
By leaving the simulator at a fixed point and changing the initial fresh feed to the system, Figure
5.5 can be drawn up. The overall recovery was based on the mass of zinc leaving the
electrowinning plant divided by the elemental zinc entering into the plant (again assumed pure
ZnS). This gives an indication of how well the plant is running. The trend seen in Figure 5.5 is
exactly as one would expect to see. By increasing the flowrate, the residence times in all of the
units are decreased and hence they are less effective. The exception of course is the leaching
tank as it was modelled at equilibrium; it is however affected as the leaching acid will remain
constant whilst the input will increase.
The milling circuit will be the most dramatically affected by the fresh feed increase as its recycle
stream will become larger with the cyclone having to remain at the given cut size while the mill
produces courser material. Individual units will all react differently and they can each be studied.
36
5.2.2 Effect of changing the Fresh Feed size distribution
45.5
45
Elemental Zinc Recovery
44.5
44
43.5
43
42.5
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Fraction passing 0.150mm
Figure 5.6 Overall process efficiency as a function of the fresh feed size distribution
The physical results of varying the size of the particles fed to the system are illustrated in Figure
5.6. Smaller particles result in better elemental zinc recovery.
The reason why this occurs is quite complex, and is a combination of effects from all unit
operations.
In this model, the leaching and electrowinning models are independent of particle size, so their
impact can be discounted.
The contribution of the mill to this recovery is simple. The mill breaks particles into smaller
particles. If the particles are already small, then the mill does not add much value to the system,
and in fact is wasting a great deal of energy. The cost of that energy is high, and so the mill could
be doing more harm than good by operating in a system that already has exceptionally fine
particles.
Flotation is very sensitive to particle size. When particles are too small, they will not float. Thus
looking at the system solely from the point of view of the flotation operation, it is surprising that
recovery increases with decreasing particle size. However, the roasting circuit counters the
flotation circuit. Small particles will practically be completely converted in the roaster. Thus it
appears that the improved roaster performance compensates for the drop in flotation
performance.
37
Development of a Spreadsheet Based Zinc Simulator Concluding recommendations
Concluding Recommendations
6 Concluding recommendations
The following section will cover some of the possibilities that exist as tasks for
students to complete before performing the practicals. It will also reiterate
some of the key findings and conclude on the overall project.
38
6.1 Student Exercises
6.1.1 Assignment 1 – Fresh Feed Exercise

This exercise sets out to investigate the effect that changing the fresh feed has on the rest of the
system as that seen and discussed in Section 5.2.1 and Figure 5.5
The following tasks are advised for the students to complete

Adjust the Fresh feed to the system for a set range of variables decided before
Whilst running the simulator, take note of the following:
o The mill feed (i.e. the recycling that is taking place)
o The overall recovery of the system
o The recovery changes in each of the unit operations
o Load on each of the units
Propagation of error in the system
With these tasks in mind, the students running the simulator will hopefully learn some valuable
lessons before they move on to completing the practicals. Some of the key learning outcomes
from this assignment are discussed below.
Knock-on Effects
Hopefully students will be able to see how all the units interact with each other and what the
effects of this interaction are. The most noticeable effect in this case would be that of the mill.
Here it can be seen that by increasing the mill’s feed and leaving the cyclone at the same cut size,
the recycle rate becomes very large. The reason for this lies in the fact that with more feed, the
mill is less efficient and will produce courser material. The cyclone will then just continue to
send the course material back to the mill and so increase its feed. This could carry on
indefinitely and the cause major plant problems.
Sensitivity of Unit operations

When looking at how the feed rate affects the individual units, it can be seen that some units will
react differently to the feed increase or decrease. This will allow students to explore a sensitivity
analysis of the unit operations. Knowing how sensitive a unit is, the students will be introduced
to the concept of control, and can see which units require a stricter control regime.
Mathematical Models
When watching how each of the unit operations react with the changing feed rate, the students
are forced to look closer at the mathematical basis of the models. This will then hopefully allow
them to understand what assumptions were made and how the model actually works. With this
in mind it is hoped that when they then move on to do the practicals, they will be able to easily
predict what would happen if they were to do something during the practical. This will hopefully
also give students grounding so that when they are writing there practical reports, they will have
a feel as to what is correct.
39
6.1.2 Assignment 2 – Optimisation Exercise

This exercise is something that is completely different to what students will experience at the
practical level, but will expose them to very necessary skills. This task is broken up into two
different types of optimisation; namely that of economic optimisation and that of performance
optimisation.
For the completion of the economic evaluation and optimisation, the following tasks should be
looked into and completed by the student.
Attempt to optimise the plant taking these key cost indicators into account
o Cost of mill power consumption
o Cost of acid required in leaching unit
o Cost of electricity usage in the electrowinning unit
What happens to the system if acid is no longer paid for
o Is the system overly sensitive to it?
o Is it a key saving not having to pay for it?
The second part of the optimisation task would be to optimise one particular unit with respect to
performance. Once this is complete, the following could be investigated.
What is the effect of the optimisation on downstream units?
Some of the key learning outcomes learnt from this task are discussed below.
Introduction to Process Economics

Students will be able to see how a plant cost is broken down and will be able to see the
sensitivity of the plant to certain variables. This will again stress the importance of control, but
from an economic view point. Completing this task will also give students experience in
searching for economic indicators. Students will also be able to see the benefits involved when
using by-products to reduce costs. The example used here is the use of SO2 from the roaster in
an acid plant to produce H2SO4 for the leaching unit.
System optimisation versus unit optimisation

By optimising one unit and seeing negative downstream effects, the students will be able to see
how it may be worth running a unit less efficiently so that the system will run better as a whole.
An example of this in the system would be the flotation and roasting setup. The flotation circuit
runs better for larger size particles, but the larger size particles will cause a drop in the
conversion of the roaster.
6.2 Possible improvements to the Simulator

The Zinc Beneficiation Simulator produces satisfactory results and should easily be implemented
as a learning tool in its current form.
However, Fletcher and Robinson feel that there is some scope to improve the simulator. The
following improvements are suggested:
40
Convert the leaching model from thermodynamic equilibrium to mass transfer limited.
This will allow the user greater control over the simulator as the effects of variables such
as tank volume, residence time, zinc concentration, leach tank temperature and sulphuric
acid concentration can be investigated.
Implement the plug flow model in the electrowinning circuit. This is not to say that the
current CSTR model should be discarded – rather both can run simultaneously, and
comparative results can be produced
6.3 Acknowledgements
We would like to thank the following individuals who were indispensable to the successful
completion of this project.
Dr. Dave Deglon for his excellent supervision. A list of everything you did for us would
take up many pages – so we won’t kill off any more trees than necessary by including it.
Emile Scheepers of Stellenbosch University who gave us a tour of the WCMPF and
was more than happy to answer all of our questions.
Peter Lekoma for arranging permission for us to make use of the UCT Chemical
Engineering bakkie to travel to Stellenbosch.
Both Emile and Peter for providing us with the 3rd year student practical reports of the
previous couple of years. The reports were extremely useful in getting an idea of what it
was that the students actually did.
41
Development of a Spreadsheet Based Zinc Simulator References
7 References
Alfaro, P., and Castro, S., 1998, The zinc refinery of IMMSA in San Luis Potosí,
Mexico, in Zinc and Lead Processing, Dutrizac et
al. (eds), Canadian Institute of Mining,
Metallurgy and Petroleum, pp 71
Blair, J.C., 1985, Fluid Bed Roasting, in SME Mineral

Processing Handbook, Vol. 1, American
Institute of Mining, Metallurgical, and
Petroleum Engineers, Inc., New York, pp
12-8
Dodson et al., 1999 Flash roasting of sulphide concentrates – Paper

presented at TMS, available at
http://www.torftech.com/publications.htm,
last accessed 09/11/2001
Fichera M.A. and Chudacek M.W., 1992 Batch Cell flotation models – A Review, Minerals
Engineering, Vol. 5, No.1, pp. 41-55, Great
Britain
Flint L.R. and Howard W.J., 1971 Collision efficiency of small particles with spherical
air bubbles, Chem. Eng. Sci., 26, pp 1155-
1168
Fogler, H.S.,1999, Elements of Chemical Reaction Engineering, 3rd

Ed. Prentice Hall International, New Jersey,
pp787-789
Frew, J.A., and Trahar, W.J., 1980 Roughing and Cleaning Flotation behaviour and the
Realistic Simulation of Complete Plant Performance,
Int. J. Miner. Process., Vol. 9, pp 101-20
Graf, G.G., Zinc, in Ullmann’s Encyclopedia of Industrial

Chemistry, Vol. A28, pp514
Gupta, C.K. and Mukherjee, T.K., 1990, Hydrometallurgy in Extraction Processes, Vol 1,
CRC Press, Boston, pp 18, 62
42
Herbst J.A., Rajamani K. and Kinneberg D.J.,

1977 ESTIMILL–A program for Grinding Simulation
and parameter estimation with linear models,
Metallurgical Engineering Department,
University of Utah, Salt Lake City, UT
Jackson, E., 1986 Hydrometallurgical Extraction and Reclamation,

Ellis Horwood Limited, West Sussex, pp 48
James, S.E., Watson, J.L., and Peter, J., 2000, Zinc production - A survey of existing smelters and
refineries, in Lead-Zinc, Dutrizac et al (eds),
The Minerals, Metals & Materials Society,
pp205 – 225
Jameson G.J., Nam S., and Moo Young M., 1977 Physical factors affecting recovery rates in flotation,
Miner. Sci. Eng., 9:103-118
Kelsall D.F., 1961, Application of probability in the Assessment of

flotation systems, Trans. Instn. Min. Metall., 70,
pp. 191-204
Morrel S and Man Y.T., 1997, Using modelling and simulation for the design of full
scale Ball mill Circuits, Minerals Engineering,
Vol 10, No 12, pp. 1311 – 1327
Picket et al., 1985 Descriptions of Specific Concentrators, in SME

Mineral Processing Handbook, Vol. 2,
American Institute of Mining, Metallurgical,
and Petroleum Engineers, Inc., New York,
pg 15-14 to 15-48
Pletcher, D., and Walsh, F.C., 1993 Industrial Electrochemistry, 2nd ed., Blackie
Academic and Professional, Cambridge, pp
96
Plitt L.R., A Mathematical model of the hydrocyclone classifier,

CIM Bull. 69, pp 114
Rajamani K., and Herbst J.A., 1991, Optimal control of a ball mill grinding circuit -I.
Grinding circuit modelling and dynamic simulation,
Chemical Engineering Science, Vol 46, No.
3, pp 861-870
43
Reay D. and Ratcliff G.A., 1973 Removal of fine particles from water by dispersed air
flotation – Effects of bubble size and particle size on
collection efficiency. Can. J. Chem. Eng., 51,
pp178-185
Reuter, M.A. and Lans, S.C., (eds), 2001 Zinc College Course Notes, Delft University of
Technology, Delft, pp 72
Seader, J.D. and Henley, E.J., 1998 Separation Process Principles, John Wiley &
Sons, Inc., New York, 1998, pp 234 -236
Tarr D.T., 1985, Hydrocyclones, in SME Mineral Processing

Handbook, Volume 1, pp 3D-10 – 3D-23
Thorne G.C., 1976, Modelling of industrial sulphide flotation circuits, in

Flotation – The A.M. Gauding Memorial
Volume, Vol. 2, American Institute of
Mining, Metallurgical, and Petroleum
Engineers, Inc., chapter 26, 1976, pg 725-
750.
Wills B.A., 1988, Mineral Processing Technology: An introduction to

the Practical aspects of Ore treatment and mineral
recovery, 4th Edition, Pergamon Press, Great
Britian
Wiseman D., 1999 Examples that ship with the LIMN Installation,
File name: mill circuit.xls
Yovanovic A.P. and Moura H.P., 1993, A New Macrophenomemological Concept of

Comminution in Ball Mills, XVIII International
Mineral Processing Congress, Sydney
Zuniga H.G., 1935, Flotation recovery is an exponential function of its

rate, Boln. Soc. Nac Min., Santiago, 47, pp
83-6
44
Development of a Spreadsheet Based Zinc Simulator Appendices
8 Appendices
Supplementary Calculations
To determining diffusivity of oxygen through the shell of the particle in the roaster, the
following equations were used:
Diffusivity for the diffusion of gas pairs of non-polar, nonreacting molecules is:
1
 1
3 1  2
0.001858T  2
+ 
 MA MB 
D AB = Equation i
Pσ AB
2
ΩD
Where MA and MB are the molecular weights of A and B respectively; T is absolute temperature
(in K); P is absolute pressure (atm); σAB is the “collision diameter” (Ǻ); ΏD is the “collision
integral for molecular diffusion (dimensionless).
The gases in the roaster are assumed to be air and sulphur dioxide. Thus MA = Mair =
28.84g/gmol; and MB = MSO2 = 64g/gmol
Pressure and temperature are chosen by the user.
σ A +σ B
σ AB = Equation ii
2
σA and σB are constant for a particular gas, and are available in literature. ΏD can be cross-
referenced in a table available in literature as it is based on the Lennard-Jones constants.
Other required variables in the calculation of the mass transfer through the shell are taken as
typical values. The typical values chosen are present in the spreadsheet.
45
Method Used to Vary Size Distribution

The method by which feed size distribution was varied was to choose a size distribution. This
was input into the LIMN spreadsheet. It is unlikely that the chosen size fractions will add up to
1, thus they must first be normalised before proceeding.
Once the size distribution is normalised, a simulator run is performed and the percentage of
material reporting to the size fractions below 150 microns was noted.
Changing the feed size distribution is a matter of shifting the size distribution curve to the left or
the right, and then normalising the curve again, before repeating the run Normalisation also has
the effect of increasing or decreasing the magnitude of the peak of the bell curve.
0.2
Shifted Spread
0.18 Original Spread
0.16
0.14
Mass Fraction
0.12
0.1
0.08
0.06
0.04
0.02
0
1 2 3 4 5 6 7 8 9 10 11 12
Size Fraction
Figure 8.1 Chart illustrating effect of shifting particle size distribution to the right
46
Simulator Spreadsheets
47
User Input Options
Instructions: Only alter cells highlighted in blue on this sheet
Major results are available on the sheet tagged "Results"
Important: Do NOT alter any values on ANY other sheet.
This will lead to unpredicable results
Fresh Feed Distrubution

Fraction
Size Fraction
Size Fraction (#) Feed rate (tph) Passing (in
(mm)
Feed)
1 >0.850 3.00 0.10
2 0.850 - 0.600 4.80 0.16
3 0.600 - 0.425 5.40 0.18
4 0.425 - 0.300 4.80 0.16
5 0.300 - 0.212 3.00 0.10
6 0.212 - 0.150 2.40 0.08
7 0.150 - 0.106 1.80 0.06
8 0.106 - 0.075 1.50 0.05
9 0.075 - 0.053 1.50 0.05
10 0.053 - 0.038 0.90 0.03
11 0.038 - 0.025 0.60 0.02
12 <0.025 0.30 0.01
Total 30.00 1.00
Mill Input Variables Cyclone Input Variables
Mill diameter [m] 3.048 Cut Size [mm] 0.1
Mill length [m] 3.66 Alpha 4
Mill speed fraction of critical 0.72
Fractional filling of mill with balls 0.35
Flotation Input Variables Roaster Input Variables
No. of cells 15 Partial Pressure of Oxygen [atm] 0.8
Volume of one cell [m 3] 2 Temperature [K] 1173
Height of cell [m] 2.9 Roast retention time [s] 0.05

3
Volumetric air flow rate [m /s] 2.5
Bubble size [mm] 1
Fraction solids in feed 0.3
W/L 0.286
Leach Input Variables Electrowinning Input Variables
Solvent flow rate (H 2SO4) [t/hr] 10 Recirculating tank volume [m^3] 0.3
Actual value used (taking Smin into account) 10.00 Time to run batch [hr] 120
Stoichiometric requirement: 21.06 Limiting current [A] 2000

Results and Stream Data
Stream Data Other Results
Particle Mass Distributions (tph in each size fraction) Milling
% Passing Less Than
Fresh Mill Cyclone Cyclone Flotation Flotation
Size Size [mm] Mill Feed Roast Product 150 microns from mill
Feed Discharge Underflow Overflow Conc Gaunge
Number discharge
ZnS [tph] 1.35 64.91
1 0.850 3 3.45 0.45 0.45 0.00 0.00 0.00
2 0.600 4.8 5.90 1.10 1.10 0.00 0.00 0.00

ZnO [tph] 17.48 Flotation
Percentage ZnS
3 0.425 5.4 7.65 2.25 2.25 0.00 0.00 0.00 Recovered (floated)
Leach Product 63.34
4 0.300 4.8 8.65 3.86 3.85 0.00 0.00 0.00
5 0.212 3 8.46 5.47 5.46 0.01 0.01 0.00

Zn [tph] 12.78 Roasting
ZnS to ZnO Mass
6 0.150 2.4 10.63 8.59 8.23 0.35 0.35 0.00 Conversion (%)
Electrowinning Product 92.84
7 0.106 1.8 8.38 8.87 6.58 2.29 2.29 0.00
8 0.075 1.5 4.08 6.61 2.58 4.02 3.97 0.05

Zn [tph] 10.38 Leaching
Zinc Recoved (%)
9 0.053 1.5 2.48 5.54 0.98 4.57 4.27 0.29
91.01
10 0.038 0.9 1.27 4.31 0.37 3.94 3.23 0.71
11 0.025 0.6 0.75 3.46 0.15 3.31 2.09 1.22 Electrowinning

Zn Recovery (%)
12 <0.025 0.3 0.44 11.36 0.14 11.23 2.61 8.62
Solids [t/h] 30.00 62.14 61.86 32.14 29.72 18.83 10.90 81.19
Flowsheet
Mill Simulator
tph Solids Mill Variables
size mm Feed Discharge mill diameter [m] 3.05
0.850 3.448 0.448 mill length [m] 3.66
0.600 5.901 1.101 mill speed fraction of critical 0.72
0.425 7.651 2.251 fractional filling of mill with balls 0.35
0.300 8.655 3.855 mill feed flow rate (solids) [tph] 62.14
0.212 8.456 5.468 % solids in mill 0.75
0.150 10.633 8.586 water flow rate to mill [tph] 20.71
0.106 8.382 8.872 slurry volume [m^3] 4.01
0.075 4.082 6.607 slurry density [t/m^3] 2.50
0.053 2.476 5.541 holdup [tons] 0.01
0.038 1.269 4.307 residence time in mill [hr] 0.0001209
0.025 0.750 3.463 discharge rate [hr-1] 8.273 Selection Function
<0.025 0.436 11.362 ball density [kg/m^3] 7380.000 k 0.05
Total 62.139 61.862 Power Expression 5.782 pk 0.00003
p
Breakage Function (Normalised) - JKMRC Model using Massive Sulphide (Pb,Zn,Cu) Selection Selection
size mm 0.85 0.6 0.425 0.3 0.212 0.15 0.106 0.075 0.053 0.038 0.025 <0.025 Function Function r/d
0.850 0 0 0 0 0 0 0 0 0 0 0 0 0.071987831 55.420418 6.698808 1 42.384
0.600 0.1081 0 0 0 0 0 0 0 0 0 0 0 0.05 38.492907 4.652736 2 28.54985
0.425 0.1442 0.1081 0 0 0 0 0 0 0 0 0 0 0.030491943 23.474471 2.837419 3 19.30694
0.300 0.1472 0.1442 0.1081 0 0 0 0 0 0 0 0 0 0.018595172 14.315645 1.730369 4 13.00515
0.212 0.1253 0.1472 0.1442 0.1081 0 0 0 0 0 0 0 0 0.011321014 8.715575 1.053474 5 8.771311
0.150 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0 0 0 0 0 0.006903994 5.3150963 0.642449 6 5.924163
0.106 0.0805 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0 0 0 0 0.004210324 3.2413523 0.391791 7 3.995546
0.075 0.06444 0.0805 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0 0 0 0.002567619 1.9767026 0.238929 8 2.698601
0.053 0.05076 0.06444 0.0805 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0 0 0.001565834 1.2054701 0.145708 9 1.820068
0.038 0.03958 0.05076 0.06444 0.0805 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0 0.000963974 0.7421229 0.089702 10 1.247888
0.025 0.03103 0.03958 0.05076 0.06444 0.0805 0.1006 0.1253 0.1472 0.1442 0.1081 0 0 0.000563929 0.4341454 0.052476 11 0.776055
<0.025 0.10829 0.13932 0.1789 0.22966 0.2941 0.3746 0.4752 0.6005 0.7477 0.8919 1 0 0.00026233 0.2019567 0.024411 12 0.369595
Cyclone Simulator
tph Solids Cyclone Model Parameters Average
size mm Feed Underflow Overflow d/d50 fraction feed in underflow Size (mm)
0.850 0.448 0.448 0.000 8.500 1.000 d50c - cut size [mm] 0.100 0.85
0.600 1.101 1.101 0.000 7.141 1.000 alpha 4.000 0.714
0.425 2.251 2.251 0.000 5.050 1.000 0.505
0.300 3.855 3.855 0.000 3.571 1.000 0.357
0.212 5.468 5.456 0.012 2.522 0.998 0.252
0.150 8.586 8.233 0.353 1.783 0.959 0.178
0.106 8.872 6.582 2.290 1.261 0.742 0.126
0.075 6.607 2.582 4.025 0.892 0.391 0.089
0.053 5.541 0.976 4.566 0.630 0.176 0.063
0.038 4.307 0.369 3.938 0.449 0.086 0.045
0.025 3.463 0.150 3.313 0.308 0.043 0.031
<0.025 11.362 0.136 11.226 0.125 0.012 0.013
Total 61.86 32.14 29.72
Flotation Simulator
flotation variables Flotation recovery and rates
Solids in Solids in
Solids Mass In Concentrate Tailings Average
Size [mm] k [funny unit] R [tons/hr] [tons/hr] [tons/hr] Size (mm)
No. of cells 15 0.850 308.567 1.000 0.00 0.00 0.00 0.85
Total Residence time [hr] 0.484 0.600 237.628 1.000 0.00 0.00 0.00 0.714
3
volume of one cell [m ] 2 0.425 141.295 1.000 0.00 0.00 0.00 0.505
height of cell [m] 3.5 0.300 84.014 1.000 0.00 0.00 0.00 0.357
3
volumetric air flow rate [m /s] 2.5 0.212 49.867 1.000 0.01 0.01 0.00 0.252
2
cross-sectional area [m ] 0.571429 0.150 29.651 1.000 0.35 0.35 0.00 0.178
-1
Superficial Air velocity [m.s ] 4.375 0.106 17.631 0.999 2.29 2.29 0.00 0.126
-1
Superficial Air velocity [m.hr ] 262.5 0.075 10.483 0.987 4.02 3.97 0.05 0.089
Bubble size [mm] 1 0.053 6.233 0.936 4.57 4.27 0.29 0.063
fraction solids in feed 0.3 0.038 3.743 0.820 3.94 3.23 0.71 0.045
total feed in [m3/hr] 61.92 0.025 2.131 0.631 3.31 2.09 1.22 0.031
<0.025 0.550 0.232 11.23 2.61 8.62 0.013
29.72 18.83 10.90
Roasting Simulator
Average size Average Initial Average Initial Critical Final Radius
Roasting Variables Size [mm] [mm] Radius [mm] Radius [m] Time [s] hours t/tc [m] X
Density of ZnS [kg/m^3] 3000 0.850 0.850 0.425 0.00042500 12.465874 0.003463 0.004011 0.000333862 0.51523157
Partial Pressure of Oxygen [atm] 0.8 0.600 0.714 0.357 0.00035707 8.7994404 0.002444 0.005682 0.000277742 0.52939356
Temperature [K] 1173 0.425 0.505 0.252 0.00025249 4.3997202 0.001222 0.011364 0.000191308 0.5650099
concentration of O2 at surface [kg/m^3] 0.26596343 0.300 0.357 0.179 0.00017854 2.1998601 0.000611 0.022729 0.000130134 0.61274077
0.212 0.252 0.126 0.00012610 1.097342 0.000305 0.045565 8.66672E-05 0.67531019
Mair 28.84 0.150 0.178 0.089 0.00008916 0.548671 0.000152 0.091129 5.59098E-05 0.75344494
MSO2 64 0.106 0.126 0.063 0.00006305 0.2743355 7.62E-05 0.182259 3.39054E-05 0.8444746
sigma (collision diameter) 3.954 0.075 0.089 0.045 0.00004458 0.1371677 3.81E-05 0.364517 1.78269E-05 0.93606111
omega (collision integral) 0.8386 0.053 0.063 0.032 0.00003152 0.0685839 1.91E-05 0.729034 1.68068E-05 0.84845474
Diffusivity [m^2/s] 0.000127686 0.038 0.045 0.022 0.00002244 0.0347492 9.65E-06 1 0 1
0.025 0.031 0.015 0.00001541 0.0163911 4.55E-06 1 0 1
volume fraction of ZnS 1 <0.025 0.0125 0.00625 0.00000625 0.0026959 7.49E-07 1 0 1
tortuosity 1.5
pellet porosity 0.4
constriction factor 0.8
Effective Diffusivity 2.72398E-05
cactus 69015218.8
Roast time [s] 0.05
Size [mm] feed [t/hour] product (ZnO) product (ZnS)

0.850 0.00 0.00 0.00
. 0.600 0.00 0.00 0.00
0.425 0.00 0.00 0.00
0.300 0.00 0.00 0.00
0.212 0.01 0.01 0.00
0.150 0.35 0.27 0.09
0.106 2.29 1.93 0.36
0.075 3.97 3.72 0.25
0.053 4.27 3.63 0.65
0.038 3.23 3.23 0.00
0.025 2.09 2.09 0.00
<0.025 2.61 2.61 0.00
18.83 17.48 1.35
ZnS 1.35
ZnO 17.48
Leaching Simulator
Leaching Variables
Fa (ZnS) [t/hr] 1.35
Fb (ZnO) [t/hr] 17.48
S (H2SO4) [t/hr] 10.00 10.00 stoichiometric requirement 21.05833552

R 0.67
W (Wash factor) 11.12
N (Number of tanks) 1.00
Xn 1.75
Y1 1.75
Weight percent solvent in underflow 40.00
Smin 0.90
Percent Recovery of soluble material 91.01
mass Zn leaving leach [t/hr] 12.78

Mass H2SO4 leaving [t/hr] 9.10
total mass flow [t/hr] 21.88
Volumetric flow rate of electrolyte [m^3/hr] 21.88
concentration of Zn [t/m^3] 0.58
concentration of Zn [mol/m^3] 8932.12
Electrowinning Simulator
Electrowinning Variables
Volumetric flow rate in [m^3/hr] 21.88
tank volume [m^3] 0.3

residence time of tank [hr] 0.01371
time to run batch [hr] 120
limiting current [A] 2000
n 2
Faraday's (F) [A.hr/mol] 2.68E+01
c infinity [mol/m^3] 8932.12
kLA 4.18E-03
fractional conversion (Xa) 0.81186
Mass Zn In (tph) 12.78

Mass Zn Out (tph) 10.3757

Zinc Beneficiation Simulator

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Zinc Beneficiation Simulator

Uploaded by

Copyright:

Available Formats

CHE445Z

CHEMICAL ENGINEERING PROJECT

Development of a spreadsheet-based simulator for a

Mineral Processing Research Unit at the University of

PREPARED BY: MICHAEL FLETCHER (FLTMIC007)

PREPARED FOR: DR. D.A. DEGLON

4.2.1 Model Selection--------------------------------------------------------------------------------------------- 23

Figure 1.1 Schematic diagram showing student learning----------------------------------------------- 3

Symbol Description Units

Symbol Description Units

Symbol Description Units

Ri Fractional Recovery of size fraction i

Introduction and Background

Operation of a Continuous Ball Mill in closed circuit with a Screen Classifier

1.2 Practical Procedure

1. Environmental impact, legislation and reasons for recycling

This project sets out to simulate the processes that

certain unit operations affects units downstream,

1.3 LIMN – The Flowsheet Processor

1.4 Aims and Objectives

1. The simulator is to be an educational tool that enhances student learning

Review of the Hydrometallurgical Route

2 Review of the Hydrometallurgical Route

The hydrometallurgical route for the beneficiation of zinc begins with

2.1 Communition and Flotation

2.2 RLE (Roasting, Leaching, Electrowinning) Process Review

2.2.1 Roasting of Zinc Bearing Ore

An alternative to conventional fluidised-bed roasting is flash roasting, using the TORBED

Spent electrolyte and

Review of Mathematical Models

3 Review of Mathematical Models

3.1 Modelling of the Ball Mill

The process is represented either chemically or physically using Microphenomenological

3.1.1 Population Balance Model

Table 3.1 Table showing the Breakage Appearance Distribution matrix

bij Breakage from size 'j'

3.1.2 Modelling the rate of breakage

P ∝ DM3 Lρ b J B Equation 3.6

3.1.3 A simplified Ball Mill Model

3.2 Hydrocyclone Modelling

14.8Dc0.46 Di0.6 Do1.21e 0.063V

Figure 3.2 Reduced efficiency curve based on partition curve

C1 = a11 d 50 + a12 Equation 3.13

Table 3.2 Table showing Rajamani and Herbst (1991)

3.3 Flotation circuit Modelling

3.3.1 First order Flotation Kinetics

3.3.2 Estimation of the First order rate Constant

Pcoll = Pc Pa (1 − Pd ) Equation 3.18

Here it must be noted that n ≠ m

3.3.3 Second order Flotation Kinetics

3.4 Roasting Models – Kinetic and Thermodynamic

3.4.1 Roasting Thermodynamics

2FeS + 7/2O2 → Fe2O3 + 2SO2 Equation 3.23

2FeS2 + 11/2O2 → Fe2O3 + 4SO2 Equation 3.24

3.4.2 Bubbling Bed Model

3.4.3 Shrinking Core Model

3.5 Zinc Leaching Models

3.5.1 Thermodynamics of Leaching

ZnO (s) + 2H+(aq) ⇔ Zn2+(aq) + H2O (Hayes, 1985) Equation 3.27

ZnO.Fe2O3 + 8H+ ⇔ Zn2+ + 2Fe3+ + 4H2O Equation 3.28

3.5.2 Leaching Kinetics – Mass-Transfer Limited Model

3.5.3 Thermodynamic Equilibrium Model

Fsol = Yout(S – RFinsol) + XoutRFinsol Equation 3.30