Energy, Mines and Énergie, Mines et

Resources Canada Ressources Canada

CANMET
Canada Centre Centre canadien
for Mineral de la technologie
and Energy des minéraux
Technology et de l'énergie

SPOC
Simulated
Processing of
Ore and Coal

Chapter 1 Summary
S.F,H
622(21)
C212sp
85-1/1E
c.,
,9Ee
(al)
cal„2sr.
85- ///é
e

■If
BIBLIOTHÈQUE
CANMET
LIBRARY,

NOV 1 1905
555 rue BOOTH 5161‘
',OTTAWA, CANADA KlA OG11
14' Energy, Mines and
Resources Canada
Énergie, Mines et
Ressources Canada

CANMET
Canada Centre Centre canadien
for Mineral de la technologie
and Energy des minéraux
Technology et de l'énergie

The

SPOC Manual
Chapter 1 SUMMary

D. Laguitton

Editor: D. Laguitton

BIBLIOTHÈQUE CANMET LIBRARY

Mineral Processing Plant Simulation

Minerais Research Program
Mineral Sciences laboratories

SP85-1 /1E
 Minister of Supply and Services Canada 1985
Available in Canada through
Authorized Book Agents
and other bookstores
or by mail from
Canadian Government Publishing Centre
Supply and Services Canada
Ottawa, Canada K1A 0S9
Catalogue No.: M38-16/1-1985E Canada: $7.00
ISBN 0-660-11855-6 Other Countries: $8.40
Price subject to change without notice
Disponible en français
14' Energy, Mines and
Resources Canada
Énergie, Mines et
Ressources Canada

CANMET
Canada Centre Centre canadien
for Mineral de la technologie
and Energy des minéraux
Technology et de l'énergie

The

SPOC Manual
Chapter 1 SUMMary

D. Laguitton

Editor: D. Laguitton

BIBLIOTHÈQUE CANMET LIBRARY

Mineral Processing Plant Simulation

Minerais Research Program
Mineral Sciences laboratories

SP85-1 /1E
 FOREWORD
High energy costs and depleting ore reserves combine to make process evaluation and optimization a challenging
goal in the 80's. The spectacular growth of computer technology in the same period has resulted in widely available
computing power that can be distributed to the most remote mineral processing operations. The SPOC project,
initiated at CANMET in 1980, has undertaken to provide Canadian industry with a coherent methodology for process
evaluation and optimization assisted by computers. The SPOC Manual constitutes the written base of this meth-
odology and covers most aspects of steady-state process evaluation and simulation. It is expected to facilitate
industrial initiatives in data collection and model upgrading.
Creating a manual covering multidisciplinary topics and involving contributions from groups in universities, industry
and government is a complex endeavour. The reader will undoubtedly notice some heterogeneities resulting from the
necessary compromise between ideals and realistic objectives or, more simply, from oversight. Critiques to improve
future editions are welcomed.

D. Laguitton
SPOC Project Leader
Canada Centre for Minerai and Energy Technology

AVANT-PROPOS

La croissance des coûts de l'énergie et l'appauvrissement des gisements ont fait de l'évaluation et de l'optimisation
des procédés un défi des années 80 au moment même où s'effectuait la dissémination de l'informatique jusqu'aux
concentrateurs les plus isolés. Le projet SPOC, a été lancé en 1980 au CANMET, en vue de développer pour
l'industrie canadienne, une méthodologie d'application de l'informatique à l'évaluation et à l'optimisation des pro-
cédés minéralurgiques. Le Manuel SPOC constitue la documentation écrite de cette méthodologie et en couvre les
différents éléments. Les retombées devraient en être une vague nouvelle d'échantillonnages et d'amélioration de
modèles.
La rédaction d'un ouvrage couvrant différentes disciplines et rassemblant des contributions de groupes aussi divers
que les universités, l'industrie et le gouvernement est une tâche complexe. Le lecteur notera sans aucun doute des
ambiguïtés ou contradictions qui ont pu résulter de la diversité des sources, de la traduction ou tout simplement
d'erreurs. La critique constructive est encouragée afin de parvenir au format et au contenu de la meilleure qualité
possible.

D. Laguitton
Chef du projet SPOC,
Centre canadien de la technologie des minéraux et de l'énergie

iii
 ABSTRACT

This first chapter of the SPOC Manual is intended as a general introduction to all the other chapters in which the
mineral process evaluation and optimization methodology is presented. This chapter also provides a brief description
of available software for mainframe and microcomputers and provides information on its distribution by the Technology
Information Division, CANMET.

RÉSUMÉ
Ce premier chapitre du manuel SPOC est un guide général qui présente les autres chapitres où l'on traite de la
méthodologie d'évaluation et d'optimisation des procédés minéralurgiques. Le lecteur y trouvera aussi une descrip-
tion sommaire des programmes de calcul développés pour ordinateurs centraux et pour micro-ordinateurs, ainsi que
la façon de les obtenir de la Division de l'information technologique du CANMET.

ACKNOWLEDGEMENTS
The SPOC project has benefited from such a wide range of contributions throughout the industry, the university, and
the government sectors that a nominal acknowledgement is bound to make unfair omissions. The main groups that
contributed are: the various contractors who completed project elements; the Industrial Steering Committee members
who met seven times to provide advice to the project leader; the various users of project documents and software who
provided feedback on their experience; the CANMET Mineral Sciences Laboratories staff members who handled the
considerable in-house task of software development, maintenance, and documentation; the EMR Computer Science
Centre staff who were instrumental in some software development; and the CANM ET Publications Section. Inasmuch
as in a snow storm, every flake is responsible, their contributions are acknowledged.
Special credits are due to Frederic Flament, Ralph Pilgrim, Jean Leung and Wayne Cameron as in-house contributors
to the manuals and software; to Louis Sirois for his constant support of the project since its conception; and to Joan
Stewart for her intensive contribution to the word processing of the manuals.

y
MIIMMMMI
 CONTENTS

THE SPOC MANUAL

FOREWORD iii
AVANT-PROPOS iii
ABSTRACT
RÉSUMÉ
ACKNOWLEDGEMENTS

1. INTRODUCTION 1
2. THE SPOC METHODOLOGY, STEP BY STEP 2
2.1 Sampling 3
2.1.1 Documents 3
2.1.1.1 Chapter 2 — Sampling methodology 3
2.1.1.2 Chapter 2.1 — SAMBA computer program 3
2.1.1.3 Chapter 2.2 — Grinding circuit sampling 3
2.1.2 Software 3
2.1.2.1 SAMBA (2.1) 3
2.1.2.2 GYFORM (8) 4
2.1.2.3 TIMESR (8) 4
2.1.2.4 STATPD (8.) 4
2.1.2.5 ANOVA (8) 4
2.1.2.6 URANDT (8) 4
2.1.2.7 FADESN (8) 4
2.2 Material Balance 4
2.2.1 Documents 4
2.2.1.1 Chapter 3 — Material balance 4
2.2.1.2 Chapter 3.1 — BILMAT computer program 4
2.2.1.3 Chapter 3.2 — MATBAL computer program 4
2.2.2 Software 5
2.2.2.1 MATBAL (3.2) 5
2.2.2.2 BILMAT (3.1) 5
2.2.2.3 NPRD (8) 5
2.3 Model Calibration 5
2.3.1 Documents 5
2.3.1.1 Chapter 4 — Modelling and simulation 5
2.3.1.2 Chapter 4.1 — Industrial ball mill modelling 5
2.3.1.3 Chapter 7 — Model calibration 5
2.3.1.4 Chapter 7.1 — STAMP computer program 5
2.3.1.5 Chapter 7.2 — FINDBS computer program 6
2.3.1.6 Chapter 7.3 — RTD and MIXERS computer programs 6
2.3.2 Software 6
2.3.2.1 STAMP (7.1) 6
2.3.2.2 FINDBS (7.2) 6
2.3.2.3 RTD and MIXERS (7.3) 6
2.3.2.4 HCAL (7) 6
2.4 Process Simulation 6
2.4.1 Documents 6
2.4.1.1 Chapter 5 — Unit models: part A 6
2.4.1.2 Chapter 5.1 — Unit models: part B 6
2.4.1.3 Chapter 5.2 — Unit models: part C 7
2.4.1.4 Chapter 6 — Flowsheet simulators 7
2.4.2 Software for process unit simulation 7
2.4.2.1 PGCRUS (5) 7
2.4.2.2 PJCRUS (5) 7
2.4.2.3 SCCRUS (5) 7
2.4.2.4 TCCRUS (5) 7
2.4.2.5 CRSHRO (5) 7
2.4.2.6 ATMILL (5) 7

Vii
 2.4.2.7 FLTCLS (5) 7
2.4.2.8 ROTARY (5.1) 7
2.4.2.9 VACRSH (5.1) 7
2.4.2.10 KSCRN (5.1) 7
2.4.2.11 WSCRN (5.1) 7
2.4.2.12 VSCRN (5.1) 7
2.4.2.13 BEND (5.1) 8
2.4.2.14 HCONE (5.1) 8
2.4.2.15 GCLASS (5.1) 8
2.4.2.16 SGSEP (5.1) 8
2.4.2.17 RMILL (5.2) 8
2.4.2.18 BMILL (5.2) 8
2.4.2.19 MOGENS (5.2) 8
2.4.2.20 VACFIL (5.2) 8
2.4.2.21 BALDRM (5.2) 8
2.4.2.22 MIXER2, MIXER3 (5.2) 8
2.4.3 Software for process flowsheet simulation 8
2.4.3.1 CRSHEX (6) 8
2.4.3.2 GRNDEX (6) 8
2.4.3.3 HCONEX (6) 8
2.4.3.4 FLOTEX (6) 8
2.4.3.5 SPLITX (6) 8
2.5 Miscellaneous Computations 9
2.5.1 Documents 9
2.5.2 Software 9
2.5.2.1 METCAL (8) 9
2.5.2.2 PARTSZ (8) 9
2.5.2.3 BONDWI and SBOND (8) 9
2.5.2.4 UCONV 9
2.5.2.5 COMPLOT 9
3. METHODOLOGY TRANSFER 10
3.1 SPOC Manual Transfer 10
3.2 SPOC Software Transfer 10
3.3 Remote Access to Demonstrations 13
3.4 Disclaimer and Conditions on Use 13
4. REFERENCES 14

TABLES
1. List of the SPOC chapters 10
2. List of the mainframe computer programs 11
3. List of the IBM-PC computer programs 12

FIGURES
1. The SPOC project objective, structure and resources 1
2. The SPOC project methodology 2
3. Methodology transfer during the SPOC project 10

viii
 FOREWORD
High energy costs and depleting ore reserves combine to make process evaluation and optimization a challenging
goal in the 80's. The spectacular growth of computer technology in the same period has resulted in widely available
computing power that can be distributed to the most remote mineral processing operations. The SPOC project,
initiated at CANMET in 1980, has undertaken to provide Canadian industry with a coherent methodology for process
evaluation and optimization assisted by computers. The SPOC Manual constitutes the written base of this meth-
odology and covers most aspects of steady-state process evaluation and simulation. It is expected to facilitate
industrial initiatives in data collection and model upgrading.
Creating a manual covering multidisciplinary topics and involving contributions from groups in universities, industry
and government is a complex endeavour. The reader will undoubtedly notice some heterogeneities resulting from the
necessary compromise between ideals and realistic objectives or, more simply, from oversight. Critiques to improve
future editions are welcomed.

D. Laguitton
SPOC Project Leader
Canada Centre for Minerai and Energy Technology

AVANT-PROPOS

La croissance des coûts de l'énergie et l'appauvrissement des gisements ont fait de l'évaluation et de l'optimisation
des procédés un défi des années 80 au moment même où s'effectuait la dissémination de l'informatique jusqu'aux
concentrateurs les plus isolés. Le projet SPOC, a été lancé en 1980 au CANMET, en vue de développer pour
l'industrie canadienne, une méthodologie d'application de l'informatique à l'évaluation et à l'optimisation des pro-
cédés minéralurgiques. Le Manuel SPOC constitue la documentation écrite de cette méthodologie et en couvre les
différents éléments. Les retombées devraient en être une vague nouvelle d'échantillonnages et d'amélioration de
modèles.
La rédaction d'un ouvrage couvrant différentes disciplines et rassemblant des contributions de groupes aussi divers
que les universités, l'industrie et le gouvernement est une tâche complexe. Le lecteur notera sans aucun doute des
ambiguïtés ou contradictions qui ont pu résulter de la diversité des sources, de la traduction ou tout simplement
d'erreurs. La critique constructive est encouragée afin de parvenir au format et au contenu de la meilleure qualité
possible.

D. Laguitton
Chef du projet SPOC,
Centre canadien de la technologie des minéraux et de l'énergie

iii
 2. THE SPOC METHODOLOGY, STEP BY STEP

The methodology for process evaluation and optimiza- is discussed in Chapters 4, 4.1 and 8 which cover mod-
tion advocated in the SPOC Manual follows a step-by- elling in general and in the particular case of ball mills.
step approach as illustrated in Figure 2. Considering
that models used to describe mineral processes are Model calibration is the fine adjustment of a particular
always either empirical or phenomenological (11), the model to describe a given set of feed, product, and
collection of data for model development or model val- operating variables. The simplest case of model calibra-
idation is a key step that determines the accuracy of tion is the linear regression of two variables. Complex
subsequent data-processing steps. The more accurate models on the other hand may involve several variables
the sampling, the more accurate the models and simula- and non-linear relationships. The topic of model calibra-
tion results. The sampling of particulate material, how- tion is discussed in Chapters 7, 7.1, 7.2 and 7.3.
ever, is by no means a trivial exercise as can be seen by
perusing the manuals of the world's experts, J. Merks
Raw
and P. Gy (12,13). Based largely on statistical models,
experimental
the theory of sampling still remains too often divorced data
from a very lax practice where collecting "enough"
material takes prominence over sample representativity
and error minimization. Chapters 2, 2.1 and 2.2 discuss Material
sampling methodology under several aspects: tools, balance
sample size, techniques for sampling, error propaga-
tion.
Refined
experimental
Material balance calculations in mineral processing data
can refer to two different applications. In the context of
SPOC, the expression Material Balance computation
refers to the calculation of coherent flow rates for various Multiple regression
non • linear optimization
species analyzed in the process streams.
The objective of such a calculation is usually to keep - - - -

track of the process inventories in the various concen- Unit models

trate stockpiles. In other cases it is performed to assess
the efficiency of process units, to determine circulating
=IL
loads, etc. Several authors, however, use the expression
material balance to describe process simulation which
[11-1 t\
of course must produce results that satisfy material
balance constraints.
Computer
Although the two applications are very related, the term simulator
material balance is restricted in the SPOC Manual, to
the computation of coherent flow rates and assays from
experimental data collected in a real circuit. This topic is Feed
discussed in Chapters 3, 3.1, and 3.2 of the SPOC
Manual. Sampling and material balance are obviously
very related subjects since the sampling error model
directly affects the material balance results.
Process Modelling in the context of SPOC is the devel-
opment of mathematical models relating product vari-
ables to feed variables and operating conditions. Model
selection is the first step at which a choice must be
made between different types of models: empirical,
phenomenological, simple, complex, etc. For instance,
if the model is to be used in real-time simulation, it is
essential that the computation time be sufficiently short.
The amount and precision of the data used to validate
the model is also critical, the range of variations for any
given variable and the number of data points stored in
the database will determine whether the variable can be
explicit in the model equations. Experimental design is
therefore essential when collecting data for modelling. It Fig. 2 — The SPOC project methodology

2
2.1.2.2 GYFORM (8) 2.1.2.7 FADESN (8)
GYFORM computes GY's simplified formula for the vari- FADESN analyzes the results of factorial experiments. It
ance of the relative sampling fundamental error. calculates the sum of squares, regression coefficients,
the experimental error (either from replicate measure-
02 =Cefgd 3 [ 1 – 1 ] ments or from interaction effects that are insignificant
FE
MS ML and negligible) and finally performs the F-test to deter-
mine the significance of each effect.
where:

0-2 FE - variance of the relative sampling funda-

– 2.2 MATERIAL BALANCE
mental error 2.2.1 Documents
M5 = sample mass (g) The following chapters contain information on material
NIL = lot mass (g) balance computation.
C = composition factor (g/cm 3)
2.2.1.1 Chapter 3 — Material balance
e = liberation factor (0 or 1)
This chapter of the SPOC Manual has been developed
f = particle shape factor (0.5) for the participants to a workshop on material balance
g = size distribution factor computation. The various methods of material balance
calculation are reviewed in order of increasing complex-
d = maximum particle size (cm). ity. The two packages BILMAT and MATBAL are largely
documented in Chapters 3.1 and 3.2 to which the reader
C, the most variable coefficient, is given by: is referred. The justification of methods requiring com-
puters is presented on the basis of a search for best
1-X estimators, i.e., estimators that take into consideration
C= L R1 - XL )pc + XLpg]
XL as much experimental information as possible in order
to provide "well informed" estimates. Each family of
where: methods is illustrated by numerical exercises.
XL = critical concentration
2.2.1.2 Chapter 3.1 — BILMAT computer program
PC = specific gravity of critical component
Material balance computations for complex ore and coal
Pg = specific gravity of non-critical compo- flowsheets are based on large experimental data sets
nent (g/cm 3). and require a computer. The BILMAT program has been
developed to solve such complex problems and com-
2.1.2.3 TIMESR (8) pute a sensitivity analysis of the estimated process
variables.
TIMESR performs time series analysis. It calculates
mean, variance, and variogram for time or space depen- This user's guide has been written for two main types of
dent data entered by the user. users: the process engineer who wants to access it as a
black box to compute his material balances and the
2.1.2.4 STATPD (8) research engineer interested in the method and the
program itself. The FORTRAN routines are fully docu-
STATPD calculates statistics of paired data. It computes mented and two sample runs are presented.
from a pair of data x's and y's the mean, the variance, the
standard deviation, the coefficients of variation, the dif-
ference and the absolute difference. 2.2.1.3 Chapter 3.2 — MATBAL computer program
MATBAL is a FORTRAN program to calculate data
2.1.2.5 ANOVA (8) adjustments required in order to satisfy mass conserva-
tion equations valid for a mineral process flowsheet. The
ANOVA performs a one- and two-way analysis of vari- calculation is done by minimizing the sum of squares of
ance for an experiment with more than two treatments. assay and flow rate adjustments weighted by their
reciprocal variance. The available data and the mass
conservation equations are coded in a suitable form and
2.1.2.6 URANDT (8)
the derivatives of a Lagrangian function are internally
URANDT simulates samples drawn from a normal pop- calculated and cancelled to obtain the adjustments and
ulation, and illustrates the concepts of a sampling error flow rates which are a solution to the problem. The
model and its application to simulation studies. precision of calculated flow rates can also be obtained.

4
Writing a program to perform such calculations is only duction to experimental design. Finally, the appendices
part of the overall task if the method is to be transferred present several useful techniques of statistical data
to the end user in the mineral industry. Program docu- analysis and regressions as well as a complete data set
mentation has traditionally been the weak point which obtained during the analysis of the Gibraltar grinding
jeopardizes and often cancels all efforts made to circuit.
develop a program. This user's manual is an attempt to
avoid such a dead-end by fully explaining both the 2.3.1.2 Chapter 4.1 — Industrial ball mill
method and the program so that the program can be modelling
installed in different computers or translated if FOR-
A simple kinetic ball mill model proved adequate after
TRAN is not supported by a given machine. The docu-
testing with 100 sets of industrial grinding data. The
mentation follows the standards of the A.S.C.E.*
essential components of the model are (a) a breakage
function determined from laboratory tests; (b) a mixers-
2.2.2 Software in-series residence time distribution; and (c) a selection
function derived from industrial data. Results show that
The following programs have been selected for material the absolute selection function (S 1 1-1) is statistically invar-
balance computation. iant with changes in feedrate, feedsize distribution and
pulp density over the normal operating range. All the
2.2.2.1 MATBAL (3.2) experiments and calculational procedures used in the
calibration of the model are fully documented. This
MATBAL computes the coherent material balance from includes an examination of model residuals and a dis-
sampling data by a least-square method that simul- cussion of model reliability.
taneously calculates flow rates and adjusts assays
according to a given normal error distribution. It can
process complex flowsheets and redundant data. Error 2.3.1.3 Chapter 7 — Model calibration
estimates for the calculated flow rates are also available. Phenomenological or empirical models used to
describe most mineral and coal process units require
2.2.2.2 BILMAT (3.1) extensive calibrations at the development stage as well
as for any subsequent update. The range of techniques
BILMAT solves the same type of problems as MATBAL required for such calibrations goes from simple linear
but was developed to process a larger variety of experi- regression to computation-intensive, multivariable non-
mental data including assays of particle size fractions. linear model calibrations.
The SPOC software library contains programs to
2.2.2.3 NPRD (8) address these model calibration requirements which are
NPRD implements the N-product formula. Given a unit reviewed through selected sample runs. References are
or a circuit with N +1 branches on which N –1 assays given to the appropriate software.
have been measured, the flow distribution is calculated
and a computation of the precision of the results is also 2.3.1.4 Chapter 7.1 — STAMP computer program
performed if the sampling error is known.
The STAMP program is a conversational package dedi-
cated to experimental data analysis. STAMP features
2.3 MODEL CALIBRATION data entry and data file maintenance, various regression
methods (including simple linear regression, multiple
2.3.1 Documents linear regression, stepwise linear regression, poly-
The following chapters in the SPOC series contain infor- nomial regression) and three non-linear optimization
mation on model calibration. algorithms (Powell, Simplex and Rosenbrock). Options
selection is menu driven.
2.3.1.1 Chapter 4 — Modelling and simulation This chapter is a user's guide to the program and
includes a detailed sample run.
This chapter introduces the concepts of modelling and
simulation of mineral processes in practical terms. The The version of the program described here (STAMP84)
general terminology applying this discipline is defined, was written in FORTRAN for a Cyber 730 computer. It
and several case stories of industrial applications are easily runs in a 70 K words memory in CDC terminology
briefly reviewed. The sequence of steps leading to a (29 K words decimal), but uses file structure and FOR-
circuit simulator are described in the case of simple TRAN routines specific to Cyber. A version is also avail-
models for ball mills and hydrocyclones with an intro- able for an IBM-PC environment.

"Engineering Computer Program Documentation Standards; Journal of the Soil Mechanics and Foundations Division SM3,
March/Mars 1973, pp. 249-266.

5
2.3.1.5 Chapter 7.2 — FINDBS computer program of coefficients by various types of regressions including
non-linear optimization by direct-search.
This manual describes the various procedures involved
in the development of a ball-mill simulator. The major
task is the estimation of the breakage rate and distribu- 2.3.2.2 FINDBS (7.2)
tion parameters which are central to the model. That is FINDBS is an interactive program to calculate the break-
why the bulk of this volume is devoted to describing the age and selection function coefficients of the kinetic
methods, mathematics and computer programs used to model of ball mills. Ore characteristic breakage func-
calculate model parameters. Section 1 outlines the gen- tions can be computed from batch mill experiments and
eral structure of the kinetic ball-mill model. Section 2 used in conjunction with industrial data to determine the
focuses on the laboratory tests and computation tech- machine-dependent selection functions that will allow
niques required to estimate the breakage distribution accurate simulations within the range of observed vari-
and rate parameters. Finally, Section 3 illustrates the ables.
methods using data from an operating industrial mill.
The computer programs are interactive and fully docu- 2.3.2.3 RTD and MIXERS (7.3)
mented and can therefore be used without a detailed
knowledge of the principles. However, it must be empha- RTD is a conversational program to calculate the resi-
sized that a correct interpretation of the results depends dence time distribution that best describes the material
on a thorough understanding of the kinetic model. For transport through the mill, from tracer experiments. This
that reason, program users should at least read Sec- program is most useful when recycled tracer overlaps
tion 1 describing the basis of the method. Further expla- the primary signal in the experimental data. MIXERS
nations and analysis appear in the remaining sections; uses a discretized residence time distribution to cali-
the mathematics have been detailed in the appendices. brate a tank-in-series model of the distribution.
The program structure should allow easy modifications.
2.3.2.4 HCAL (7)
2.3.1.6 Chapter 7.3 — RTD and MIXERS computer HCAL is proposed for the calibration of hydrocyclone
programs efficiency curves. Two kinds of models are available, the
The transport properties of material through various ore equation developed by Plitt and used in the simulation
and coal processing units are important factors which module HCONE or the model proposed by Lynch.
control the performance of those units. This manual
describes the basic tools for determining the flow pat-
tern for a piece of equipment. The residence time dis- 2.4 PROCESS SIMULATION
tribution is defined, as well as three approaches to repre-
sent it. Then, several experimental methods based on 2.4.1 Documents
tracers are presented and compared. The FORTRAN
programs used to process the tracer data are fully docu- The following chapters contain information on process
mented, as are the mathematics on which they are simulation.
based. All the methods and programs are illustrated with
actual data from industrial grinding and flotation circuits. 2.4.1.1 Chapter 5 — Unit models: part A
This manual is directed to plant process engineers. All Specific mathematical models of comminution and flota-
the necessary definitions are given, and only limited tion equipment are described and the need for quantify-
mathematical ability is required to apply the methods ing equipment constraints is discussed along with the
and use the programs. For those more familiar with limitations of each model.
process modelling, extended appendices give details of
the mathematics. This should allow ongoing improve- The mathematical description of a primary gyratory
ments and modifications to the packages as well as crusher, a primary jaw crusher, a secondary cone
independent programming of the methods for users crusher, a tertiary cone crusher, a crushing rolls, an
who want to have their own program in a language other autogenous grinding mill and a flotation cell have been
than FORTRAN. converted to FORTRAN programs for subsequent use
as subroutines in an executive steady state simulator.

2.3.2 Software Documentation should permit potential users to install

subroutines in different computers or to translate them
The following programs are related to model calibration. into language other than FORTRAN.

2.3.2.1 STAMP (7.1) 2.4.1.2 Chapter 5.1— Unit models: part B

STAMP is a conversational regression package for easy Specific mathematical models of classification and coal-
entry and manipulation of experimental data, basic sta- processing equipment are described. Constraints and
tistical calculations, study of correlations and calculation limitations are discussed for the following mod-

6
els: rotary beaker, a general coal crusher model, four 2.4.2.5 CRSHRO (5)
screen models (Karra, Whiten, Vaillant and a sieve bend
CRSHRO (Crushing rolls model). The concept used is
model), a hydrocyclone model, a gravity classifier model
that breakage leads to material smaller than the par-
and a specific gravity separator model. For each FOR-
ticular size range of interest. Material remaining within
TRAN program, a small driver and test data are pro-
the size range after passing through the rolls is consi-
vided.
dered material that is unbroken (by-passed).

2.4.1.3 Chapter 5.2 — Unit models: part C

2.4.2.6 ATMILL (5)
Mathematical models of the following units are
ATMILL (Autogenous mill model). The basic model is
described: rod mill, ball mill, balling drum, rotary vac-
after Gault and Lynch.
uum filter, Mogensen sizer. For each model, a brief
engineering discussion as well as test data are pro-
vided. 2.4.2.7 FLTCLS (5)
FLTCLS (Flotation cell model). This is based on the
2.4.1.4 Chapter 6 — Flowsheet simulators model of a flotation cell and bank of cells by Mular and
Bull. An interactive method using linear interpolation is
This crushing plant simulator is based on industrial data
used to determine estimates of tailing volume, con-
and can be used for the optimization of existing plants or
vergence being judged by the difference between suc-
the design of new plants. It allows fast computations at a
cessive estimates.
lower cost and reduces the chances of error. For a
meaningful use of the program and a good interpretation
of the results, the user must be familiar with crushing 2.4.2.8 ROTARY (5.1)
plant practice.
ROTARY (Rotary breaker model). This is Valliant's
model which supposes that coal is processed serially in
a number of tumbling events. Each event consists of
2.4.2 Software for Process Unit coal breakage followed by screening.
Simulation
The following programs are related to process simula- 2.4.2.9 VACRSH (5.1)
tion.
VACRSH (Valliant model of crusher). This is Valliant's
model which identifies three crushing zones in coal
2.4.2.1 PGCRUS (5) breakage. This model is said to be capable of simulating
PGCRUS (Primary gyratory crusher model). This model gyratory/jaw, single roll, multiple roll or cage mill
was proposed and developed by Mular and uses the crushers, either in primary or secondary mode.
concept of selection/breakage/classification. The only
design variable included in the model is the open side
set. 2.4.2.10 KSCRN (5.1)
KSCRN (Screening model by Karra). This model incor-
2.4.2.2 PJCRUS (5) porates all the parameters which are deemed significant
in evaluating a screen's performance. The model rela-
PJCRUS (Primary jaw crusher model). This model tionships were established from data obtained from 20
treats the Primary jaw crusher as a special case of the dry screening tests.
Primary gyratory crusher model using scalped
(screened) feed. The only design variable included in
the model is the open side set. 2.4.2.11 WSCRN (5.1)
WSCRN (Whiten screen model). This model has been
2.4.2.3 SCCRUS (5)
used in the simulation of coarse screens in mineral
crushing plants. The model computes the mean proba-
SCCRUS (Secondary cone crusher model) is the bility that particles within a narrow size inten.fal will report
Whiten cone crusher model modified by Hatch and to the screen oversize product.
Mular.
2.4.2.12 VSCRN (5.1)
2.4.2.4 TCCRUS (5)
VSCRN (Valliant screen model). This is a relatively sim-
TCCRUS (Tertiary cone crusher model) is similar to the ple model for coarse screening requiring a minimum of
modified Whiten model for the secondary cone crusher. user inputs. It is capable of simulating both wet and dry
It was developed for a 7-foot short head cone crusher screening for both top and bottom decks and for proj-
equipped with a 300 hp motor at 4.16 kv. ected throughfall apertures from 18 to 0.01 in.

7
2.4.2.13 BEND (5.1) 2.4.2.20 VACFIL (5.2)
BEND (Sieve bend model). This model consists of a VACFIL (Vacuum filter model). This model is based on
series of empirical equations which relate the partition the work of H. Rudolf. Using the operational parameters
curve parameters in terms of the feed slurry per cent of a unit filter an estimate of the unit throughput and filter
solid, the mass flow rate of water, and the slot width cake thickness and moisture is obtained.
(width of opening between bars).
2.4.2.21 BALDRM (5.2)
2.4.2.14 HCONE (5.1)
BALDRM (Balling drum model). This model is based on
HCONE (Hydrocyclone model). The Plitt model is based the work of M. Cross. The balling drum circuit output is
on a large database of over 200 tests involving both described as a function of time. There are two growth
laboratory and industrial scale cyclones. HCONE is mechanisms, with fresh feed and in the absence of fresh
available as three different models: feed.

— HCONE1 calculates the partition

curve parameters from cyclone 2.4.2.22 MIXER2, MIXER3 (5.2)
dimensions and operating condi- MIXER2 and MIXER3 simulate the mixing of two or
tions. three slurry streams, respectively.
- HCONE2 calculates the partition
curve from values of the cyclone
parameters D50C, EM and RF.
- HCONE3 uses a fixed partition (effi- 2.4.3 Software for Process Flowsheet
ciency) curve.
Simulation
2.4.2.15 GCLASS (5.1) 2.4.3.1 CRSHEX (6)

GCLASS (Gravity classifier model). The model is CRSHEX (Crushing plant executive) has been devel-
applicable to any type of gravity classifier where the oped for the interactive simulation of crushing plants on
settling area can be defined, e.g. Hydroseparators, spi- an IBM-PC type of microcomputer.
ral or rake classifiers, classifier cones.
2.4.3.2 GRNDEX (6)
2.4.2.16 SGSEP (5.1) GRNDEX is used to simulate grinding plants involving
SGSEP (Specific gravity separator model). This model rod mill, ball mills, hydrocyclones and mixers. The unit
is the partition curve used in the coal industry for many modules that can be used are: RM ILL, BMILL, HCONE,
years. The partition curve is, in general, considered to and two or three stream mixers.
be characteristic of the unit used and independent of the
feed density. 2.4.3.3 HCONEX (6)
HCONEX is used to simulate a multicycloning unit invol-
2.4.2.17 RMILL (5.2) ving solids recycling and water addition. It uses two
RMILL (Rod mill model). This is the classical Lynch- kinds of unit modules, HCONE unit module and the
Callcott matrix model using breakage and selection MIXER2 and MIXER3 unit modules.
functions.
2.4.3.4 FLOTEX (6)
2.4.2.18 BMILL (5.2) FLOTEX is used to simulate a flotation plant involving
BMILL (Ball mill model). This is the kinetic model of ball- solids recycling and water addition. It uses two kinds of
milling based on concepts of breakage and selection unit modules, the FLTCLS module and the MIXER2 and
functions, and tank-in-series residence time distribu- MIXER3 modules.
tion.
2.4.3.5 SPLITX (6)
2.4.2.19 MOG ENS (5.2) SPLITX is a general flowsheet simulator based on a
MOGENS (Mogensen sizer model). This routine simu- split-coefficient model of two or three product sepa-
lates the operation of a linear, multi-deck, probability rators and mixers. It features automatic flowsheet
screening machine known as the Mogensen sizer. The decomposition and tear-stream selection as well as a
probability screening machine sacrifices a degree of convergence acceleration method.
screening accuracy in favour of higher throughput and
reduced blinding. •

8
2.5 MISCELLANEOUS COMPUTATIONS 2.5.2.3 BONDWI and SBOND (8)
2.5.1 Documents BONDWI is a conversational implementation of the
bond work index (WI) formula used to process results of
During the course of the SPOC project, several compu- laboratory tests.
ter programs have been developed or adapted to per-
form ancillary tasks or routine laboratory computations. SBOND calculates the simplified bond formula for work
They are documented in Chapter 8 of the manual. A index determinations using a reference material of
brief description of these programs is given below. known WI.

2.5.2 Software 2.5.2.4 UCONV

2.5.2.1 METCAL (8) UCONV performs unit conversion from and to metric in
conversational mode.
METCAL computes the species distribution and abso-
lute content from assay values of several fractions of a
feed material. The program offers options for ore and 2.5.2.5 COMPLOT
coal, combinations of fractions, table format. COMPLOT is computer data plotting utility program. x-y
graphs with optional axes and labels can be previewed,
2.5.2.2 PARTSZ (8) modified and sent to hard copy plotter. It requires a
Textronix graphic terminal, and software libraries. Con-
PARTSZ manipulates particle size data in various repre- version to other machines may be more difficult.
sentations (cumulative, non-cumulative, etc.).

9
 3. METHODOLOGY TRANSFER

From the start, the SPOC project was aimed towards a Table 1 — List of the SPOC chapters
methodology transfer between R&D centres and the
mineral industry (17-21). The release of software and the
distribution of the SPOC Manual by the Technology Chapter Title
Information Division of CANMET provide a durable 1 Summary
channel to achieve this transfer, but it is clear that the 2 Sampling Methodology
human factor plays a major role in drawing the full 2.1 SAMBA Computer Program
benefits from the SPOC methodology. In other words, 2.2 Grinding Circuit Sampling
what is needed is not only a series of soft disks and a
collection of manuals, but also a new profile of mineral 3 Material Balance
processing engineer for whom interaction with a compu- 3.1 BILMAT Computer Program
ter terminal is as natural as reading the technical liter- 3.2 MATBAL Computer Program
ature. This integration of the computer as an everyday 4 Modelling and Simulation
tool is done routinely in the engineering schools of the 4.1 Industrial Ball Mill Modelling
mid-80's but requires special training for engineers from
previous vintages. This has been the objective of the 5 Unit Models (Part A)
many seminars given between 1980 and 1985 (Fig. 3) 5.1 Unit Models (Part B)
and will remain a critical element of the post-project 5.2 Unit Models (Part C)
activities. 6 Flowsheet Simulators
7 Model Calibration
7.1 STAMP Computer Program
7.2 FINDBS Computer Program
r 0 WORKSHOP PARTICIPANTS 7.3 RTD & MIXERS Computer Program
• PROGRAM TRANSFERS 8 Utility Programs

3.2 SPOC SOFTWARE TRANSFER

40 -120
The FORTRAN programs generated by the SPOC proj-
ect can be acquired in source form from the Technology
20 -110
Information Division of CANMET.
The source programs come in three categories for
8 AVAILABLE CHAPTERS ordering purposes:
1 1
1980 1981 1982 1983 1984 — CDC CYBER magnetic tape of all
Fig. 3 — Methodology transfer during the SPOC project SPOC programs as operating on the
CDC computer and in a version and
style of FORTRAN (4.8) that makes
3.1 SPOC MANUAL TRANSFER transfer to other mainframes quite
straightforward (see also Table 2);
The SPOC manual consists of 18 separate chapters as
listed in Table 1. They can be acquired separately or as a — IBM-PC soft disks (5 1/4 in.) of individ-
set from: ual programs or group of programs
related by theme. The FORTRAN ver-
CANMET, sion is microsoft and the operating
Technology Information Division system is MS-DOS (see also Table 3);
555 Booth Street
Ottawa, Ontario — a magnetic tape containing a copy of
K1A 0G1 all the IBM-PC source programs.

10
 Table 2 — List of the mainframe computer programs

MAGNETIC TAPES, UNLABELLED, 9 TRACK, 1600 BPI*, ASCII or

EBCDIC, 80 CHARACTERS/RECORD, 50 RECORDS PER BLOCK (*OPTION)
SYSTEM USED: CDC CYBER 730
OPERATING SYSTEM: NOS/BE
FORTRAN COMPILER: CDC FORTRAN EXTENDED 4.8
AVAILABLE AS: TAPE No. 1: THE FORTRAN SOURCES TESTED ON
A CDC CYBER
TAPE No. 2: A COPY OF THE IBM-PC SPOC
DISKETTES
PROGRAMS ON TAPE 1: SAMBA, GYFORM, TIMESR, STATPD, ANOVA,
URANDT, FADESN,
MATBAL4, BILMAT,
ADSTAMP, STAMP, FINDBS, RTD, MIXERS, HCAL,
PGCRUS, PJCRUS, SCCRUS, FLTCLS, ROTARY,
KSCRN, VSCRN,
WSCRN, GCLASS, SGSEP
TCCRUS, CRSHRO, ATMILL, VACRSH, BEND,
HCONE, RMILL,
BMILL, MOGENS, VACFIL, BALDRM
GRNDEX, HCONEX, FLOTEX, SPLITX,
METCAL, PARTSZ, BONDWI, SBOND, UCONV, NPRD,
COMPLOT
PROGRAMS ON TAPE 2: SEE LIST OF SPOC IBM-PC COMPUTER
PROGRAMS (TABLE 3)

11
 Table 3 — List of the IBM-PC computer programs

SOFT DISKS, 5 1/4 INCH, DOUBLE DENSITY

SYSTEM USED: IBM-PC XT (256K), 1 HARD DISK DRIVE AND
ADAPTER (10MB), 1 SOFT DISK DRIVE AND
ADAPTER (320K), 1 MATH COPROCESSOR 8087,
PRINTER, KEYBOARD, COLOR/GRAPHICS MONITOR
ADAPTOR (OPTIONAL), 1 SYSTEM MONITOR OR
SYSTEM GRAPHICS MONITOR (OPTIONAL)
OPERATING SYSTEM: MS-DOS VERSION 3.00
FORTRAN COMPILER: MICROSOFT FORTRAN 77 VERSION 3-2 COMPILER
FOR MS-DOS
8086 OBJECT LINKER VERSION 2.41
PACKAGE NO. OF DISKETTES PROGRAMS
1 1 SAMBA, GYFORM, TIMESR, STATPD, ANOVA,
URANDT, FADESN
2 1 MATBAL4
3 1 BILMAT
4 2 STAMP
1 STAMP (ADSTAMP)
5 1 FINDBS, RTD, MIXERS, HCAL
6 1 PGCRUS, PJCRUS, SCCRUS, FLTCLS, ROTARY,
KSCRN, VSCRN, WSCRN, GCLASS, SGSEP, ATMILL,
VACRSH, MOGENS
2 TCCRUS, CRSHRO, BEND, HCONE, RMILL, BMILL,
VACFIL, BALDRM
7 1 GRNDEX
2 CRSHEX
8 1 HCONEX, FLOTEX
2 SPLITX (1 of 2)
3 SPLITX (2 of 2)
9 1 METCAL, PARTSZ, BONDWI, SBOND, UCONV, NPRD

12
3.3 REMOTE ACCESS TO — the programs or manuals may not be
DEMONSTRATIONS copied for use by persons or organiza-
tions other than the purchaser except
Since late 1983, a system of demonstrations of the with the written permission of CAN-
SPOC programs has been made available to the Cana- MET;
dian mineral processing industry to help users become
familiar with the various programs of the SPOC libraries. — due acknowledgement is expected in
Software development had limited the resources allo- the publications resulting from the use
cated to the exploitation of this training system which of programs or of modifications involv-
should experience a peak of activity during the post- ing less than 50% of the executable
project period. Request for temporary access to this statements;
remote training facility can also be routed to the Tech- — the use of the programs in commercial
nology Information Division of CANMET. applications must be agreed in writing
with CANMET.
3.4 DISCLAIMER AND CONDITIONS ON
USE The liaison officer can be contacted at the following
address:
The software and manuals distributed by CANMET are
released under the following understanding: SPOC Liaison Officer
Technology Information Division
— the charges do not cover any support 555 Booth Street
and CANMET cannot undertake to Ottawa, Ontario, Canada
provide assistance of any kind nor K1A 0G1
guarantee that the programs and doc-
umentation are free from errors. How-
ever, we will try to remain available for
questions from users and we invite
written suggestions to improve the
programs;

13
 4. REFERENCES
1. Ford, M.A. "The simulation of ore dressing plants"; 11. Finch, J.A.; Laplante, A.R.; Leung, J. "SPOC Man-
Ph.D. thesis; U. Witwatersrand, Johannesburg, ual, Chapter 4: Modelling and simulation meth-
S.A.; 1979. odology for ore and coal processes"; Report
SP85-1/4; CANMET, Energy, Mines and
2. McMullen, J. "Calibration of grinding and classifica- Resources Canada; 1985.
tion models for the optimization of the Heath
Steele Mines grinding circuit"; A thesis submitted 12. Merks, J.W. "Sampling and weighing of bulk solids";
to the Faculty of Graduate studies; Laval Univer- Series on Bulk Materials Handling; Vol 4; Trans
sity, for the degree of Master of Science; Quebec; Tech Publications, Karl Distributors; Rockport,
1981. MA; 01966; USA; 1985.

3. Ruebush, J.C. "Simulator for individual mineral 13. Gy, P. "Sampling of particulate materials - Theory
behaviour in an integrated grinding and flotation and practice"; Developments in Geo-
plant"; A thesis submitted to the Faculty of the mathematics 4; Elsevier Scientific Publishing
University of Utah, for the degree of Master of Co.; New York; 1979.
Science; Salt Lake City; 1982.
14. Control Data Corporation, "FORTRAN extended,
4. Adel, G.T. 'An interactive simulation package for Version 4, Reference manual"; Publication No.
mineral processing systems'; A thesis submitted 60497800.
to the Graduate Division of the University of Cal-
ifornia, Berkeley, for the degree of Doctor of Engi- 15. Control Data Corporation, "CDC NOS/BE, Version
neering; Berkeley; 1982. 1, Reference manual"; Publication No. 6049380.

5. Laguitton, D. "Process simulation software: is there 16. Control Data Corporation, "CDC INTERCOM, Ver-
progress towards universal systems?"; Proceed- sion 4, Reference manual"; Publication No.
ings of the 4th IFAC Symposium; Helsinki; 1983. 60494600.

6. Laguitton, D. and Flament, F. "On-line software 17. Reid, K.J.; Smith, K.A.; Voller, V.R.; and Cross, M.'A
demonstrations for mineral and coal pro- survey of material balance computer packages in
cessors"; Proceedings of the 16th Canadian the mineral processing industry; Proceedings of
Mineral Processors' Meeting; Ottawa; 1984. the 17th APCOM Conference; T. B. Johnson, R.J.
Barnes, Editors; AIME 41-62; 1982.
7. Agar, G. "Comminution - grinding and liberation";
In Research needs in mineral processing; 18. Carmola, R.; Hoover, M.R.; Kim, J.H.; and Kloda,
Report of a workshop held at Columbia Univer- S.J. "Computer simulation of synthetic fuels feed
sity; U.S. Department of Commerce; PB preparation circuits"; Proceedings of the First
260-735; August 1975. Conference on Use of Computers in the Coal
Industry; Y.J. Wang, R.L. Sanford, Editors; AIME
8. Mular, A.L. "Empirical modelling and optimization of 177-186; 1983.
mineral processes"; Mineral Sci. and Eng. 4; July
1972. 19. Madge, D. "Evaluation of the SPOC project"; Report
submitted to CANMET; Contract OGR84-00499;
9. Hodouin, D.; Bérubé, M.A.; Marchand, J.C.; and DSS File No. 06GR.23440-4-9322; February
Everell, M.D. "Computer simulation of the Heath 1985.
Steele grinding circuit; Paper presented at the
17th Annual Conference of Metallurgists; Metal- 20. Flintoff, B.C. "Evaluation of the SPOC project";
lurgical Society of the CIM; Montreal; 1978. Report submitted to CANMET; Contract 0GR84-
00499; DSS File No. 06GR.23440-4-9322; Feb-
10. Laguitton, D. "Methodology transfer for the simula- ruary 1985.
tion of mineral and coal processing plants"; CIM
Bulletin 75-840:166; 1982. 21. Laguitton, D.; Flament, F.; and Sirois, L. "Le calcul
par ordinateur dans l'industrie Minéralurgique
Canadienne en 1984"; Proceedings of the 15th
International Minerai Processing Conference;
Cannes; June 1985.

14
 85-1/1E C212sp
622(21) c.2
Laguitton, D.
Summary. -- (The SFOC manual.
Chapter 1) --

c1985 14 p.
Canada
Printed in Canada