1

Comminution Bern Klein, Ph.D., P.Eng.

Professor and Head

And Norman B. Keevil

Institute of Mining
Size Engineering

Classification University of British

Columbia

Vancouver Canada
May 2013

Bern Klein Ph.D P.Eng

Professor and Head, Norman B. Keevil Institute of
Mining Engineering, University of British
Columbia
Vancouver, Canada

[email protected]
604 822 3986

Process Design Engineer 2000-1998

Professor Mineral Processing 1998-present

1
 Teaching:
Comminution and Size Classification
Process Design
Process Mineralogy
Processing of Precious Metal Oes

Research Areas:
Comminution - High Speed Stirred Milling, High
Pressure Grinding Rolls
Rheology Hydraulic Transport, Paste and
Thickened Tailings
Sensors and Sorting Systems
Continuous Centrifugal Gravity Concentrators
Weathering of Waste Rock

Course Outline
UBC Norman B. Keevil Institute of Mining Engineering
Comminution Overview
Comminution Theory
Process Development and Plant Design
Process Development
Metallurgical Testing
Sampling
Process Mineralogy
Physical Properties
Plant Design 30 min
Comminution Technologies
Sampling
Crushing
Screening
Features and Design
Sizing and Selection
Crushing and Screening Plant Design

2
 Course Outline
HighPressureGrindingRolls
SensorsandSorting
Introduction to Grinding
Grinding Circuits
Grinding and Classification Ore Characterization
Grinding and Classification Circuit Design
Mill Power
Mill Sizing
Mill Sizing Example
Ball and Rod Mill Sizing Olav Meijo
Size Classification
Fine Grinding
Energy Efficiency in Mining
Statistical Experiment Design

CourseObjectives
Tolearnaboutthemainunitoperationsthatareusedto
processmineralsincluding
Introducenewcomminution technologiesandsystems

Describingthefundamentalphysicalprinciplesthatare
exploited/employedtoachievethepurpose
Demonstratinghowtosizeandselecttheequipment

Demonstratingtheuseoftheequipmentinmineral
processing

3
 7

List of Recommended Publications

Mineral Process Plant Design, A.L. Mular, D.N. Halbe, D.J. Barratt,
SME, 2002
Mineral Comminution Circuits, T.J. Napier-Munn, S. Morrell, R.D.
Morrison, T, Kojovic, JKMRC Mining and Mineral Processing, 2005
Advances in Comminution, S.K. Kawatra, SME, 2006
Mine to Mill Conference, A. Scott, S. Morrell, Aus IMM, 1998
Advances in Autogenous and Semiautogenous Grinding
Technology, Proceedings, 1989, 1996, 2001, 2006, 2011 (2015)
Proceedings of the Annual General Meeting of the Canadian
Mineral Processors, 1964 - present

COMMINUTION
OVERVIEW

4
 Comminution Overview

Comminution at Face Coarse Breakage Fine Breakage

Particle Weakening Size Classification

Sensing and Sorting

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

Comminution at Face
- Mine to Mill
- Drill & blast optimization
- Continuous miners
- Caving methods
- Hydrofracturing

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

5
 Particle Weakening

- Sellfrag
- Electric Pulse Treatment
- Microwave

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

Coarse Breakage
- Crushers (gyratory, jaw, cone)
- Vertical roller mills (VRM)
- High Pressure Grinding Rolls (HPGR)
- Vibrocone
- SAG milling

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

6
 Sensing and Sorting Systems

-Sorting
- High capacity sorting machines
- Sensors
-Other Sensing Applications

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

Classification
-Classification equipment
-Coarse classification
-Screening
-De-agglommeration
-Cyclones

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

7
 Fine Breakage (Grinding)
-Ball milling
-Low speed stirred mills
-High speed stirred mills

CHARACTERIZATION
Lithology Mineralogy
Geometallurgy

16

Definitions
Mineral Processing

The technology of economically converting mineral bearing

raw material into individual mineral constituents; the minerals
remaining essentially unaltered in physical and chemical form
throughout. The temperature of the system normally is less
than the boiling point of water. Mineral processing is also
known as mineral beneficiation, milling or concentration.

Unit Operation
An individual process with a specific function, which is a
component or forms part of a complex process.

8
 17

Mineral Processing
Ore

Comminution

Tailing

Mineral Separation

De-watering

Concentrate

18

Primary
Primary
Mine Crushing PlantDesign
Crushing

Mineralogy&
2nd/3rd
2nd/3rd Autogenous
Autogenous Process
Screening
Screening
Crushing
Crushing Grinding
Grinding Development

RodMill/Ball
RodMill/Ball Sampling
Cyclone
Cyclone Gravity
Mill
Mill

Material
Flotation/ Transport
Regrind
Regrind
Leaching
Tailing

Thickening Filtering
Filtering

UnitOperations

9
 19

THEORIES OF
COMMINUTION

20

Theories of Comminution
Largeparticle+Energy=SmallParticles+Sound+Heat
Energydissipation(sound+heat)accountsfor99%of
inputenergy
Mostexpensiveunitoperationrequiring5 40kWh/t

10
 21

Breakage Mechanisms
Properties of solids that influence breakage mechanisms:
Elastic versus Plastic (stress-strain relationship linear or
nonlinear)
Strain behavior (fracturing) depends on:
Microstructural differences in physical properties of
adjacent minerals e.g. hardness, brittleness, cleavage
Macrostructural weaknesses e.g. along joints, bedding
planes, grain boundary cementation/impurities
Microstructural weaknesses e.g. schistosity, number of
flaws/ number and type of defects including cracks,
crystal surfaces, impurities, minute inclusions

22

Surface Properties
From fracture mechanics, for an isotropic material:
F/A = 2Es/L
F - critical force to initiate fracture
A - cross sectional area
L - length of specimen
Es - surface energy
- Youngs modulus
Critical stress to initiate fracture is proportional to
surface energy which depends on the number of flaws
on surface.

11
 23

Breakage Energy

cracks or flaws = energy required

brittleness = energy required
coarse grain crystals = energy required
water = energy required

24

1st Theory Rittinger (1867)

Constant energy per unit of surface area generated.
New surface area produced by crushing and grinding
is directly proportional to the useful work input.

Surface area is
1 1 inversely
E k
X 2 X1 proportional to the
diameter of the
particle

where E - energy consumed

X2 - product size
X1 - feed size

12
 25

2nd Theory Kick (1885)

Constant energy per unit mass for similar relative reduction.
The work input required to deform a homogeneous rock to
the yield point and to break it is proportional to the reduction
in diameters of the particles concerned.

x
E k ln 1
x2

Theory: Work required to reduce rock from 4 cm to 2 cm

equals work required to reduce rock from 2 cm to 1 cm.
Rock is not homogeneous due to flaws and breakage is
controlled by number of flaws. Rock breaks at far below the
stress required by its theoretical homogeneous elastic limit.
Overestimates work input at coarse sizes and
underestimates work input at fine sizes.

26

3rd Theory Bond (1951)

Useful work input per ton is inversely proportional to the
square root of the new surface area produced.
Derived empirically from operating data and experimental test
results.
Compromise between Rittinger and Kick Theories and is still
used for most mill designs.

1 1
E k
x2 x1

Bond Equation gives us indices for Work Index and Operating

Work Index for ores

13
 27

General theories of comminution

Consider the incremental energy dE required to produce
an incremental change in size dD. More energy is
required to achieve a similar relative degree of size
reduction as the product becomes finer:

Where E = specific energy to introduce new surface

energy; K = constant; D = particle size; n = value to
describe behaviour in different size ranges.
Rittinger: n=2; Kick: n=1; Bond: n= 1.5

28

Log-Log plot of Energy Consumed vs Size

Rittinger
slope=-1

Bond
Slope =-1/2 Kick slope=0

-1000m, Rittinger Particle Size (m) 1cm+, Kick

14
 29

Comminution Research
Main objectives:
Reduce unit operating cost ($/t)
Increase throughput
Improve downstream process performance as a result of
an improved size specification.
Improve energy efficiency.

Two kinds of improvements

Fundamental change, novel technologies (e.g. ultrasonic,
microwave, impact and electricity)
Incremental (design, operating practice)

30

Factors Affecting Fracturing

Ability to fracture rocks depends on degree of internal strain

which is influenced by:
composition
nature of chemical bonds
grain boundary cementation/ impurities
number of internal flaws
number of surface flaws which decrease surface energy

15
 31

Breakage Mechanisms

Four breakage mechanisms

1. Impact
2. Compression
3. Abrasion
4. Chipping

32

Impact/Compression Breakage
Particle shatters into fragments with minimal secondary
breakage (re-breakage)

Size distribution data often fits the Gaudin-Schuhmann

Size Distribution Equation

Wp = cumulative fraction passing size X

K = size modulus
m = ln (Y1/Y2)/ln (X1/X2)
for impact/compression breakage n = 1
e.g. crushing

16
 33

Attrition Breakage - Rod and Ball Mills

abrasion + chipping
abrasion e.g. chalk on board
chipping e.g. off center loading

Size Size
Abrasion Chipping

34

Bond Work Index

Power Draw vs. Product Particle Size & Throughput
Bond measured grindabilities of various ores using a batch
mill in closed circuit with a screen. For ball mills he
maintained a circulating load of 2.5 and for rod mills 1.0.
Bond conducted parallel tests using an 8 ft diameter ball
mill and rod mill. Obtained ratio of net power to feed rate.
A plot of Work input vs F80 - P80 produced a straight
line. The proportionality constant is the Work Index. The
Work Index was obtained using an empirical equation for
size reduction in a ball mill.

W = 10 Wi [1/P80 - 1/F80] (kWh/t)

P = W x TPH

17
 35

Therefore to estimate W:
1. Measure Rod/Ball Mill Grindability using the Standard
Procedure.

2. Calculate Work Index using Bonds empirical

equation.

3. Use the Work Index to calculate the Work Input, W.

4. Total Power Required = W x Feed Rate.

5. Estimate size of grinding mill using equation relating

net power per mill versus mill geometry and operating
conditions.

36

Typical Work Indices

Determined by:
Crushability Test
Rod Mill Grindability Test
Ball Mill Grindability Test

Typical Work Index Values (kWh/t)

Bauxite 11
Cement clinker 16
Corundum 33
Dolomite 14
Feldspar 13
Granite 12
Gypsum 8
Hematite 15
Limestone 15
Pyrite 11
Quartz 16

18
 37

Example P Calculation

What is the total power required for to reduce particle

size from F80 = 1 mm to P80 = 50 m for ore with Wi =
15 kWh/t at Capacity = 2500 tpd?

P = (2500/24)*10*15*(1/(50)1/2 -1/(1000)1/2)

P = 1716 kW

38

Energy Efficiency

Trommans andMeech
SelectiveComminution

19
 39

Process Development

HVC: 0.3% Cu
Industry Avg.: ~ 0.8%

40

Mine Life Cycle

All mining projects pass through a series of stages over the
project life:
Exploration
Discovery
Development
Production
Reclamation/Abandonment

20
 41

Pre-Feasibility & Feasibility Studies

Feasibility Study is conducted to determine the economic
and environmental viability of a project and includes the
following sections:

1. Geology and Geological Reserves

2. Mineral Reserves, Mining Plan and Mining Methods

What is the difference between geological and mineable

reserves?

1. Mineral Processing Plant Design

2. Environmental Review
3. Capital and Operating Costs
4. Net Cash Flow
5. Marketing Study

42

Mineral Process Plant Design

- including Plant Expansions and Retrofits

Trade-off studies
Process Selection:
Process Design Criteria
Flow Sheets
Piping and Instrumentation Diagrams
Process Equipment Specification/Selection
Equipment Specifications
Process Calculations
Commissioning and Operations Support

21
 43

Steps of a Mineral Geological Exploration

Development
Project Mineralogical Studies
Mine Design
Lab and Pilot Tests

Engineering Studies

Increasing Project No
Certainty company
Yes
Basic Engineering

Detailed Engineering

Implementation

44

Process Development

Process Development is a blend of science, technology and

economics.

The objective is to develop a process which, when combined

with all other aspects of the project, will optimize the overall
economics of the project.

22
 45

Transforming Rocks into Ores

Rocks or minerals

1. Raw Material
Mineral Processing 2. Technology
3. Market
4. Economic Aspects
5. Environmental &
Social Issues

Ore or Valuable Minerals

Saleable

46

Process Development
The Flow sheet is critical to establish:
Design Criteria
Mass balance
Water balance
Energy balance
Tailing disposal
Plant lay-out
Equipment sizing
COSTS

23
 47

Flow Sheet Development Objectives

Determine processing parameters

Determine mass/water/energy balance of each

unit operation

Evaluate processing alternatives

Establish the list of main pieces of equipment

required

Create the foundation for the conceptual

engineering (feasibility study)

48

Flow Sheet Development Goals

Create a procedure (process) to concentrate minerals with

high recovery (hopefully >90%)

Create a process to obtain concentrates with a high grade

(saleable)

Create an economically feasible process

Create a safe & environmentally sound process

24
 49

Key Information for Process

Flowsheet Definition
Grades (chemical analysis of elements)
Ore-minerals
Gangue-minerals
Mineralogical Factors
Physical Properties
Chemical Properties
Liberation
Process Recovery

50

Role of Metallurgical Testing in Project

Development
Is very project specific, however, like projects have like
metallurgical requirements

Should be based on creating existing, saleable products,

as well as using known technologies

In most projects, early metallurgical testing is concerned

with fatal flaw detection

Detailed project test work should ideally follow after ore

reserve data is secure.

25
 51

Metallurgical Testing
Determine Processing Parameters
Lab Scale Testing
Evaluate and Monitor Process Performance
Evaluate Changes to Process Prior to Implementing
Evaluate Reagents
Equipment
Size and Select Equipment
Process Development
Pilot Scale
Confirm Process Selection
Confirm Scale-Up & Test New Technologies

52

Prefeasibility &
Lab-Scale Feasibility
Testing
Confirm flowsheet & identify
reagents
Establish recovery &
concentrate quality
Study variability (met
mapping)
Determine preliminary
concentrator design criteria

26
 53

Steps to Define Process Flow Sheet

Sample

Pilot Testing Mineralogical Lab Testing

Studies

Mine Plan Process Flow sheet Modeling

54

Rock (full of minerals)

1. Raw Material
2. Technology
3. Market
4. Economic Aspects Ore (with defined ore-
5. Environmental & minerals and gangue
Social Issues minerals)

Crushing and Grinding Comminution

Mineral Separation Processing

Tailing Concentrate

27
 55

Sampling

Mostcritical aspect of any

metallurgical test work.

Needs rigorous planning

Must link the knowledge of
geologists, metallurgists mine
planners and consultants
Courtesy of SGS Lakefield Research

56

Sample Selection
All metallurgical test work is limited to the validity &
representativity of the sample(s) tested

Courtesy of SGS Lakefield Research

28
 57

Representative Sample

Representative sample (head sample): similar to the

mineable material

This sample must be used to define all process parameters,

concentration routes and preliminary costs

Many problems in the mining industry are caused by bad

choice of the head sample for process development

58

Sampling

Sampling to Establish Geological Reserves

Field samples
Drill cores
Bulk sampling (e.g. Trenches, Adits)

Sampling to Develop/Evaluate Process Flow sheet

Representative head samples (bulk samples:
trenches, adits, composite sample from drill holes, etc)
Tailing (provide good information about what is wrong
in the plant)
Concentrates

29
 59

Considerations for Composite Samples

Criteria for composites or met mapping matrix

Rock types(s)
Alteration type(s)
Mineralogy
Head grade
Oxidation state
Mine plan
Unusual occurrences

Courtesy of SGS Lakefield Research

60

Compositesamplesarebest,but

Excessive compositing can

mask valuable
metallurgical response
information and give
misleading conclusions
about actual plant
performance.

Courtesy of SGS Lakefield Research

30
 61

Head Sample Collection

Particle size (too much fines cause problems in tests)

Grade (if its too low, the lab tests are not
representative)

Ore oxidation (affects flotation testing)

Contamination (oil could affect flotation)

Different types of ore in the same mineral deposit (hard

to make a representative composite sample; better test
many different types separately)

62

Process Mineralogy

Micrography of back-scattered electrons

pyrite (Py) grain with

inclusions of
galena (Ga)
sphalerite (Sp)
chalcopyrite (Cp)

Ref. MASc Thesis of Valerie Bertrand, DMMPE-UBC, Vancouver, 1998.

31
 63

Mineralogical Studies

Mineralogy characterizes the physical and chemical

characteristics of the ore-minerals and gangue-
minerals

Mineralogical analyses identify the particle size at

which the ore-mineral is liberated from the gangue

Properties of ore-minerals with respect to the gangue

can be measured/evaluated

These factors largely determine the mineral

processes to be used in beneficiating the ore

64

Mineralogical Studies

Mineralogy: identification and quantification of

minerals to establish concentration and/or leaching
techniques

Mineralogical factors: characteristics and

properties of minerals determine the technological
routes, mineral liberation, impurities, etc.

32
 65

Importance of Mineralogical Studies

Which valuable minerals can be concentrated?

Which contaminants will be in the concentrate (penalties)?

Which technology is available and suitable?

What are the environmental impacts?

Which market the product is suitable for?

How much will be spent to produce saleable mineral products?

66

Mineralogical Factors
Mineral types
Texture
Grain shape
Grain size
Mineral associations
Mineral surface
Inclusions
Crystal Structure
Alteration Products
Physical and Chemical Properties
Porosity

33
 67

Mineral Analysis Methods

Hand Specimens (visual mineral identification)
Rough Chemical Tests (e.g. fizz test)
Physical Tests (e.g. scratch tests)
Polarized Optical Microscopy
X-ray Diffraction
Electron Microscopy (Scanning & Transmission)
Other Techniques: Thermal Analysis, Infrared Spectrometry,
X-ray Photoelectron (XPS), Auger Spectroscopy
Mineral Liberation Analysis

68

Mineral Types
Classes Examples
Native Elements Gold, Au
Sulfides Pyrite, FeS2
Sulfosalts Enargite, Cu3AsS4
Oxides/Hydroxides Hematite, Fe2O3; Goethite, FeOOH
Halides Fluorite, CaF2
Carbonates Calcite, CaCO3
Nitrates Nitratite, NaNO3
Borates Borax, Na2B4O5(OH)4.8H2O
Phosphates Apatite, Ca5(PO4)3(F,Cl,OH)
Sulfates Barite, BaSO4
Tungstates Scheelite, CaWO4
Silicates Plagioclase, NaAlSi3O8-CaAl2Si2O8

34
 69

Impurities and Inclusions

70

Penalty Elements
Typical limits for Cu Concentrate

Pb <6% Its important to know where

Zn <5% are these elements in the
As <0.5% ore:
Ni <0.3%
Sb <0.2% - Forming minerals
F <0.1% - Structure of ore-minerals
Bi <0.05%
Cd <0.05% - Structure of gangue-
Hg <0.01% minerals

35
 71

Specification Cu-concentrate
Escondida Mine, Chile

72

Degree of Liberation

Mineral of interest 0.07 mm Mineral of interest

not liberated liberated

Assuming that the black particles are the mineral of

interest (ore-mineral)

36
 73

Gravity Separation

High degree of
liberation

74

Liberation
Reduce particle size to improve liberation

37
 75

Mineral Liberation and Separation

Recovery and grade are
two measures of
separation performance.

Recovery compares the

quantity of valuable metal
in the product stream with
that in the feed stream.

Grade usually refers to

one stream, such as the
grade of the concentrate.

76

38
 77

MLA

Source: Teck

78

MLA

Source: Teck

39
 79

Grade Recovery Relationship

Recovery (R%)

Grade of the
Concentrate (G)

0 Low High

Mass of Concentrate

80

Grade Recovery Curve

Trade-Off Between Grade
and Recovery:
Grade and recovery are
interdependent for a given feed
composition.

Because liberation is usually

incomplete, even in a well-run
separation unit, there is a trade-off
between grade and recovery.

If the grade of a product

increases, recovery drops. If the
grade decreases, recovery rises.

40
 81

Physical Properties
Specific Gravity (ore reserve estimation, gravity
concentration)
Moisture
Magnetic and electrical properties
Color/shape characteristics
Specific surface area
Degree of friability, hardness, toughness
Particle Size

82

Size Analysis

Rotap with stack of sieves with largest sieve at the top.

Mesh size is the number of openings per square inch (i.e.

larger mesh number corresponds to smaller size).

Tyler Sieves, US Mesh Number, Canadian Mesh Number

Convention is 2 series (successive meshes vary by 2)

41
 US Sieve Tyler Equivalent Opening
83
Size mm in
- 2 Mesh 8.00 0.312
- 3 Mesh 6.73 0.265
No. 3 3 Mesh 5.66 0.233
No. 4 4 Mesh 4.76 0.187
No. 5 5 Mesh 4.00 0.157
No. 6 6 Mesh 3.36 0.132
No. 7 7 Mesh 2.83 0.111
No. 8 8 Mesh 2.38 0.0937
No.10 9 Mesh 2.00 0.0787
No. 12 10 Mesh 1.68 0.0661
No. 14 12 Mesh 1.41 0.0555
No. 16 14 Mesh 1.19 0.0469
No. 18 16 Mesh 1.00 0.0394
No. 20 20 Mesh 0.841 0.0331
No. 25 24 Mesh 0.707 0.0278
No. 30 28 Mesh 0.595 0.0234
No. 35 32 Mesh 0.500 0.0197
No. 40 35 Mesh 0.420 0.0165
No. 45 42 Mesh 0.354 0.0139
No. 50 48 Mesh 0.297 0.0117
No. 60 60 Mesh 0.250 0.0098
No. 70 65 Mesh 0.210 0.0083
No. 80 80 Mesh 0.177 0.0070
No.100 100 Mesh 0.149 0.0059
No. 120 115 Mesh 0.125 0.0049
No. 140 150 Mesh 0.105 0.0041
No. 170 170 Mesh 0.088 0.0035
No. 200 200 Mesh 0.074 0.0029
No. 230 250 Mesh 0.063 0.0025
No. 270 270 Mesh 0.053 0.0021
No. 325 325 Mesh 0.044 0.0017
No. 400 400 Mesh 0.037 0.0015 * 1.0 mm = 1000 microns (m

84

Sieve Analysis Report

Sieve Size Individual Cumulative Cumulative
% Retained % Retained % Passing
Sieve fraction Weight Aperture size
(m) (g) (m)

+210 0.75 210 0.3 0.3 99.7

-210 + 149 6.25 149 2.5 2.8 97.2
-149 + 105 45.51 105 18.2 21 79.0
-105 + 74 63.01 74 25.2 46.2 53.8
-74 + 53 41.80 53 16.7 62.9 37.1
- 53 + 44 13.01 44 5.2 68.1 31.9
-44 + 37 12.50 37 5.0 73.1 26.9
- 37 67.25 26.9
Total 250.08 100.0

42
 85

Cummulative % passing vs. particle size

100
90
Cumulative % passing
80
70
60
50
40
30
20
10
0
20 100 500
Particle size [microns]

86

Particle Size Distributions Equations

Determining PSD is laborious and repetitive

Typical particle size distributions belong to families of

curves, with normal, or log-normal distributions

You can represent data with mathematical functions,

using a small number of parameters.

The coefficients of the mathematical equation can be

used to monitor operations or can be used in models for
process simulation.

The two mathematical functions used most commonly in

mineral processing are the Gaudin-Schuhmann and the
Rosin-Rammler equations.

43
 87

Gaudin-Schuhmann Equation

where, Wp = Cumulative percent passing

X = size in microns
K = size modulus (size at Wp = 100)
(measure of top size)
m = distribution modulus (slope of
log-log plot of Wp vs X)

The coefficients can be determined graphically or from

linear regression. The function is most appropriate for
coarse, crushed material, which has been screened at
some top size.

88

Rosin-Rammler Function

where, Wr = cumulative weight percent retained on X

X = size in microns
a = size at which (100/exp) = 36.8% of
particles are retained
b = constant
(slope of plot of ln ln(100/Wr) vs ln x)

Special graph paper available to plot cum. % retained values

directly on the Y-axis. A line at cum.% retained = 36.8 is
included for estimation of a

Originally developed for coal, but fits many mineral size

distributions very well, especially finely ground material (e.g.
ball mill product)

44
 89

Generated using
Matlab, 2010

90

Particle Size Analysis Methods

Sieving (wet/dry)
Cyclosizer
Coulter Counter (Elzone PSA)
Laser beam diffraction methods (Malvern)
Sedimentation Methods -Andreassen Pipette

The resultant particle size depends on method used

particle size that passes through a sieve
equivalent spherical diameter of a settling particle

45
 91

Particle Size Analysis Methods

Gupta et al, 2006

92

Assay / Chemical Analysis Methods

Wet chemical assays

Fire assays
Atomic Absorption
Inductively Coupled Argon Plasma Emission
Spectroscopy (ICP)
X-ray Fluorescence Analysis

46
 Mineralogical Factors (grain size) 93

Size-Assay Analysis Grain Size

Sieve size Wt retained Au DAu
(mesh) (%) (ppm) (%)
+6 3.31 8.60 10.87
-6 +8 4.50 7.00 12.02
-8 +10 7.80 4.09 12.18
-10 +14 6.70 3.45 8.82 83% of gold
-14 +20 6.75 4.01 10.33 + 48 mesh
-20 +28 8.90 2.70 9.17
-28 +35 9.80 2.93 10.96
-35 +48 8.90 2.54 8.63
-48 +65 6.02 0.81 1.86
-65 +100 6.20 1.36 3.22
-100 +150 7.55 0.98 2.82
-150 +200 6.30 0.82 1.97
-200 +270 5.97 1.12 2.55
-270 +400 6.30 1.13 2.72
-400 5.00 0.98 1.87
Total 100.00 2.62 100.00

Not necessarily coarse: Au can be fine but not

liberated

94

Metallurgical Testing

The behavior of a sample under a well-defined set of

chemical and physical conditions

The technically and economically optimum conditions

for concentration or separation to specific project
requirements, and

The ultimate plant design incorporating well-informed

selections of processing unit operations, equipment
types and sizes, materials of construction and physical
arrangements

McNulty,T.P.,MineralProcessingPlantDesign,PracticeandControl,SME2002

47
 95

Metallurgical Testing
Grain Size Analysis
Assays (grades)
Geotechnical properties
Bond Work Index Determinations
Abrasion Tests
Concentration Tests (Flotation, Gravity, etc)
Leaching Tests

96

Metallurgical Testing
In Bench Scale
Evaluate and Monitor Process Performance
Evaluate Changes to Process Prior to
Implementing
Reagents (quality and quantity)
Equipment Selection Based on Parameters Obtained
in the Bench Scale and Mineralogical Studies

48
 97

Pilot Testing - Purpose

Verify processes in a continuous operation
Identify differences between batch bench and full scale
continuous
batch versus continuous grinding
circuit stability
Provide scale-up information
Bench Pilot Full Scale
Evaluate and test equipment designs and processes
To test conditions on large scale
To evaluate new equipment
Confirmation of material and energy balances, equipment
selection and plant design
Produce adequate sample size for downstream testing

98

Simplified
Process
Flowsheet

Cerro Verde

49
 99

Process Development - Overview

Review pertinent background information
Sampling (representative sample)
Obtain good mineralogical information
Evaluate concentration possibilities/ alternatives (lab
tests)
Determine important process information
Economic evaluation of process alternatives
Process optimization
Pilot testing?
Process plant design and feasibility study

100

Process Plant Design -

Objectives
Reviewthemajorstepsinprocessdevelopment.
Introducethekeydocumentsaprocessdesignengineermustbe
familiarwith.

50
 101

Project Phases

TYPICAL PROJECT PHASES

STUDIES

FRONT-END ENGINEERING

DETAILED ENGINEERING

PROCUREMENT

CONSTRUCTION

STARTUP

Typical Total Project Duration About 2-3 Years

102

Actual Project Phases

STUDIES

FRONT-END ENGINEERING

DETAILED ENGINEERING

PROCUREMENT

CONSTRUCTION

STARTUP

Project Duration: ASAP

51
 103
100
90 A Order of magnitude
80 B Preliminary Feasibility
70 C Bankable standard
60 D Definitive

% Progress
50 Preparation period
40

Project
30
20
10

Definition 0

A B C D
0 10 20 30 40 50 60 70 80 90 100

% Engineering & Design Duration

50

Probable Accu racy W/O contingency

Order of magnitude estimate
40

Preliminary Feasibility Estimate

30
Bankable Standard

20 Definitive Estimate
Mechanical Completion
Project
10 Completion

0
Project Definition

104

Studies / Conceptual Engineering

Developingprojectrequirements,proposing&optimizing
solutions,estimatingthecosts,andevaluatingthe
economicsofaproject
Conceptualworksuchasdevelopingconfigurationsand
materialbalances.
Conceptualengineeringworksuchasdeveloping
preliminarymaterialbalancesandprocessflowdiagrams.

52
 105

Project Phases Front End Engineering

Preliminary Engineering
Completing first-pass process, control systems, and
mechanical engineering design activities.
Developing cost estimates for the project.

Supporting environmental studies and permitting.

What documents would typically be prepared for Front-End
Engineering?
Design Basis and Scope, PFDs, Material Balance,
Material Selection Diagrams, Plot Plan (layout), P&IDs
Equipment Data Sheets, Instrument Data Sheets, Utility
Balances

106

Project Phases Detailed

Engineering

Completing the process, control systems, piping,

structural, and electrical design. Incorporating vendor
information. Procurement of equipment and bulk
items.
What additional documents would typically be
prepared for Detailed Engineering?
Isometrics (piping design), Structural Drawings,
Electrical Drawings.

53
 107

Project Phases - Construction

Building and testing

Engineering responsibilities include:
Supporting construction questions and changes.
Completing field checkout and developing punch
lists.
Supporting testing of equipment and systems (Pre-
operations).

108

Project Phases - Startup

Commissioning and starting unit operation

Engineering responsibilities include:

Operator training
Supporting operations during startup
Monitoring startup and unit operation
Supporting performance tests.

54
 109

Typical Engineering Documents

Process flow diagrams Concrete, steel & building
P&IDs drawings/sketches
Process data sheets Piping drawings/sketches: alloy
Plot Plan/Layout
large dia. C.S., special fabrication
Equipment list
Motor list
Design criteria
Single-line wiring diagrams
Piping Line list
Area classification (electrical)
Equipment specifications and
Electrical equipment specifications
vessel sketches Conduit/cable schedules
Utility requirements Electrical design drawings/layouts
Soils data Instrument list
Design specifications (all Insulation schedules (equipment
accounts) and piping)
Sewer and paving layouts

110

Design Criteria
Set basis for all design and calculations

Criteria cover
life of mine
throughputs
wastes
operating parameters
maintenance schedules
feed properties
product qualities
+++

55
 111

Design Criteria - Example

112

Equipment Specifications - Example

56
 113

P&IDs
Diagram which shows the piping of the process flow
together with the installed equipment and instrumentation
Process and Instrumentation Diagrams (P&IDs):
- Schematic representation of the equipment, piping,
and instrumentation of a plant
- Formal documentation of a plant engineering design.
P&IDs are required by authorities in many areas of the
world.

114

P&ID Development

Input
Process Flow Diagrams
Process Description
Design Criteria
Equipment Datasheet
Instrument Datasheet
Line sizing
Piping Spec

57
 115

P&ID Development - Basic Steps

Conduct The Joint P&id Review

Issue For Hazop Review
Issue For Design
Issue For Construction
Issue P&IDs For Record

Flow Sheet - Ball Mill Circuit

116

58
 117

P&ID Ball Mill

118

Process Design - Summary

Different project stages include:
Studies
Front-end engineering (feasibility)
Detailed Engineering
Construction
Commissioning

At each stage metallurgists create and provide definition for

the project, including the following key documents:
Process flow diagrams (Flow sheets)
Material Balances (Mass Balances)
Design criteria
Equipment specifications
P&IDs

59
 119

COMMINUTION
TECHNOLOGIES

120

Objectives
Understand basic principles of comminution
Review common comminution equipment
Review common comminution circuits

60
 121

Introduction
Def: Comminution is the size reduction of solid materials through
the application of energy, usually by means of mechanical forces.

Objectives:
To liberate valuable minerals from waste prior to concentration

To increase surface area available for chemical reaction (e.g. lime,

leaching processes)

To produce minerals particles of required size and shape (e.g.

industrial mineral products)

1st stage of comminution: Blasting

122

Comminution Process Significance

Power Requirements
Typically accounts for 30% - 50% of total plant
For hard ores, up to 70%

Operating costs typically 50%

Capital costs 20-50%

Only a small percentage of power actually used for

comminution:
Grinding efficiency may be as low as 1%
Most energy is used transferring heat to the ore.

61
 123

Breakage Mechanisms

1. Impact
2. Compression
3. Abrasion
4. Chipping

Crushing Impact/Compression
Grinding Abrasion/Chipping

124

Crushing Circuits

Crushing:
Relatively Coarse Sizes
Usually include screening equipment
Usually dry process

Three classes:
Primary
Secondary
Tertiary

62
 125

Primary Crushers

Jaw Crusher
Gyratory Crusher
Roll Crushers
Impact Crushers

126

Jaw Crusher

63
 127

Gyratory Crusher

128

Impact Crusher

64
 129

Roll Crusher -
MMD Sizer

130

Secondary Crushers
Jaw
Reduction Gyratory Crusher
Cone Crusher
Hammer Mil
Impact Crushersl

65
 131

Cone Crusher

132

Hammer Mill

66
 133

Impact Crusher

134

Tertiary Crushers

RollCrusher
ShortHeadConeCrusher
HighPressureRollCrusher
ImpactCrusher
HammerMill
FinerReductionGyratoryCrusher

67
 135

High Pressure Grinding

Roll - HPGR

136

HPGR Roll

68
 137

Grinding Equipment
Producing relatively Fine Product Sizes - Usually include
size classification equipment typically hydrocyclones

Tumbling Mills
Autogenous (AG) Mills
Semi-autogenous (SAG) Mills
Rod Mills
Ball Mills
Stirred Mills
Tower Mills
Vertical Pin Mills
Horizontal Pin Mills

138

Ball Mills

69
 139

Horizontal Stirred (Bead) Mills - ISA

140

Circulating Load
OpenCircuit
Feed Product
Comminution

ClosedCircuit
Feed Product
Size
Comminution
Classification

Oversize

Circulatingloadexpressedasapercentageofnewfeed:
CL=100xO/F

70
 141

Sizing Classificaton Technologies

Screens
Staticgrizzlys
Inclinedvibrating
SieveBends

SizeClassifiers
Cyclones
Hydraulic
Rake/spiral

142

Plant Availability
Plant availability is the percentage amount of time the plant
is actually running.

If a plant is designed to produce a set tonnage, a certain

amount of downtime for maintenance must be planned.

Example: 100,000 tpd design at:

90% availability; tph = 100,000 tpd/24 h/0.90 = 4,629 tph
100% availability: tph = 100,000/24 = 4,167 tph

71
 143

Availability Example Ball Mill Screens

Plant Availability 93%

Circulating Load 90%

Ore S.G. 2.75

Screen Operating
Density (wt %
solids) 50%
Ball Mill Screen
Feed % Solids 94%
Ball Mill Screen O/S
% Solids 90%
Ball Mill Screen
Undersize (t/d) 13,500
Screen Deck Sprays
(m3/h) 200

Feed Grade (%Cu) 0.64

Mass Balance

Solids (dry basis) Water Wet basis Copper

t/d Avg t/h S.G. t/h % Solids S.G. m3/h % Cu

144

Myra Fals
- Crushing
Circuit

72
 145

Myra Falls
- Grinding
Circuit

146

Highland Valley Copper

73
 147

Highland Valley Copper AG Mills

148

Highland Valley Copper SAG Mills

74
 149

Cerro Verde
Crushing Circuit

150

Cerro
Verde
HPGR
Grinding
Circuit

75
 151

152

Sampling

76
 153

Definition
Sampling is the process of securing, in either weight or a
sample, a representative fraction / lot for some purpose such
as assaying.

Basic Rule for Correct Sampling

Each particle of ore or concentrate must have an equal
probability of being collected and becoming part of the final
sample for analysis

154

Sampling
Sampling for feasibility
Field samples
Drill core
Bulk sampling
Trenching
Mined sample

Plant /Operations sampling

ROM samples

Head samples

Mill feed

Crusher, mill, cons & tailings samples

77
 155

Why do we want to sample a plant?

Determine material characteristics

hardness, abrasivity, BWI, angle of repose
Assess size distributions
Obtain samples for assay
- Determine Feed, Concentrate Grade
- Mass Balances
- Assess Process Performance (Recovery)
- Estimate Metal Production
- Identify deleterious elements

156

Representative Sample
precision, accuracy and confidence

Repeat 2 Mean Precision

ASSAY

Repeat 1
Accuracy

True Value

Sample

78
 157

Accuracy and Precision

158

Sample Variance
From Statistics, recall that for a set of values y1, y2yn, the
mean value is: n

y i
y i 1
n
and the variance of x is:
Var(Y) = s2 = (Yi Y)2/(n-1)
Where, s is the standard deviation.
n-1 = the degrees of freedom

For several sets of results, the variance of the mean value is:

Var(Y)= s2/n (1)

n - number of sample increments

s - standard deviation associated with

determining Y

79
 159

Confidence Interval
The true mean can be expressed as the estimated mean
plus/minus a confidence interval as indicated in the following
expression.
= Y t,1-V(Y)1/2 (2)

- true mean value

Y- estimated mean value
t,1- - t-statistic at
degrees of freedom, and
probability
(see statistic reference)

Equations (1) and (2), can be used to determine number of

increments for a desired precision

n = [t,1-s/(-Y)]2 (3)

160

Example 1- Precision and Accuracy

Determine the standard deviation and the 95%
confidence interval for the following Au grades.

Assay Au (g/t)
1 5.45
2 4.73
3 4.66
4 5.39
5 4.71
Mean 4.99

80
 161

Example Confidence Interval

V(Y) = S2 = (Yi - Y)2/(n-1)
S = 0.396

Y = Y t,1-V(Y)1/2 [Eq. 2]
= n-1 = 4
= 0.95

from table of t-statistics

t,1- = t4,0.05 = 2.776

therefore,
Y = 4.99 2.776x0.396
Y = 4.99 1.099 g/t Au

162

Example 2 - Confidence
Estimate the number of samples required, at 95%
confidence, to obtain a difference of not more than 0.1 g/t
Au between the true mean assay estimate and the
estimated mean. Assume infinite degrees of freedom.

n = [t,1-S/(-Y)]2 [Eq. 3]

(-Y) < 0.1 g/t Au

from t-statistic table

t, 0,0.95 = 1.96
use calculated S,
S = 0.396

81
 163

Example 2 - Confidence
Therefore, the number of increments required is:

n = (1.96x0.396/0.1)2

n = 60.2 increments or cuts of a stream.

Similarly,
Max. Difference Number of Samples
0.2 g/t 15.1 (15)
0.3 g/t 6.7 (7)
0.4 g/t 3.8 (4)

164

Sample Size for Desired Precision,

Accuracy and Confidence Gys Method

W = C x [d3/2]

Sampling Error Variance determined by Pierre Gy, 1982

2 = Cd3/W
Where,
2 - sampling error variance
C - sampling constant which is a
function of material characteristics.
d - nominal top size, cm
W - sample mass, g

82
 165

Gys Method
Sampling Constant, C
C = fgmL
f - shape factor
g - size distribution factor
m - mineral composition factor
L - liberation factor

Re-arranging the equation, provides an expression

for sample size.
W= fgmLd3/S2

166

f Shape factor

f=1 f = 0.5 f = 0.1 f = 0.2

g size distribution factor

Wide range in size (d0.95/d0.05 >4); g = 0.25

Medium range in size (2 < d0.95/d0.05 <4); g = 0.50
Small range in size (1 < d0.95/d0.05 <2); g = 0.75
Uniform size - pulverized (d0.95/d0.05 = 1); g = 1.0

83
 167

m mineralogical composition factor

m = (1-a) [(1-a)m + ag]/a (units g/cm3)

where a = fractional average mineral content

m = specific gravity of the mineral
g = specific gravity of the gangue

168

l liberation factor

l = (dL/d)0.5
dL= Liberation size
d = 95% passing size

Francois-Bongarcon and Gy (2002) proposed general

form of l=(dL/d)b

- where in the case of gold mineralization, the value of b

is almost always experimentally found close to 1.5

84
 169

Example

Information is given for a copper/gold process.

If W= fgmLd3/S2

What size sample should be obtained for assay?

a. Mill discharge
b. Flotation feed

170

Mill details

Ore type Massive sulphide copper

Feed Rate 2500 tpd
Mill Discharge 95%-0.1 cm
Flotation Feed Size 95% -48 mesh (0.0297 cm)
Flotation Pulp Density 40% solids.
Liberation size, d1 200 mesh (0.0074 cm)
m (CuFeS2) 4.2 g /cm3
g (gangue) 2.7g /cm3
Sampling error <0.01% Cu (0.0289%CuFeS2)

%Cu ore = Atomic weight of Cu x % Chalcopyrite in ore

Mol. Wt. Chalcopyrite

85
 171

a. Find f - Particle Shape Factor

f = 0.5 for most ores, f= 0.2 for Au ores

b.Find g - Size Distribution Factor

select according to ratio d/d
d - top size; d - lower size (5% passing size)
d/d g
>4 0.25
24 0.5
<2 0.75
1 1.00
Mill discharge d = 1000 m d = 100 m
Flotation feed d = 300 m d = 20 m
g = 0.25

172

c. Find m-Mineralogical Factor

m = (1-a) [(1-a)m + ag]/a

From mineralogical analysis, chalcopyrite grade

estimated to be 1%
for a = 0.01
m = 414.3

d. Find L Liberation Factor: L = (dL/d)0.5

Mill Discharge: L = (0.0074/0.1)0.5= 0.272

Flotation Feed: L= (0.0074/0.0297)0.5 = 0.50

86
 173

e. Calculate W sample mass

W = fgmLd3/ S2

Gy recommends a factor of 2 to 3 times larger than

calculated

f. Calculate pulp volume

Flotation Feed @ 40% solids
Volume ore = 24,366/2.7 = 9,024 mL
Volume water = 1.5 x 24,366/1.0 = 36,549 mL

Total Volume = 45.57 L

174

Comparison od Sample Sizes

Parameter Mill Discharge Flotation Feed

d (cm) 0.1 0.0297
S2 8.35 x 10-8 8.35 x 10-8
f 0.5 0.5
g 0.25 0.25
m 414.3 414.3
L 0.272 0.499
W (kg) 168.6 8.12
Gy x factor of 3 505.8 24.4

87
 175

If the desired copper assay to be within +/- X% Cu, how

do you calculate desired value of s?
a) Select desired sampling accuracy e.g. +/- 0.05% Cu, 95
out of 100 times

b) For a probability of 95%, the number of standard

deviations needed to give the above confidence level
is1.96 (For 99%, it is 2.576)

c) Then 1.96s = ao/a

where ao is desired sampling accuracy (0.05% Cu)
a is the Cu assay in the ore

176

Coning and Quartering

88
 177

Jones Riffle splitter

178

Jones Riffle splitter

89
 179

Jones Riffle splitter for -1 cm

180

Rotary splitter

90
 181

Sampling system

Regular increment
Uniform speed
Normal direction
Proper cutter
Minimize error

182

91
 183

Crushing

184

Overview - Crushing

Applications
Crusher features & terminology
Primary crushing
Secondary and tertiary crushing
Crusher selection

92
 185

Crushing & Screening Part 1

Introduction
Terminology
Crushing Principles
Crusher Types

I 186

Introduction why do we crush?

Improve material handling characteristics

Generate products of a particular size fraction e.g.

aggregates

Prepare for downstream processes increase surface

area, reaction rate, match feed properties

93
 187

Applications

188

Preparation of heap leach dumps

Primary Gyratory Crusher

Open Pit
(oxides)

Conveyor

O/S

Screen

Heap Leach Heap

To leach or U/S
Pressure OX

94
 189

ROM/Overland conveyor

Primary Crusher

Mill Feed Bin

Secondary Crushers

Prepare mill feed product

190

Terminology
Terms
F80 80% passing size fraction in feed F80
P80 80% passing size fraction in
product
Gape
Gape feed opening dimension
OSS maximum jaw gap at discharge
CSS- minimum jaw gap at discharge
Throw = OSS - CSS
Mechanical reduction ratio = Gape/OSS
(jaw)
CSS
or Gape/CSS (cone/gyratory) OSS
Particle reduction ratio = f80/p80

P80 P80

95
 191

Terminology

Frame
Pitman
Jaw
Bowl
Mantle
Toggle
Eccentric
Liners

192

Types of Crushers
Jaw Crusher
Gyratory crusher
Cone crusher
Standard
Short-head
Roll crushers
Single roll
Double roll
Impact Crusher (interparticle crusher)
Hammer mills

96
 193

Features
Jaw Crusher
Spec by throat dimensions
eg 80x60in
Sized by max particle size
in feed
Gape 440-1200mm
Feed ~ 80% gape
OSS > P80 > CSS
Reduction ratio ~ 5:1 max
45-250 kW
10-1600 tph
Prefer blocky, coarse
material, can be wet
Robust, simple, compact
design
Manual or semi-auto
operation

194

Jaw Crusher

97
 195

Features
Gyratory Crusher
Specify by Gape/Mantle
dimension e.g. 60x102in
Sized by throughput
Gape 0.7-2,5m
Max feed size 80% of
gape
P80 ~ OSS
Reduction ratio ~ 8:1 max
500 7000 tph
200 1000 kW+, mantle ~
100rpm
Can accept wide range of
feed types
Expensive and complex
vs. jaw, but higher
throughput
Manual or auto operation

196

Gyratory Crusher

98
 197

Gyratory Crusher

198

Cone Crusher
Features
Spec by mantle diameter, e.g. 6ft
Sized by product spec & throughput
F80 - 50% of mean gape
P80~CSS (fine)
- Theoretical reduction ratio can be
13:1 max, prefer 3:1
- 90-650 tph typical
- 45-350kW , up to 750kW
- Compact but complex, higher shaft
speeds, finer applications only
- Automatic operation only

99
 199

Roll Crushers MMD Sizer

Features
Softer materials (coal and
chrome)
In-pit and underground sizing
for material handling
High throughputs in soft
material
Single- or double rolls
Allows fines to fall through
F80 0,3 2m
Coarse product profile, low
reduction ratio

200

Impact Crushers and Hammer Mills

Interparticle crushing
Softer materials
Lower throughput, 5 100 tph
Low wear / unit throughput
Require dry, regular sized feed

100
 201

Crushing Part 2

PrimaryCrusher
selection
Duty&Capacity
Feedcharacteristics
Productrequirements
Workedexamples

202

Primary Crusher
Duties
Primary crusher feed variable tonnage, topsize, size
distribution
Product requirements not usually strict
Typically prepare feed for conveying, stockpiling, or feed
preparation for secondary crush
Capacity dependent on feed size, Work index, crusher size,
speed, throw, CSS

101
 203

Selection parameters
Duty P80
Feed arrangement Work Index
Location Abrasion Index
Topsize Hardness
F80 Product size distribution
Fines/Clay preferred sizes
Throughput Discharge arrangement

204

Crushing Principles
Understand feed characteristics
Throughput (tons per hour)
Size distribution (f80)
Work Index (kWh/ton)
Abrasion & Hardness (Mohs or or mass-loss test)
Moisture content (%H2O by mass)
Understand desired product properties
Number of products (separate size fractions)
Size distribution p80 = f80 to next process
Crushing is a route from one state to the other

102
 205

Crushing Principles
Power Estimate

From Comminution Theory:

Wi = 10BWi(1/p80-1/f80)
Where Wi = specific work index
BWi = Bond Work Index for material

Power = k (Q x Wi )
Where k = 0,75 (primary), 1 (secondary)
Q = throughput

Apply safety factors for surge, feed size variation, environment

~ maybe 25-30%

206

Crusher Selection

103
 207

Typical Crusher Ranges

208

Preliminary Crushing Sizing

1. Estimate Free Run in Feed (if screened prior to
crushing)
2. Estimate Crusher Capacity
3. Estimate F80 and P80
4. Estimate Power Requirements
5. Determine top size
6. Select crusher

104
 209

Jaw Crusher Selection

Example: See Metso Handbook:

Quartzite BWI ~ 12 kWh/t Gape = Topsize/80%
Throughput = 200 tph CSS < p80 < OSS
f80 = 300 mm, topsize Choose crusher
450mm
Feed decline conveyor Check capacity
Discharge to secondary cone Size motor
~ f80 = 80mm

210

Jaw Crusher Section

Jaw
CSS ~ P80 ~ 80 mm
F80 300000um Top size, 450 mm
P80 80000um Top size = 80% gape
Q 200t/h Gape = 563 mm
BWi 12kWh/t
Wi 0.205kWh/t Metso C-Series Jaw
k 0.75 Crusher:
SF 1.3 C106
P 40.0kW

105
 211

Typical arrangement

212

Features Gyratory Crusher

Specify by Gape/Mantle
dimension e.g. 60x102in
Sized by throughput
Gape 0.7-2,5m
Max feed size 80% of gape
P80 ~ OSS
Reduction ratio ~ 8:1 max
500 7000 tph
200 1000 kW+, mantle ~
100rpm
Can accept wide range of
feed types
Expensive and complex vs.
jaw, but higher throughput
Manual or auto operation

106
 213

Preliminary Crushing Sizing

1. Estimate Free Run in Feed (if screened prior to

crushing)
2. Estimate Crusher Capacity
3. Estimate F80 and P80
4. Estimate Power Requirements
5. Determine top size
6. Select crusher

214

Gyratory Crusher Selection

Example:
Copper Porphyry Ore BWI ~ 15 kWh/t
Plant design throughput = 50,000 tpd
Crusher operating time = 7 x 2 x 8 hr shifts/week.
F80 = 420mm, top size 1200mm
Feed = haul truck
Discharge overland conveyor ~ 150mm

107
 215

Gyratory Selection - Example

F80 420000um
P80 150000um
Q 3125t/h
BWi 15kWh/t
Wi 0.156kWh/t
k 0.75
SF 1.3
P 475kW
Top Size 1200mm
Gape 1500mm

216

Gyratory Crusher Capacities

Source: Metso Crushing Handbook

108
 217

Gyratory Selection - Example

F80 420000um
P80 150000um
Q 3125t/h
BWi 15kWh/t
Wi 0.156kWh/t Metso: 62-75
k 0.75 Gyratory, increase
SF 1.3 availability or move
P 475kW to 165 OSS.
Top Size 1200mm
Gape 1500mm

218

Typical Arrangement

109
 219

Summary: Jaw vs Gyratory

Jaw Gyratory
Lower max capacity, f80 Highest capacity, f80
than gyratory
Complex, robust, expensive,
Compact, robust, cheap
but low cost/tph
Must screen out fines
Can accept high fines ratio
Prefer reduction ratio ~ 3:1
Better reduction ratio
Limited by feed
arrangement Accepts all feed methods
Discharge arrangement needs
care high tph

220

Screening:
Features, Design

110
 221

Objective

Understand principles of screening

Review screening equipment
Learn how to size a screen.

222

Purpose of Screening
Definition:
- Screening is a mechanical process which accomplishes a
separation of particles on the basis of size and their
acceptance or rejection by a screening surface.

Prepares products of appropriate sizes for downstream

process or final sale.

Efficiency is determined by the perfection of separation

based on the aperture size.

111
 223

Screening

Effective from 300mm to 40m

Less efficient at finer sizes
Typically:
Dry screening >5 mm
Wet screening <250m

224

Screening Applications

Metso

112
 225

Screening Applications
Scalping(oversizerejection)
Sizing
Oversize/recycle
Intermediatesizes,feedsplitting
Finalsizing(millproductscreens)
Feedpreparation
Densemediarecoveryscreens(DrainandRinseScreens)
Dewatering/desliming
Trashremoval

Screening Theory 226

ScreenBed

113
 227

Screening Principles
A - feed zone
B - stratification Saturated zone
C - separation

C B A

f, a

Oversize Undersize

228

Particle flow rate through deck

related to screen length
Zones
a. Feed
b. Stratification
c. Separation

114
 229

Screening Mass Balances

F=200 t/h
fx = 0.7
ox = 0.2
O?
U?

230

Mass Balance Example

F=O+U
Ffx = Oox + U

If F = 200 t/h, fx = 0.7, ox = 0.2, determine O, U.

O = F(1-fx)/(1-ox)
O = 200(1-0.7)/(1-0.2)
O = 75 t/h
Solve for U
U = 125 t/h

115
 231

Screen Efficiency

Undersize Removal Efficiency in Oversize

F (1 f x )
Eu (1 ox )
O (1)

Example. Eu = 80%
Efficiency of Undersize Recovery

U f x ox
Ru
(2)

x f x (1 ox )
Ff
Ex. Ru = 89%

232

Types of Screens
Vibrating Screens
Inclined,
Grizzly,
Horizontal,
Dewatering,
Banana screens
Static
Self cleaning grizzly
Trommel
Linear

116
 233

Scalping screen
Screen Types
Multi-deck screen

Linear Screen
Trommel

234

Features

Feed plate Side

plates

Flow

Frame
Drive
Top deck
2nd deck
3rd deck

117
 235

Screen Surfaces/Medium
Surface Characteristics:
Must withstand stress and loads, and be abrasion and
corrosion resistant.
Materials:
Monel, stainless steel, abrasion resistant high carbon steels,
rubber, and reinforced polyurethane.
Best surfaces provide:
-Required opening size and capacity
-Wear resistances
-Minimum replacement cost per unit of throughput

236

Woven wire cloth -all sizes: Media Selection

Rail grizzly bars Poly panels wear

and corrosion
- coarse sizes: resistant, medium fine
to medium coarse

Wedge wire
- fine and
difficult
screening duty

118
 237

Screen Aperture Shapes

Square: coarse applications, accurate sizing

Rectangular / Parallel: - higher capacity (higher area),

less susceptible to blinding, suited to needle shape
particles, good for high moisture ores with clay.

Rectangular / Perpendicular - less blinding for dry

screening, longer screen life, higher efficiency.

238

Screen Surfaces

Woven Wire - Traditional

Profile Wire/Bar
Parallel to flow used for coarse screening
Perpendicular to flow used for wet fine screening,
desliming and dewatering.
Perforated plate
Pros: high wear resistance, less blinding, higher
efficiency, higher accuracy.
Cons: more expensive, less open area
Polyurethane/rubber screens now standard:
less expensive, robust

119
 239

Influence of Variables on Screen

Performance
Screen Area/Open Area
Effective Area < Actual Area
Capacity screen area

For a given area

Capacity width
Efficiency length

Length is usually 2 to 3 times width

Best capacity and efficiency when solids 1 particle layer in

depth at end of screen

240

Influence of Variables on Screen

Performance
Aperture Size/Shape
aperture size, capacity
aperture size, efficiency
aperture size, blinding

Slope
slope, capacity
slope, effective aperture size
slope, Constant Efficiency up to Critical Slope, then .
Typical Slopes: 20 - 25

120
 241

Influence of Variables on Screen

Performance
Deck Motion (Speed/Throw)
Purpose: To lift material causing stratification and
conveying of particles.
Vibration: Inflow vs. Counter Flow, circular/elliptical motion
Amplitudes: Typical 3-15 mm
Frequency: Normal 700-1000 cycles/min
Frequency: High Speed 3600 cycles/min

242

Influence of Variables on Screen

Performance
Amplitude:
Too small allows blinding
Too large reduces efficiency
Too large reduces bearing life

Elliptical Circular Motion

In flow increases capacity
In flow may decrease efficiency
Counter flow decreases capacity
Counter flow increases efficiency
Counter flow may increase blinding

121
 243

Influence of Variables on Screen

Performance
Speed:
High speed used with small throws, small particles
Low speeds used with large throws, large particles

244

Properties of Feed Material

Particle Size/Shape/Distribution
At fixed screen opening particle size, Capacity

Near size particles = 0.5 to 1.5 of screen size.

Amount of near size is rate determining
near size, capacity, blinding, efficiency
To maximize capacity, exposure of fines and near size
to screen

Use upper screen deck to reduce oversize, ensure good

stratification, optimize throw

122
 245

Properties of Feed Material

Moisture Content
Moisture + Clay leads to agglomeration and blinding
In severe cases:
can heat wire screen,
Switch to wet screen
Add rubber ball tray under screen

246

Properties of Feed Material

Feed Rate/Bed Depth

Bed depth = function of (Feed rate, slope, size distribution,
circulation direction)
Bed depth increases with increasing feed rate
Screen width selected to maintain bed depth at discharge,
therefore screen width determines capacity.

123
 247

Screen Sizing and Selection

Two methods presented (many more exist):

1. VSMA Screening Surface Area Calculation,

Developed collaboratively by VSMA, screen
manufacturers to ensure consistency and compatibility of
screening equipment. Based on theoretical surface area
of a perfect screen in an application.

2. Metso Handbook

248

Screen Sizing and Selection

Screening Area = US
A x B x C x D x E x F x d x SF
Where:
US = undersize tonnage (t/h)
A = basic capacity (m3/h/m2)
B = oversize percentage factor (of the deck concerned)
C = efficiency screening factor
D = halfsize percentage factor (of the deck concerned)
E = efficiency screening factor for wet screening
F = deck factor
d = bulk density (t/m3)
SF = free area factor
In calculating SA, other factors are important:
M = split (mm)
OS = oversize tonnage (t/h)
HS = half size tonnage (t/h)

124
 249

Screening Unit
Efficiency and Capacity
Select:

Factor A = basic capacity for woven wire cloth

Factor E = efficiency screening factor for wet screening
Factor C = screening efficiency factor
normal screening C = 1
high efficiency screening C = 0.8
light screening C = 1.2
Factor F = deck factor
1st deck F = 1
2nd deck F = 0.9
3rd deck F = 0.8
4th deck F = 0.7

250

Oversize and
Undersize
Factors

125
 251

Free Screening
Area and Efficiency

252

SF
Surface
Factor

126
 253

Check Bed Depth

If the motion is with
flow: Bed depth at end of deck :
s = 1100 m/h = 0,3
m/s
D= OS
If the motion is
lds
counterflow:
Where
s = 800 m/h = 0,22 D = bed depth (m)
m/s OS = oversize tonnage (t/h)
l = screen width (m)
Average screen d = bulk density (t/m3)
Dry process, we s = material travel speed (m/h)
must have D < 4 x
aperture
Wet process, we
must have D < 6 x

254

Choosing media aperture

and material for application
Relief deck
can be required for 2 reasons :
to have a smaller bed depth at the split
considered
because of the excessive feed size falling on
the deck considered (see table).
Important :
Data based on observations of field life of
screen deck
Increase of in size 20% is suitable for gravels
Coarser feed at smaller apertures requires
poly decks
Coarser feed at larger apertures requires
perforated plate or grizzly decks

127
 255

Correcting for Moisture

Moisture = H20%
This factor influences the efficiency of fine screening. When the split (M) is
less than 10mm, we consider :
If H20% < M Use conventional woven mesh
8
If M < H20% < M Use stainless or wedge wire cloths
8 4

If M < H20% < M Use anti-blinding stainless cloths

4 2

If M < H20% < M Use ani-blinding or self cleaning deck

2

If M < H20% Wet process is required

*Note - if there is clay content, screening capability must be checked in

laboratory.

256

Screening:
Sizing and Selection

Primary Source: AJ Gunson

128
 257

Objective

To review the Metso screen sizing method.

To size a screen through an example problem.

258

Reference:
Screen
Conversions

129
 259

Screen Sizing Data

a) Features of material to be c) Product separation ranges
screened: d) Desired efficiency
Density e) Type of job:
Maximum feed size Washing
Product granulometry Final classification
Particle shape Intermediate classification etc.
Moisture content f ) Any room and weight
Presence or lack of clayey limitations
material g) Degree of knowledge of the
Temperature etc. material and desired product
b) Capacity

260

Screen Area

Screen area determination (Metso):

Qu S
Area
A B C D E F G H I J K L

Qu = t/h undersize in the feed

S = Safety factor (1 to 1.4)
A = Screen capacity for required size (t/h/m2)
B to L = Screen Area Modifying Factors

130
 261

Screen Sizing Factors:

A Screen Capacity

Metso

262

Screen Sizing Factors:

A Screen Capacity

Metso

131
 263

Screen Sizing Factors:

B Oversize Fraction Factor

Metso

264

Screen Sizing Factors: C

Metso

132
 265

Screen Sizing Factors

D: Position of screen deck (from top) factor

E: Wet screening factor, desired separation size.

(1 if dry screening)

F: Material weight factor (can be graded)

Metso

266

Screen Sizing Factors

G: Open surface factor Actual _ open _ area%
G
50%
H: Shape of screen surface opening factor

I: Particle shape factor

J: Screen efficiency factor (%)

Metso

133
 267

Desired separation
size vs. actual required
screen size
Due to screen slope, actual
screen size must be larger
than the desired separation
size.

3% to 5% of the screen
undersize may be slightly
greater than the specified
size this difference is taken
into account in the sizing
factors and does not need to
be separately calculated.

268

Screen Free Open Area: A

134
 269

Screen Free Open Area: B

270

Screen
efficiency,
based on
screen
loading

Metso

135
 271

Screen Sizing Factors

K: Screen type factor

L: Feed moisture factor (% moisture by mass)

Metso

272

Screen Sizing
Width & Bed Depth
Q
Width: B
3.6 v d
Where,
B = nominal screen width (m)
Q = oversize (discharge) capacity (m3/h)
Not (t/h): Typical bulk density, 1.6 t/m3
d = material layer thickness (mm)
V = material transport speed (m/s)
Dry process, d should be < 4 x separation size
Wet process, d should be < 6 x separation size

136
 273

Screen Sizing
Material Transport Speed

Metso

274

Screen Sizing
Recommended Feed Bed Depth

Metso

137
 275
Screen Sizing:
Recommended Discharge Bed Depth

Metso

276

Typical Screen Models

Metso

138
 277

Screen Sizing Example

a) Features of feed material:

Mesh (mm) 100 25 13 10 5
Density: 2.7 t/m3 % Passing 100 75 45 30 22
Max. feed size: 100mm
Product granulometry
Particle shape: flaky
Moisture content: 3%
Presence or lack of clayey material
b) Capacity: 380 t/h

278

Screen Sizing Example

c) Product separation ranges

+25mm, 10 to 25mm, -10mm
d) Desired efficiency: 90%
e) Type of job:
Dry screening
Dual sloped, variable elliptical screen
Inclined screen at 20 degrees, circular motion (coarse)
f ) No room/weight limitations
g) Degree of knowledge of the material and desired product

139
 279

Problem Solution
Set up mass balance 380 t/h

Mesh (mm) % Passing t/h (Cumm.) Size Fraction t/h (fraction)

(Cumm.) 95 t/h
100 100 380 +25 95

25 75 285 -25+13 114

13 45 171 -13+10 57 285 t/h

10 30 114 -10+5 30.4 171 t/h

5 22 83.6 -5 83.6

114 t/h
1st Screen:
Qu =285 t/h; Assume Safety Factor S=1.0
A = 54 t/h/m2 (Either Factor A chart)

280

Problem Solution
B = 1.35, from Factor B chart.
%O/S in feed = 100% - 75% = 25% @ +25 mm.
C = 1.1, from Factor C chart.
%U/S the opening size: Opening size: 25mm
% passing 25/2 ~ 13mm: 45%
D = 1, first deck.
E = 1, dry screening.
F = 1, solids density = 2.7 t/m3

140
 281

Problem Solution
G = Open Surface factor
Assume a heavy square hole, which is appropriate for the
size range and the flaky material.
From the Desired separation size vs. actual required
screen size table, for at 25mm product size, the screen
size must be between 27-30mm, or 28.5mm, or around 1
1/8". For a 1 1/8" heavy screen, the actual screen free open
area is 61%.
G = actual open area/50% or 61%/50% = 1.22
H = Screen surface opening factor, or 1.0 for square
openings.

282

Problem Solution

I = Particle shape factor, or 0.9 for flaky particles (from table).

J = Screen efficiency factor, or 1 assuming standard 90%
efficiency (from table)
K = Screen type factor, or 1.3 assuming a dual slope variable
elipitical screen (from table)
L = Feed moisture factor, or 1.0 assuming 3% moisture (from
table)

141
 283

Problem Solution

Qu S
Area
A B C D E F G H I J K L

A1 = (2851)/(541.351.11111.210.911.31)
A, deck 1 = 2.5 m2

A2 = (1141)/(330.90.790.9111.0410.911.31)
A, deck 2 = 4.4 m2

284

Problem Solution B
Q
Width & Bed Depth 3.6 v d
Reviewing the typical screen sizes, the Metso 5 x 12
model meets the minimum area required for the 2nd deck
(25mm & 10 mm). Bed depth.

Variable Deck 1 Deck 2

D is less than 4 x
B (m) 1.5 1.5
separation size for
O/F (t/h) 95 175
both decks, 5 x 12
Bulk Den 1.6 1.6
(t/m3)
model passes depth
Q (m3/h) 59 107
test, screen size is
v (m/s)* 0.58 0.58
adequate.
d (mm) 19 34

*Inclined Screen at 20, circular motion, coarse classification

142
 285

Crushing &
Screening:
Plant Design

Primary Sources: AJ Gunson, B Klein

286

Outline

Screen Efficiency and Circulating Loads

Factors Affecting Crusher Design
Crushing Plant Design Procedure
Flow Sheet Examples
Design and Layout
Design Criteria, Operability & Cost
Operation & Control
SAG vs. HPGR

143
 287

Screen Efficiency
Undersize Removal Efficiency in Oversize

F (1 f x )
Eu (1 ox )
O (1)

Efficiency of Undersize Recovery

U f x ox
Ru
(2)

x f x (1 ox )
Ff

288

Crushing Circuits
Closed Circuit A C=F U
D, dx F, fx Product
Crusher Size
C, cx Classification

Oversize O, ox
cx=% passing x in C
D = tph fresh feed
Closed Circuit B
dx = % passing x in D
F = t/h screen feed
F, fx Product fx = % passing x in F
Feed D, dx
Size O = tph screen oversize
Classification U ox = % passing x in O
C, cx
U = tph screen undersize
O, ox C = tph crusher discharge
Oversize
Crusher
C=O D = U at steady state

Circulating load expressed as a percentage of new feed

CL = 100 x O/D

144
 289

Screen Efficiency and

Circulating Loads
Circuit A
As F = O + D, substitute in efficiency eqn 1 & solve for O/D
O 1
1
D Eu

1 f
x

As D = U, substitute in equation 2 and solve for O/D

O 1
1
D ( Ru f x )

290

Screen Efficiency and

Circulating Loads
Circuit B: As D = U, C = O
F(1-fx) = D(1-dx)+O(1-cx)
Substitute in equation 1 & solve for O/D
O (1 d x )

D Eu (1 c x )
Similarly: Ffx = Ddx+Ocx

O 1 Ru d x

D Ru c x

145
 291

Factors Affecting
Crusher Design
Plant throughput / availability
Desired product size for downstream process
Ore Characteristics
Size distribution
Moisture content
Density
Crushability
Abrasiveness
Climatic Conditions

292

Crushing Circuit Planning

Choose flow sheet and select equipment sizes for efficient

metallurgical performance at designed capacity

Ensure good access for maintenance and that future

expansion or modification can be carried out without
difficulty

Plan for minimum capital and operating costs, while

allowing for efficient metallurgical and mechanical
performance.

146
 293

Crushing Plant Design Procedure

Know the feed size and tonnage, product size reqd
Choose # of stages of crushing (reduction ratios!)
Decide open or closed circuit at each stage
Draw flow sheet. Check your logic!
Select crushers
Optimum crusher open/close side settings
Estimate product stream characteristics (tph, P80)
Select screens
Estimate product stream characteristics (tph, P80)
Determine if parallel circuits are required
Determine capacity of surge bins and conveyors
Size motors
Draw up equipment list
Remember its an iterative exercise.

294

Crushing Reduction Ratio

147
 295

Crushing Reduction Ratio

Metso

296

Circuit Design & Layout

Feed method and orientation

Material handling criteria set chute/bin angles and opening
sizes (angle of repose & 3 x max size)

Conveyor length ~ feed height

For bins, feed height ~ bin capacity

Check recycle streams, conveyor or structural clashes

Check centre centre distances

Review maintenance access (personnel, cranes & hoists)

148
 297

Design Criteria, Operability and Cost

Multi stage crushing usually required

Open circuit means less size control
Closed circuit usually means larger capacity units
Scalping prior to a crusher reduces unit size
Crusher, chutes, feeders & conveyor size ~ particle size
Surge capacity > cost, increases availability
Unit capacities must account for tonnage, size, grade, moisture
and operational variabilities
Good maintenance facilities >cost, but < downtime
Design last crusher circuit running @ 100% load.

298

Crusher Circuit Design Basics -1

Multi-stage crushing usually required more efficient and typically
a single crusher cannot provide required reduction ratio.

Closed circuit of final stage necessary, of earlier stages may not

improve efficiency

Crushers, feeders & surge bins need to be able to handle largest

rock to stop bridging (3-5 x top size)

Capital & power costs per ton at same closed side setting does not
decrease significantly with crusher size feed size and capacity
more important than capital and power costs oversize
equipment.

149
 299

Crusher Circuit Design Basics -2

For a given crusher, power draw increases linearly with
feed rate.
At a given power draw, product size with throughput.
fines leads to throughput, power
Steady feed leads to throughput surge capacity
important
Primary crushers have intermittent feed, so need to be
oversized.

300

Crusher Cavity Operation

150
 301

Operation and Control

Maintain product size and throughput targets, maximise

throughput or optimise size
Equipment health!
Electrical status (on/off/trip/emergency stop)
Temperatures and pressures
Interlocks startup, shutdown and safety
Monitor system variables vs. design criteria (throughputs,
levels, densities etc.)

Operation and Control Variables 302

PRIMARY CRUSH STOCKPILE/BIN

-Motor on/off + power -Level
-If hydraulic then hydraulic healthy, CSS, -Feed in
Tramp warning -Feed out
-Feeders
on/off
1
SCALPING GRIZZLY -Feed rate
-on/off (v/s)
-New feed rate 2 12 (
3
SECONDARY
SCREEN
11 14 -Motor on/off
4
13

CONVEYORS GENERAL
-motor on/off
-belt condition
-maybe variable speed 9 5
-pull cord

6
8 SECONDARY CRUSH
SIZING SCREEN TERTIARY CRUSH -Motor on/off/power
-Motor on/off 7 -motor on/off -Cavity level
-Cavity level -CSS
-CSS -Hydraulic status
-temperatures
-hydraulic status
10 -temperatures

151
 303

Flow Sheet Examples

High capacity crushing (iron ore)

Coal plant feed preparation
Hard rock crushing circuit
HPGR vs SAGB

304
MID SOUTH IRON ORE, SA: 10 000 TPH
-1000mm
IN-PIT
GYRATORY
1

2 3

OVERLAND CONVEYOR
4

5 6

7
SECONDARY
SIZING SCREEN CRUSHING
8
(-30mm) (P80 80mm)

11

TERTIARY CRUSH
9 (P80- 25mm) FEED BINS

10

152
 ROM PIT FEED 305
-? @ 500 TPH OVERSIZE DUMP
PRIMARY
JAW
1 (P80 = 80mm)

STATIC SECONDARY 3
GRIZZLY 9 SCREEN (-80)
(-400mm)
2
RECYCLE CONVEYOR 10

TAVISTOCK COLLIERY, SA: 500 tph SIZING

SCREEN
(-80+20;
WASH WATER -20+3)
4
U
-20 + 3

DMS 12
CYCLONE 7
5
FEED PREP
SCREENS -80 + 20
(-1mm) U
DMS
DRUM 11

6 U = -1mm UNDERSIZE & FINES TO MILL & SPRIALS

8
U

SCALPING GRIZZLY 306

(-120mm)
1 STOCKPILE
-400mm @ (12000 T)
350 TPH
FROM 12 SECONDARY
2
U/G) 3 SCREEN (-40)

11 14
4
PRIMARY CRUSH 13
(P80 100mm)

MESSINA PLATINUM, SA 2000 tpd

RECYCLE 9 5
CONVEYOR

6
SIZING SCREEN 8 TERTIARY CRUSH SECONDARY CRUSH
(-12mm) (P80 12mm) (P80 40mm)
7

10

153
 High Pressure Grinding Roll Technology

HPGR at Boddington Gold

(http://www.womp-
int.com/images/story/2009vol10/13a.jp
g)

Presentation Outline
1. HPGRIntro&History
2. MainComponentsandWearItems
3. TestingandSizingFactors
4. Flowsheets andApplications

154
 Typical HPGR Comminution Duty

HPGR

History of HPGR Technology

Comminution method was patented by Dr. Schnert in 1979
First HPGR installed in a cement application in 1985
HPGRs became established in the cement industry due to
recognized energy benefits
1987 - HPGRs first applied in the diamond industry
~1995 Unsuccessful trials in hard rock applications (eg.
Cyprus Sierrita)

155
 History of HPGR Technology
1995 till present- HPGRs installed in hard rock applications
due to Improvements in roll wear linings and gaining
momentum (more than five vendors participating in the
market)
2012, Expiration of studded lining patent. Increase in HPGR
vendors, now including CITIC/KHD, Polysius,
Koeppern/Outotec, Metso & FLSmidth

HPGR - Function

156
313

314

157
HPGR Main Components

HPGR Edge Effect

158
317

318

159
 HPGR Wear Components
Studded Lining (~2000 to ~10,000 hours)

Cheek Plates (~1500 hours)

from Weir Minerals Brochure

courtesy of Koeppern Machinery Aus.

ROLL SURFACE - STUD LINING Wear Parts Roller Changeout

Hart et al (SAG2011)

Newmont roll changeout

Koski et al (SAG2011) Hart et al (SAG2011)

160
 321

HPGR Test Work and Sizing

Test Work Carried out to Determine:
HPGR Sizing Parameters
Suitable specific pressing force
Specific throughput Mdot
Net specific energy consumption (kWh/t)
HPGR operating gap / Feed top size
Flake density

Process Flowsheet Parameters

Size reduction
Influence of feed parameters
on HPGR comminution
Influence of transfer size
and circuit configuration

161
HPGR Test Work
and Sizing

Agglomerated HPGR product (Flake)

324

162
325

326

163
327

328

164
329

330

165
331

332

166
333

334

167
335

336

168
 Existing & Upcoming Operations
Project Company Location HPGR TPD Ore Op.
s Type Since
Cerro Verde Freeport Mc. Peru 4->12 120 -> Copper 2006
360 ktpd Porphyry
Grasberg Freeport Mc. Indonesia 2 ~70ktpd Copper, 2007
Gold
Mogalakwena Anglo Platinum S. Africa 1 ~25ktpd Platinum 2008
Boddington Newmont Australia 4 ~100ktpd Gold, 2009
Gold Copper
Penasquito Goldcorp Mexico 1 ~+100 ktpd Poly- 2010
(peb. metallic
crusher
circ.)
Salobo Vale Brazil 2 ~33ktpd Copper, 2012
Gold
Sierra Gorda KGHM/Sumitomo Chile 4 ~110ktpd Copper - 2014?
Moly
Morenci Freeport Mc. USA 1 -> 115ktpd Copper 2014
Porphyry

Reported Benefits of Using HPGR

Energy efficiency

Reduced steel consumption (in

comparison to SAG milling)
Courtesy of Koeppern Machinery Aus.
Not sensitive to ore variability (in
comparison to alternative
comminution equipment)

Breakage along grain boundaries

(promoting liberation)

169
 Reported Disadvantages of Using HPGR
Relatively small number of operations and experienced
engineers

Maximum HPGR throughput is approximately 2500 tph

(increasing in near future to ~3000+ tph)

Sensitive to feed moisture

Assessment of HPGR is expensive (no lab scale test)

Approach to Application
Feed Size: Top size related to roll diameter and gap.
Typically a maximum of 50 mm top size

Feed Moisture: less than 8%

Circuit Configuration: Typically tertiary application with

closing screen. Quartenary (Grasberg) and pebble
crusher duty (Penasquito and Empire Mine)

Material Handling: Choke fed feed hopper located

directly above HPGR. Product is typically wet screened

170
 Approach to Application
Tramp Metal: Needs to be removed to
protect roll lining

Wear Linings: Spare roller set needed to

reduce downtime during liner changes

METSO HRC (METSO Catalogue, 2013)

HPGR Operation
Machine Control:
Product Size: Controlled by changes in pressing force
(hydraulic setpoint) not roll gap!
Throughput: Controlled through changes in roll speed (VFD)
Roller Skew: control depends on vendor and can be
mechanical or hydraulic (adjusted via control loop).

Machine response to changes in roll speed or pressing force

setpoints is almost instantaneous

171
 HPGR Operation

Influences on Roll Wear:

Feed Moisture: Wear generally increases with moisture

Roll Speed: Wear increases with higher roll speeds

Pressing Force: Wear increases when greater pressing

forces are used

Feed Size: An HPGR feed top size that exceeds the width of
the operating roll gap is particularly detrimental to roll wear

344

172
 345

Typical Flowsheet:
Tertiary Application

Cerro Verde Flowsheet (Vanderbeek 2006)

173
HPGR Quartenary
Role (Grasberg)

Villanueva et al (SAG2011)

HPGR Pebble
Crusher Role

Peasquito
(Mexico)
Palmer et al
(SAG2011)

174
 HPGR & The Future
2 stage HPGR & Stirred Milling

Wang et al (CMP2013)

Novel Flowsheet for Ores with Clays

Rosario (2010)

175
COMPARISON OF HPGR - BALL
MILL AND HPGR - STIRRED MILL
CIRCUITS TO THE EXISTING
AG/SAG MILL - BALL MILL
CIRCUITS

Chengtie (Fisher) Wang

Presented at CMP Conference, Ottawa, 2013

352

176
 353

354

Outline
Introduction

Objectives

Experiment program

Results and discussion

Conclusions and recommendations

177
 355

Introduction

Comminution is energy
intensive and energy inefficient
process

Low-grade fine-grained deposit

increases energy consumption
and carbon emission -61

Energy efficient comminution

technologies include high
pressure grinding rolls and
stirred mills
*US Department of Energy, Industrial technologies program, June 2007

356

High pressure grinding rolls

(Napier-Munn et al., 1996)

178
 357

Horizontal stirred mill

(Arburo & Smith, 2009)

358

Objectives

HPGR-ball mill circuit

HPGR-stirred mill circuit

179
 359

Experimental program
Existing Operation

Circuit
Identification

Plant DCS data Plant Survey and

Equipment data Sampling

Bulk Sample

Characterization: Pilot HPGR Testing

JK DW parameter
Bond work index
Size distribution
Specific gravity
Density Pilot Stirred Mill
Testing

Circuit Modelling Identification of

and Simulation Key Parameters

Comparison Analysis
Comminution equipment energy
Complete circuit energy
Operating and capital costs

360

Test flowsheet

180
 361

JK SimMet simulation

Mill Dimension
Ball Charge
JK DW Test
BBWi
PSD
%S

(Napier-Munn et al, 1996)

362

Case A - SAB circuit

Copper-Molybdenum porphyry

889 tph

JK DW A x b = 65, Ta = 0.45

BBWi = 13.8

F80 = 108 mm

P80 = 0.19 mm

181
 363

Case C - SAB circuit

Copper-Molybdenum porphyry

1332 tph

JK DW A x b = 64.9, Ta = 0.31

BBWi = 13.6

F80 = 92 mm

P80 = 0.27 mm

364

Case D - AGBC circuit

Copper-Molybdenum porphyry

765 tph

JK DW A x b = 74.2, Ta = 0.58

BBWi = 13.8

F80 = 95 mm

P80 = 0.24 mm

182
 365

Case H - SABC circuit

Copper-Molybdenum porphyry

766 tph

JK DW A x b = 31.3, Ta = 0.59

BBWi = 18.0

F80 = 66 mm

P80 = 0.16 mm

366

Sample

183
 367

HPGR testing results

Scaled HPGR product

Specific FSP M-dot ESP net
Test No. (90% Center, 10% Edge)
2 3
[N/mm ] [ts/hm ] [kWh/t] P80 [mm] P50 [mm]
A1 3.0 257 1.37 6.30 1.91
A2 4.0 191 2.22 1.67 0.54
C1 3.0 266 1.23 6.54 1.58
C2 4.0 208 1.87 1.88 0.76
D1 3.0 244 1.55 4.70 1.17
D2 4.0 142 2.90** 1.71 0.55
H1 3.0 184 1.89 6.50 3.00
H2 3.0 222 1.25 3.83 1.75

368

Bond ball mill work indices

RoM HPGR product Difference

Circuit
[kWh/t] [kWh/t] [%]
A 13.8 12.1 -12.3
C 13.6 12.6 -7.4
D 13.8 12.8 -7.2
H 15.4 15.4 -14.4

184
 369

IsaMillTM testing results

Test Description Units ISA A1 ISA C1 ISA D1 ISA H1

Feed top size [m] 710 710 1000 710
F80 [m] 310 326 420 343
Target P80 [m] 100 100 100 75
Specific Energy [kWh/t] 3.8 4.4 5.0 4.8
Media Consumption [g/kWh] 6 7 5 3

370

Pure comminution energy

HPGR-BM @ 75 um, 10%

HPGR-IsaMill @ 75 um, 37%

HPGR-BM @ 160 um, 24%

Note: A power factor of 120% and 95% of net specific energy was used to determine the total motor power draw
of the HPGR and IsaMill for the process capacity, respectively.

185
 371

HPGR - ball mill circuit

372

HPGRs - stirred mill circuit

186
 373

Complete comminution energy

HPGR-IsaMill @ 75 um, 34%

HPGR-BM @ 160um, 21%

Note: A power factor of 120% and 95% of net specific energy was used to determine the total motor power draw
of the HPGR and IsaMill for the process capacity, respectively.

374

Comparison breakdown

SAG mill
HPGR

187
 375

Comparison breakdown (contd)

376

Capital cost

*determined from vendor quotes and installation costs

188
 377

Operating cost

378

NPV and IRR*

F80 P80 HPGR/ball mill to SABC HPGR/stirred mill to SABC

[mm] [um] NPV, M$ IRR, % NPV, M$ IRR, %

66 160 33 22 n/a

66 75 22 23 5 7

*@5%, 15 years

189
 379

Conclusions
The combination of HPGR and stirred mill in a single flowsheet,
without tumbling mills, has been demonstrated to be technically
feasible, with the implementation of two passes of HPGRs in the
flowsheet, and large-diameter ceramic media in IsaMill for
coarse stirred milling.

The work has demonstrated that the HPGR - ball mill circuit and
HPGR stirred mill as alternatives to existing SAB/AGBC/SABC
comminution circuits has significant potential in energy saving.

Economics of HPGR - ball mill option and HPGR - stirred mill

option are more favourable compared to existing SABC circuit
larger operation and long mine life
more expensive energy supply area

380

Recommendations

Evaluation of the influence of ore hardness variability

Further evaluation of size classification for HPGR product

Further evaluation of coarser stirred milling

190
 381

Acknowledgements

Questions?

191
 Sensors and Sorting

Towards Tomorrows Smart Mine:

Embedded Sensor Telemetry and Sensor-Based Sorting

Acknowledgements

Andrew Bamber, CEO

MineSense Technologies Ltd, Vancouver, Canada.

N. Emre Altun, Associate Professor

Mula University, Mining Engineering Department, Mula,
Turkey.

Malcolm Scoble, Professor

Norman B. Keevil Institute of Mining, UBC, Vancouver,
Canada.

192
 Mines of the Future
Low grade, complex geology, deep and remote
Clean - less waste, improved waste management
Healthy and Safe
Energy efficient
Invisible - underground mining and processing
Smart best use of information eg sensors

Sensing and Sorting Technologies

Hand sorting - pre-Roman times
Automated sorting
Uranium radiometric sorting Ontario 1958
Diamonds X-Ray fluorescence W. Australia 1985
Recent large scale examples (est. 300 sorters
installations)
Nickel, Kambalda W. Australia
Platinum, Amplats, Rustenburg UG2 Section
Sensors - Surface versus Bulk Properties
Challenges Better sensors, higher throughput
machines

193
 Sensor Technologies
Method Analysis Application

Photometric Surface Coal, sulphides, phosphates,

(reflection, oxides
brightness, grey level,
RGB, IR, UV, texture)
Radiometric Bulk Uranium, gold
Conductivity, Bulk Metal sulphides, native
magnetic metals, iron oxides
susceptibility
X-Ray Fluorescence Surface Diamonds, metal sulphides,
limestone, iron
X-Ray Transmission Bulk Coal, sulphides

Conductivity Sorting

Conductivity Testing at UBC

PC

Sort Signal

A/D Converter:
Signal generation
and analysis

CommoDas
ROM Secondary EM
Sensing Coil 1 Sensing Coil 2 Sensing Coil 3 Conductivity Sorter

Amplifier
Bridge/
Power Supply Balancing Coil 1 Balancing Coil 2 Balancing Coil 3

194
Courtesy C.
Bergman
Mintek, 2009

390

195
391

392

196
393

394

197
395

396

198
397

398

199
399

400

200
 401

Sorting Economics

MiningValueChain(afterPorter,1980)

201
 Sorting Economics

ValueChain(withsorting)

Sudbury Ni-Cu Operations

Energy Assessment

202
Sudbury Operations - Conductivity Sorting
Deposit Feed Grade (%) Conc. Mass (%) Conc. Grade (%) Recovery (%)
Ni Cu Mg Ni Cu Mg Ni Cu Mg
Craig 8112 1.16 0.47 5.54 72 1.50 0.57 5.16 93.49 87.40 67.46

Craig LGBX 2.10 0.35 2.57 83 2.43 0.37 2.39 95.85 86.70 77.07

Fraser Ni 0.81 0.36 4.21 80 0.94 0.40 3.73 92.73 89.43 70.67

Fraser Cu 0.83 11.42 1.81 41 1.65 20.92 0.68 81.12 74.89 15.42

TL Footwall 1.29 9.08 1.90 66 1.85 12.05 1.08 94.66 87.88 37.51

TL Zone 2 1.40 0.87 3.41 62 2.03 0.87 3.41 90.35 83.84 59.11

TL Zone 1 0.68 0.43 6.00 44 0.98 0.48 5.58 63.07 48.43 40.47

Montcalm East 1.66 0.56 4.61 75 2.06 0.63 4.17 93.60 85.48 68.22

Montcalm West 0.32 0.15 5.97 30 0.64 0.30 6.05 59.23 57.50 29.93

McCreedy East Mine - U/G Sorting

203
 McCreedy East Mine U/G Sorting

Sudbury Operations Sorting

Overall reduction in energy consumption 20%

10000.00

9000.00
Base

8000.00
Precon
7000.00

6000.00

Power (kW) 5000.00

4000.00

3000.00

2000.00

1000.00

0.00
Montcalm Thayer Fraser Fraser Craig Onaping Ni Rim S Ni Rim S
Lindsley Copper Nickel Depth F/W

Thayer Fraser
Operation Montcalm Lindsley Copper Fraser Nickel Craig Onaping Depth Ni Rim S
Hoisting $399,995 $1,319,625 $505,001 $684,364 $2,391,748 $1,891,163
Haul $786,583 $302,422 $884,600
Pre-con -$1,342,180 -$843,569 -$615,687 -$979,603 -$1,285,380 -$1,285,380 -$1,167,864
Grinding $560,607 $273,248 $236,058 $320,410 $476,930 $476,770 $418,730
Processing $1,397,813 $698,906 $436,817 $873,633 $1,310,450 $1,310,450 $1,135,723
Overall
Savings $1,402,823 $831,002 $1,376,812 $719,440 $1,186,364 $2,893,589 $3,162,352

204
 Sorting Past Present - Future?
Proven Technology
Sorting machines exist
Metallurgy proven
Concepts for mine designs developed
Economics demonstrated
Challenges of implementation
Better sensors
Higher capacity sorters
Technology transfer - Risk averse industry

How can we make better use of sensors?

Sensors organic part of mining system
Apply to all aspects from exploration (geophysical,
borehole) to mining to processing
Embedded sensors in material handling systems
(ore passes, scoops, shovels, bins, chutes,
conveyors)
Transmission, recording, analysis technologies
Wireless data transmission (WiFi)
Data available to GEMCOM, MineSight, process
control
Intelligent connected mines with active online telemetry

205
 Innovative Use of Sensors

Core logging equipment

Boreholes
Blast hole drill rigs
Face shovel
Belts
Sorter

Multi-Sensor Product Platform

ConductOreXTM Desktop Evaluation System
BeltSense - Multi-modal Mineral Telemetry System (completed
Pilot SortOre HFEMS or HSXRF @ 10 tph (on demand)
ShovelSense - Scoop/Shovel HFEMS System (in progress)
SortOre High Capacity Sorting System (in progress)

SortOre40TM ShovelSenseTM BeltSenseTM 412

206
Sensor Based Systems in Surface Mining

Sensor-based technologies and U/G Mining

207
 Conventional vs Sensor Based Mining
Conventional mining:

- people-orientated, plan-based, subjective, time

consuming
Future mining:
Application of on-line telemetry from in-mine
sensors: Production scheduling, grade control,
plant process control settings:
- flexible
- objective
- real-time
- simultaneous

Conclusions

The outcomes of sensor-based technologies and sorting are

significant
in economic and environmental measures

Challenges to the application of these technologies relates

primarilly to aspects of technology transfer and mining culture
rather than technical issues

208
 NBK Research
Centre

418

Introduction to Grinding

209
 419

Grinding & Classification Outline

1. Types of Mill Equipment and Circuits

2. Factors In Grinding Circuit Selection
3. Ore Properties and Grinding Testwork
4. Mill Power and Sizing Grinding Mills
5. Importance of Grinding Media
6. Ultrafine Grinding
7. Classification Principles and Equipment

420

Types of Mill Equipment and Circuits

Introduction
Grinding Fundamentals Recap
Why Grind? Breakage vs. Enrichment and Upgrading
Grinding Economics
Grinding Mechanism Characteristics

Types and Characteristics of Grinding Equipment
Overview of Ball Mill Feed Preparation Systems
Grinding Mill Equipment Types
Characteristics of Grinding Mills

210
 421

Introduction
Course covers grinding equipment typically used in the
ferrous and non-ferrous mineral industry.

Internationally the technology is fundamentally similar with

minor variations to local conditions.

Over past 100 + years ball mills remain the central

component and workhorse of most grinding circuits.

Circuit differences are mainly in feed preparation.

422

Introduction
Technology is specialized because of the need to grind
mostly siliceous, highly abrasive ores.

Some adaptation of cement industry equipment has

resulted in power savings with harder rock (High Pressure
Grinding Rolls).

Wet grinding is almost universally practiced.

The notable exception of dry-grinding, air-swept, double

compartment ball mills (also adapted from cement
industry) that grind refractory gold ores prior to roasting.

211
 423

Grinding Fundamentals
Why Grind? Breakage vs. Enrichment and Upgrading
In the size reduction stages of grinding we are creating the
necessary mineral liberation conditions, or surface area, for
subsequent separation and enrichment, upgrading and recovery.

424

Grinding Economics
Rule of Thumb. Mills consume about two-thirds of the
entire process plant power, or about 20-25 kWh/t (65% of
35-40 kWh/t).

Mills consume about 1-2 kg/t (C$1-2) of grinding media &

liner steel.

Assuming 10c/kWh/t overall grinding costs (power+media)

are about $3-$5/t, or about 40-50% of overall mill
consumable costs.

212
 425

Grinding Economics

Lost performance in separation due to miss-grinding

represents a major problem for many operations, eroding the
process economics.

An economic balance is required between the marginal cost

of grinding and revenue to maximize net revenue.

426

Grinding Mechanism Characteristics

a) Breakage Mechanism

Impact / Compression Attrition/ Chipping Abrasion

b) Dominant Grinding Action

Tumbling/Impact Cataracting Cascading

c) Energy Efficiency
Lowest (-) Improving Highest (+)

213
 427

Tumbling Mill Ball & Energy Distribution

428

Ball Mill Feed Preparation Systems

1. Crushing the ROM feed by primary crusher to a top size

of about 300 mm at crusher settings of 150 to 200 mm to
permit conveyor transport.
2. Further size reduction by either:
2 + stages of crushing by cone crushers to a ball mill
feed size of 10 to 15mm.
2 + stages of crushing & rod milling to a ball mill feed
size of 1.5 to 2mm.
Semi-autogenous (SAG) or autogenous (AG) grinding
to a ball mill feed size of 1 to 4 mm.
Cerro Verde Crusher/HPGR: 2.8 mm.

214
 429

Ball Mill Feed Preparation Systems

As concentrator capacities have increased, SAG mills have

become the standard method of preparing ball mill feed.

The capacity of secondary and tertiary crushers has not kept

pace with increasing plant capacity, as well as limitations in
rod mill capacity.

Currently the largest cone crusher commonly in service is the

MP1250 driven by a 1,250 hp motor.

430

Ball Mill Feed Preparation Systems

A 50k tpd+ secondary and tertiary crushing plant is

complex with many lines, screens, conveyors, bins, etc.

SAG mills were the only practical way to prepare ball feed
at medium and high tonnage rates.

SAG (or AG) mills have been installed in most mineral

processing grinding circuits in the last 20 years.

Now HPGRs have been shown to be viable alternatives.

215
 431

Types of Grinding Equipment

SAG/AG Vertimill Vibrating Ball

Ball Isamill
Rod Stirred Media
Pebble Detritor (SMD)

432

Common Mill Components

1 Shell 7 Frame
2 Mill Heads 8 Feed spout
3 Trunnion Bearings 9 Discharge Trommel
4 Grinding gear & pinion 10 Discharge Chute
5 Grinding Mill Reduction Unit 11 Mill Liners
6 Mill Motor

216
 433

Types of Grinding Equipment

Ball Mill (Grate Rod Mill

Discharge)

434

Insides of rod and ball mills

Ball mill

Rod mill

217
 435

Types of Grinding Equipment

Vertimill Vibrating Ball

436

Regrind Mills

218
 437

Grinding Circuits

438

Outline

Grinding Equipment Selection

Types of Grinding Mills
Evolution of Grinding Equipment
Grinding Circuit Arrangements
Mongolian ASM Circuit

219
 439

Source: Mt Polley

440

Reduction Ratio: Grinding vs. Crushing

Crushers have a limited reduction ratio - due to the design,
there is a limit to the residence time for the material passing
through.

Grinding in a mill takes place in more open space, thus the

retention time is much longer and can easily be adjusted
during operation.

In practice size reduction by grinding is done in optimized

stages.

220
 441

Grinding Equipment Selection

442

Grinding Equipment Selection

Mill Type Grinding Media Feed Size Product Size Diameter / Length

a) Autogenous Coarse Ore 2 feet -100 mesh 3 to 1

b) Semi- Coarse Ore / 2 feet -10 mesh 2 or 3 to 1

autogenous Balls
c) Rod Steel Rods 2 inch -10 mesh 0.5 to 1

d) Ball Steel Balls 1 inch to -200 mesh 0.5 up to 1 to 1

4 mesh
e) Pebble -8 + 4 Pebbles -1 inch -200 mesh 0.8 to 1

f) Verti-Mill Sand/Ceramic - 2mm 25-10 Vertical

microns

221
 443

Autogenous (AG) and

Semi-Autogenous Mills (SAG)

444

Autogenous (AG) Mill (D:L is about 3:1 US and

1:1 Europe/RSA)
Wet or dry grate discharge

Product: ~ -100 mesh (149 micron)

Primary, coarse grinding (up to 2 feed)

Grinding media is the feed material (min load of 15% > 6 in)

High capacity (short retention time)

Sensitive to feed size & material composition (critical size)

222
 445

Semi-Autogenous (SAG) mill (D:L is about 2-3:1

US and 1:1 Europe/RSA)
Wet or dry
Product: ~ 10 mesh (2 mm)
Higher capacity than AG mill
Primary, coarse grinding (up to 2 ft feed size)
Grinding media is feed plus 4-12% ball charge (4-5
inches)
High capacity (short retention time)
Less Sensitive to feed composition (critical size material)

446

Gibraltars New SAG (28)

223
 447

Fully Assembled 40 ft. x 22 ft. SAG Mill @ Cadia

Metso Minerals, Inc. 2003

448

HVC Line C: 43'x16, 2 x 4700 kW motors

224
 449

Rod Mills (D:L is about 0.5:1)

Overflow is wet only End & Center are mainly dry

Mostly open circuit (secondary) Special Applications
Grinding media is rods Capacity < 200 t/h limited by rod
Primary grinding secondary/tertiary length (22 ft max)
crushed Power < 1,500 kW
product (19-50 mm)
Coarse grind 600-2500 micron

450

Rod Mill Dimensions

Rod length to mill diameter 1.4 to 1.6

6.8 m is practical limit on rod length

Mill length should be 0.1 to 0.16 meters (4 to 6) longer than

the rods.

225
 451

Rod mill @ Mount Polley

452

Overflow Ball Mills (D:L is about 0.5-1: 1)

- Wet only, Robust & Simple

- Primary on 1 in. to 4 mesh crushed feed
- Mostly closed circuit (secondary) on
AG/SAG/Rod/HPGR product
- Finer Grind (longer retention time) to > 20 microns
- Higher risk of over grinding
- Ball charge 35-45%.

226
 453

Grate Discharge Ball Mills (D:L is 0.5-1:1)

- Wet or dry
- Discharge grate more complex
- Primary on 10-19 mm crushed feed
- Mostly closed circuit (secondary) on rod product
- Coarser grind (short retention time) > 74 microns
- Lower risk of over grinding
- Can take 5-10% more balls

454

22 ft x 36.5 ft Ball Mills @ Cadia

227
 455

24 ft x 36 ft Ball Mills @ Cerro Verde, 13 MW

456

Pebble Mill (D:L is about 0.8:1)

- Wet or dry grate discharge (product 200 mesh)

- Secondary grinding (AG/SAG/Rod/Ball Product) of 1 inch feed
- Grinding media pebbles (-8 + 4 inch) screened from feed, flint
pebbles, porcelain balls
- Larger than ball mills at same power draw

228
 457

Grinding @ Mount Polley

458

Vertimills
Vertical stirred
From 10 HP through 1500 HP.
For wet application 2 mm feed
to as fine as 10 microns.
Secondary/Regrind/Lime
Slaking

229
 459

VTM-1250-WB
Vertimills at Chino

460

Main Grinding Mill Suppliers

(Sales 1990 2002)
Metso Metso
Minerals
Metso
FFE
FFE Fuller-Vecor
Bradkin Outo- Outokumpu
Krupp
(purchased
Outo.
Nordberg/Morgardsh
Others
ammar)

230
 461

AG/SAG Mill Evolution

1959 - 1st 18 ft/5.49m. diameter AG Mill @ 600 HP/448 kW
1959 - 1st 22 ft./6.71m diameter AG Mill at 1,250 HP/933 kW
1962 - 1st 24 ft./7.32m diameter AG Mill @ 1,750 HP/1,306 kW
1962 - 1st 28 ft./8.54m diameter AG Mill @ 3,500 HP/2,612 kW
1965 - 1st 20 ft./6.10m diameter AG Mill @ 500 HP/373 kW
1965 - 1st 32 ft./9.76m diameter AG Mill @ 6,000 HP/4,478 kW
1970 - 1st 26 ft./7.93m diameter AG Mill @ 3,000 HP/2,239 kW
1970 - 1st 30 ft./9.15m diameter AG Mill @ 7,000 HP/5,224 kW
1973 - 1st 36 ft./11.0m diameter AG Mill @ 12,000 HP/8,955kW
1979 - 1st 34 ft./10.4m diameter AG Mill @ 8,800 HP/6,567 kW
1986 - 1st Gearless SAG Mill @ 11,000 HP/8,209 kW
1996 - 1st 38 ft./11.6m diameter SAG Mill @ 26,800 HP/20,000 kW
1996 - 1st 40 ft./12.2m diameter SAG Mill @ 26,800 HP/20,000 kW
Proposed 42 ft./12.8m diameter SAG Mill @ 37,500 HP/28,000 kW

462

Ball Mill Evolution

1965 - 1st 14 (4.27m) dia. Ball Mills @ 1,306 kW
1966 - 1st 15.5 (4.73m) dia. Ball Mills @ 1,493 kW
1967 - 1st 16.5 (5.03m) dia. Ball Mills @ 2,612 kW
1970 - 1st 18 (5.49m) dia. Ball Mills @ 3,172 kW

Following poor performance of 18 ft mills at Bougainville,

there was speculation that the limit of ball mill size had
been reached. It was subsequently proved that operating
conditions were the cause of observed lower grinding
efficiency and not size.

231
 463

Ball Mill Evolution

1980 - 1st 21 (6.4m) dia. Ball Mill @ 8,060 kW

1990 - 1st 20 (6.1m) dia. Ball Mill @ 5,597 kW
1996 - 1st 22 (6.71m) dia. Ball Mills @ 8.955 kW
1996 - 1st 24 (7.32m) dia. Ball Mills @ 10,448 kW
1999 - 1st 25 (7.62m) dia. Ball Mills @ 13,433 kW
2001 - 1st 26 (7.93m) dia. Ball Mills @ 15,500 kW
Current 22 MW+

464

Ball Mill Evolution

Ball mill sizes have continued to increase and there is
currently no evidence to suggest that efficiency drops as
diameter increases.

Economics is driving selection of the fewest number of

mills lines.

A large SAG mill followed by a large ball mill could enable

a single mill line to mill up to 150,000 tpd of ore.

232
 465

Ball Mill Evolution

34 ft. SAG & 20 ft. Ball Mills @ Fairbanks Gold

466

Grinding Circuit Arrangements

Secondary and Tertiary Crushing plus Single-
stage Ball Mill Grinding
This circuit and the following rod/ball mill circuit were
almost universal pre-1975.

Variations such as pebble or tube mills and deleting

tertiary crushing for softer ores or low tonnage
operations.

The 3-stage and single-stage ball remains one of the

most energy efficient compared to AS/SAG but
crusher/rod mill sizes did not keep pace to industry
leading to its demise.

233
 467

Secondary and Tertiary Crushing

plus Single-stage Ball Mill
PRIMARY
CRUSHER FEED PRODUCT

CYCLONE
SCREEN SCREEN

BALL
MILL

SECONDARY TERTIARY
CRUSHING CRUSHING

468

Primary, Secondary and Tertiary Crushing plus

Rod Mill and One or Two-Stage Ball Mills
PRIMARY
CRUSHER FEED PRODUCT

CYCLONE
SCREEN

ROD BALL
MILL MILL

SECONDARY TERTIARY
CRUSHING CRUSHING

234
 469

SAG/AG Mill (Pebble Crushing) + Ball Mill

These circuits have been the workhorse of the industry for

the last 20 years.

External pebble crushing improves power efficiency and is

necessary for competent ores that exhibit a propensity to
form critical size material.

Early AG installations in the iron industry have operated

well for many years.

470

SAG/AG Mill (Pebble Crushing)

+ Ball Mill
Similar installations in the copper industry were not so
successful, typically grinding too fine at low tonnage rates.
Some circuits were modified to SAG operation.

There are a few single stage SAG mills operating

successfully. This type of circuit is well suited to uranium
sandstone deposits (Colorado Plateau Ores) in which
uranium coatings are released for leaching.

235
 471

SAG/AG Mill + Ball Mill

PRIMARY
CRUSHER FEED PRODUCT

CYCLONE

SCREEN
BALL
MILL

472

SAG/AG Mill + Ball Mill +

Pebble Crushing
PRIMARY MAGNETIC
CRUSHER FEED SEPARATOR PRODUCT

PEBBLE
CRUSHER

CYCLONE

SCREEN
BALL
MILL

236
 473

Double Rotator Dry Grinding

This circuit has been adopted by two gold roasting operations
in Nevada. The circuit was adapted from cement industry
practice and combines drying with two stages of grinding.

The handling and classification circuit is relatively complex

airslides, bucket elevators, dynamic and static classifiers and
product recovery baghouses.

474

Double Rotator Dry Grinding

PRIMARY
CRUSHER FEED PRODUCT

CYCLONE

HOT GAS

DRYING COARSE FINE

GRIND GRIND

237
 475

Mongolian ASM Circuit

476

238
477

478

239
479

480

240
481

482

241
 483

Grinding & Classification

Ore Characterization

484

Introduction Ore Testing

Grinding Ore Testing - To quantify what type and size of

grinding circuit is best suited to the ore.
Test work can range from simple tests, based on a small
sample of rock or core, to comprehensive pilot testing
requiring hundreds of tonnes.

Objective to become familiar with commonly used ore tests

for grinding.

242
 485

Common Ore Tests

Bond grinding indices (rod, ball and abrasion)
Unconfined compressive strength (UCS)
Impact crushing tests
Autogenous Media Competency (Tumble Test)
JK drop weight tests
McPherson
SPI Minnovex (Starsky)
Pilot scale milling
Circuit Surveys

486

Standard Bond Ore Testing

Four Most Relevant Indices:

a) Bond Ball Mill Work Index (BMWI)
b) Bond Rod Mill Work Index (RMWI)
c) Bond Abrasion Work Index
d) Standard Bond Crushing Work Index (see Impact Crushing
Tests)

243
 487

a) Bond Ball Mill Work Index

BMWI standard test was developed by Fred Bond in the
1920s, published in 1952 and modified in 1961.
Test enables basic grinding power requirements to be
determined, from the feed 80% passing size (F80) to the
circuit 80% passing size (P80).
BMWI test determines the standard Wi of a sample of ore, or
the specific power (kWh/t) required to reduce the P80 of a
sample of material from infinite size to 100m.
BMWI is used in designing new equipment and in simulating
existing equipment to improve performance.

488

What is a Ball Mill Wi Test?

The BMWI is a measure of the resistance of the material to
crushing and grinding.
It is a 'locked cycle' test conducted in closed circuit with a
laboratory screen.
Requires 10 kg of drill core or rock, crushed to 3.35mm (6#
Tyler)
The closing screen size is selected so that the product P80
from the test is as close as possible to the product P80
expected from the circuit under design.
Note: Wi is linked to the tested closing sieve size.

For full details, refer to the original Bond paper (Ref: Bond, F.C. 1961.
Crushing and Grinding Calculations Part I and II, British Chemical
Engineering, Vol 6., Nos 6 and 8).

244
 489

When would a BMWI Test be required?

A BMWI is required for the design of a new mineral processing
plant. Tests should be on a samples of ore that are typical of
the proposed feed to the plant.
A BMWI may also be used in the simulation and subsequent
optimization of existing mill(s) and the associated grinding
circuit.
The Bond Equation can be used to calculate:
The specific energy requirement for a given grinding duty,
and
The feed size and required product size.
It is then possible to determine the size of mill required based
on throughput, and therefore motor power.

490

Detailed Ball Mill Work Index

Test Procedure
1.Stage crush the feed to 3.35mm (- 6 mesh) and take a
representative sample.

2.Undertake a series of batch grinds in a standard Bond mill. A

Bond mill is 0.305m x 0.305m (12), with rounded corners,
smooth lining, running at 70rpm. The charge consists of 285
balls, weighing a total of 20.125kg.

3.Initially, a 700ml sub-sample of feed is prepared for use in the

first batch grind. It is ground in the mill for 100 revolutions. All
grinding is dry.

4.After each batch grind, the contents of the mill are sieved on
the selected 'closing' screen to remove the undersize. This is
replaced by an equal weight of fresh feed to bring the weight
back to that of the original charge.

245
 491

Detailed Ball Mill Work Index

Test Procedure
5. This sample is then ground in the mill for a predetermined
number of revolutions calculated to produce a 250% circulating
load. The number of revolutions required is calculated from the
results of the previous period to produce sieve undersize equal
to 1/3.5 of the total mill charge.

6. Repeat at least 7 times until the weight of undersize produced

per mill revolution reaches equilibrium.

7. The average of net mass per revolution from the last three
cycles is taken as the ball mill grindability (Gbp) in g/revolution.

8. A representative sample of product is sized to determine the

P80.

492

Detailed Ball Mill Work Index

Test Procedure
9. Calculate the BMWI using the Bond equation:

Wi = 44.5 / [(P1)0.23 x Gbp0.82 x 10 (1/P80 - 1/F80)]

Where:
Wi = Ball mill work index
P1 = opening in microns of the sieve size tested
Gbp = the average of the last three net grams per
revolution, or grindability.

246
 493

How are BMWI results reported and

what do they mean?
The standard report details the Bond test procedure method,
and presents the results including F80, P80, Grindability and
Work Index.

The Bond BMWI provides a measure of how much energy is

required to grind a sample of ore in a ball mill.

Typical BMWI results and their relative measure include:

Property Soft Medium Hard Very Hard

Bond WI (kWh/t) 79 9 14 14 20 > 20

494

Additional Bond BMWI Comments

A typical BMWI test takes 1 week.

As a rule of thumb, for a given closing sieve size, the resulting product
P80 will be ~ one root 2 series sieve size smaller. For example, if the
required product P80 is ~ 106 m then use a 150 m closing sieve size.

Wet sieving is only used if the material is likely to agglomerate or if the

closing sieve size is 45m.

Wet sieving significantly increases the test time, as the test must be
carried out on dry material. The sample must be oven-dried after each
wet sieving process.

There may also be issues of material degradation either in water or at

the high drying temperatures, which needs to be considered before the
test is carried out.

247
 495

b) Bond Rod Mill Work Index

BRWI test requires 20 kg of material, which is crushed to -

12.7mm (-1/2) and is tested in a standard Bond Rod mill.

The sample is ground to -1.18 mm (14# Tyler) to emulate

the duty of a primary rod mill in front of a secondary ball
mill.

The rod mill index derived from this test is used in

conjunction with the ball mill work index to determine the
rod mill power demand, again using the Bond power
equation.

496

Rod Mill Grindability Test Procedure

1. Weigh 1250 cc of crushed 1/2 inch rock
2. Conduct sieve analysis and determine, F80
3. Grind dry in closed circuit with 100% circulating load in 12
inch diameter x 24 inch long rod mill
4. Screen and weigh undersize of product
5. Add fresh feed to original 1250 cc weight
6. Calculate number of revolutions to produce 100% circulating
load
7. Repeat cycle until the net grams of undersize produced per
revolution is constant
8. Conduct Sieve Analysis on product and determine P80
9. Calculate Wi:
Wi = 62 / [(P1)0.23 x Gbp0.625 x 10 (1/P80 - 1/F80)]

248
 497

c) Bond Abrasion Index

The abrasion index test requires only 5 kg of material, which
is crushed and screened to an exacting size range of +12.7
19.0 mm (+ ).
The test uses a small laboratory scale mill with a test paddle
that is weighed before and after being rotated in contact with
the dry test sample.
The difference in weight is designated as the abrasion index,
and is used in conjunction with Bond formulae to predict liner
wear and media consumption in rod and ball mills, as well as
in crusher liners.

498

d) Impact Crushing Tests:

Standard Bond Method
These tests can take two forms. The first is the Standard
Bond Crushing Test, which has a requirement of twenty
pieces of rock or core of size +50 75 mm (+2 3). Pieces
are placed in a twin pendulum device and impacted to failure
to produce an impact crushing strength, measured in kWh
per tonne of ore.
Twenty specimens are tested to provide a measure of
variability of results, as there is a tendency towards
heterogeneity in rocks of larger sizes. The standard index is
used primarily by crusher manufacturers to assign down
rating factors for ore toughness in crusher selection.

249
 499

Impact Crushing Tests: Standard Bond

Method

500

Impact Crushing Tests:

Standard Bond Crushability Method
Crushability Test Procedure

1. 3 + 2 inch rock mounted between

two-30 lb weights on wheels
2. Weights strike rock simultaneously on
smallest dimension
3. Increase height until rock breaks
4. Calculate impact crushing strength, C
(ft. lb/inch)
5. Determine rock SG
6. Calculate Wi from average of 10
breaks

Wi = 2.59 x C / SG

250
 501

Impact Crushing Tests: Modified Bond

Method
Test uses a larger sample of rock or drill core.

120 kg of material is tumbled in the standard 1.83 m diameter

x 0.3 m wide Bond autogenous media competency test mill
for 500 revolutions at 26 rpm, to eliminate imperfections in
rocks and to mimic seasoned pebbles in a mill charge.
The product is screened to remove 19 mm material. The
oversize are sorted into 4 or 5 classes, depending on the feed
size. The size classes are 19 x 25 mm, 25 x 38 mm, 38 x 51
mm, 50 x 75 mm, and 75 x 100 mm.
Select 20 rocks in each size class and subject to standard
Bond Impact Crushing Work Index Test.

502

Single Particle Methods (AG and SAG)

a) Bond Impact Test Method: Barratt Approach

Barratt (1986) proposed a method for predicting SAG power involving the
use of a combination of Bond Work Indices over a range of sizes from F80
to a defined P80, applying a correction factor to resultant power, and
deducting the ball milling component of the power:

E (SAG) = [10Wic(Sp) + 10Wir(Sr)*Kr + 10Wib(Sb)*Kb] * 1.25 - 10Wib(Ssb)

where: E (SAG) is the specific SAG mill power in kWh/t

Wic,r,b are the Crushing, Rod and Ball mill Work Indices
Sc,r,b are [1/P - 1/F] for the equivalent stage size ranges

It was noticed that the method can be used unless the Wic and Wir are
significantly higher than the Wib, in which case SABC is indicated and E
(SAG) can be discounted by 10% to arrive at a power efficient SABC design.

251
 503

Single Particle Methods (AG and SAG)

b) Bond Impact Test Method: Siddall Approach
Siddall, et al., (1996) classified the responses obtained from impact
testing the products of a tumbling drum and related them to a correction
factor, designated f(SAG) which is applied to the Bond Ball Mill Work
Index to predict the total power required to grind from F80 = 150 mm to P80
= 75 micron. The equation takes the form:

P(TOT) = 10 WI * f(SAG) [1/75 - 1/150000]

By subtracting the ball mill power requirement and correcting for feed
size, the SAG mill power can be predicted.

P(SAG) = P(TOT) P(cr) P(bm)

P(TOT ) is the total circuit power

P(cr) is the correction for feed F80 size
P(bm) is the correction for ball mill power

504

SINGLE PARTICLE METHODS

- AG AND SAG

The Barratt Approach and Siddall Approach methods have

been found to predict the single stage grinding power
required in a AG/SAG mill.

In both methods, there is a reliance on either pilot plant data

or database correlations in order to establish T80 (SAG
transfer size), and hence the SAG mill power in a two-stage
grinding circuit.

252
 505

Impact Crushing Tests: Results

The test provides the raw data required to derive an impact
crushing profile, used to identify the type of comminution
circuit is best suited to the ore.

506

Additional Ore Characterization Tests

Unconfined compressive strength (UCS)
Autogenous Media Competence Test
JKDrop Weight Test
McPherson Test
SPI Minovex
Pilot Scale Testing
Plant Circuit Surveys

253
 507

Ore characterization test requirements

-SAG/AG mills
Must test particles over the entire size range of SAG mill feed
for both impact and abrasion breakage, to determine energy
levels expected in commercial mills
Must determine media competency;
Must allow examination of steady-state mill load characteristics
(critical sized material);
Must generate a breakage vs. energy level map for simulation
Must be reproducible (need representative samples);
Must determine total grinding power required; and,
Must use a small sample mass.

508

Unconfined Compressive Strength Test

This test determines the strength of a rock sample under
compression by a single vertical force.

The test requires the use of a specialized compression

device which applies an evenly controlled force to the
rock until failure.

Unfortunately, the test is undertaken in many different

types of devices, with widely varying sample
specifications, which makes cross-comparison of results
difficult at times.

254
 509

Unconfined Compressive Strength Test

One international standard that is used widely is the

ASTM2938-86.
A sample specimen is machined into a cylinder featuring a
length twice that of the specimens diameter, ideally 50 mm
(2).
The test produces two outputs:
The mode of breakage, providing insight into the nature of the
rock.
The actual UCS value, usually quoted in MPa.
The UCS value is used to guide crusher manufacturers in
selecting the appropriate equipment, and to assist the grinding
consultant in assessing an ores competency.

510

Autogenous Media Competency Test

When a sample is available in the form of lump rock, the
standard Autogenous Tumbling Test can be carried out using 10
rocks in five size ranges between 102 and 165 mm.

The rock is normally tumbled for 500 revolutions in a

6 ft x 1 ft drum and the product sized.

The product provides data to enable evaluation of the

following:
Interpretation of the product distribution against generic
curves.
Production of media in AG and SAG mills.
The amount of critical size build-up.
The tendency for ore to generate fines (-6 mm material).
Overall amenability to autogenous milling.

255
 511

Autogenous Media Competency Test

The test provides excellent insight into impact breakage and
auto abrasion characteristics of ores, but is currently only
performed in a few laboratories around the world.

512

JK Drop Weight Test

This test has been devised by the

Julius Kruttschnitt Mineral Research
Centre (JKMRC), and is used to derive
impact breakage and abrasion
parameters for use in their simulation
package, JKSimMet.

The method involves dropping a metal
weight from a set height onto a test
specimen and sizing the daughter
products from the resultant rock failure.

256
 513

JK Drop Weight Test

JK Tech introduced a Drop Weight Test to replace the

Pendulum Test.
A number of specimens of varying sizes are tested to
generate breakage curves from which the JKSimMet
simulation parameters are calculated.
The test is a useful adjunct to the other media
competency tests: once the type of circuit that is best
suited to the ore is identified, JKSimMet can be used to
verifying initial mill sizes.

514

JK Rotary Breakage Tester

257
 515

JK Rotary Breakage Tester

Minerals Engineering 22 (2009) 602612

516

McPherson Test
The most notable method in use in the Americas is the
test developed by Art MacPherson.
It uses dry grinding on material typically crushed to 38
mm to ascertain the ores autogenous characteristics in a
450 mm diameter mill.
The results are compared to a standard Bond test, and an
empirically scaled value for the amount of power that is
theoretically required to grind the ore is determined.
The test is normally used as a precursor to pilot scale
AG/SAG milling.

258
 517

McPherson Test
The main area of concern with the test is the underlying
assumption that ore at large lump sizes behaves in a
similar manner to the small sizes used in the test at 38
mm, which is not consistently correct.
This is particularly so with tough siliceous ores (typical of
the greenstone belts in Australia and parts of Africa).
However, the test can serve as a useful adjunct to the
other testes discussed above in providing some insight
into the autogenous characteristics when whole ore is not
available for testing.
It generates an estimate of the product size from an
AG/SAG mill.

518

SPI Minnovex Test (Starkey )

This test has been devised by Minnovex in Canada to predict SAG mill
specific power requirements using only 12.7 mm material.
The test uses a small 300 mm dia. x 100 mm long laboratory scale mill
with a small ball charge of 25 mm balls to grind a 2 kg test sample.
The objective is to establish the grinding time required to grind the ore to
80% passing 1.7 mm (10#), the closing screen size.
This test claims to demonstrate a strong correlation between grinding time
for ores and their corresponding SAG mill specific power draw. It provides
an attractive alternative to tests requiring large sample size.
Like the McPherson test it draws on a large database for comparison with
actual operations, which also provides the basis for calibration of the
model against laboratory results.

259
 519

Pilot Scale Testing

Most greenfields projects can not access whole ore in the early stages
of the study; hence, the focus on drill core testing.
If whole ore is available from a current operation or from a development
audit or shaft it is possible to undertake pilot scale SAG mill testing.
For circuits of less than 2 MTPA capacity, piloting is usually not justifiable,
with the cost of such a venture usually incorporated into extra mill length
and/or motor rating.
For simple AG or SAG mill piloting without online downstream piloting of
other unit processes (such as flotation or solvent extraction) 100 150
tonnes of ore are required, with campaign duration being 10 to 15 working
days in a test facility.
For more complex arrangements, campaigns have been known to run
over two months, with corresponding escalating costs.

520

Plant Circuit Surveys

When the mill selection being considered is the result of an
intended plant upgrade, obtain plant survey data in the form
of mass balances and sizing data.
Supplement with process information such as milling rate,
power draw and equipment configurations (operating ball
charge, total mill charge volume, milling speed, cyclone
parameters, etc.).
This data can be used to provide input information for power
based modeling, or for more sophisticated breakage rate
based on stimulators.

260
 521

Grinding & Classification

Circuit Design Mill Power

522

Sessions Outline

1. Types of Mill Equipment and Circuits

2. Factors In Grinding Circuit Selection
3. Ore Testing
4. Mill Power
5. Sizing Grinding Mills
6. Grinding Mill Design and Operation

261
 523

Milling Power

1. Introduction
2. Power Method Considerations
3. Bond Grindability Method
4. Mill Speed
5. Liner Profile and Speed Effects
6. Classification and Circulating Loads
Reference Papers
Bench-Scale and Pilot Plant Tests for Comminution Circuit Design, Mosher &
Bigg & Selection of Rod Mills, Ball Mills and Regrind Mills Rowland. SME, 2002.

524

Introduction

In 1951 Mr. Fred C. Bond of the AllisChalmers Co.

proposed his third theory of comminution.

Mr. Bond developed his work index (Wi), which is used

extensively to determine power input.

This session describes methods of determining mill power

(used as the basis of mill sizing) and ore testing.

262
 525

Introduction
Ore characterization for comminution is to provide
parameters to design circuits that economically achieve
the throughput and grind that is suited to the balance of
the plants equipment and capabilities.

Circuit design is a balance between:

Finding the minimum operating cost to attain the
desired final grind (consumables and power) &
Efficiently using installed capital.
The most efficient circuit is the one that allows the
greatest rate of return to a project.

526

Introduction

Certain circuit configurations require more ore

characterization that others.

A conventional crusher-rod mill-ball mill circuit requires less

characterization than an AG or SAG circuit.

AG/SAG circuits power draw is dynamic and greatly affected

by changes in operating conditions.

263
 527

Mill Power
Around half the energy used in most mineral processing plant
is consumed in grinding. Usually, it is the single biggest
operating cost item, and good energy utilization is critical to
project economics.
Sizing of grinding mills is mostly carried out by determining
the energy required for the duty and selecting an appropriate
unit to deliver that energy.
Determining the energy required can often be done by
laboratory testing. Two forms of testing are common:
The Bond grindability tests
Single particle tests e.g. the Impact test or the Drop Weight
Test.

528

Power Method Considerations: Testing

Bond grindability (ball and rod mills)
The Bond Grinding Indices are for predicting rod and ball
mill power requirements. They can also be used by
operators to assess the power efficiency of an existing
circuit, as explained below. However, the Bond BWI is not a
good predictor of AG/SAG mills unless adapted using
empirical factors.
Single particle (AG and SAG mills)
In order to assess AG/SAG behavior, single particle tests
have been devised which look at the energy required to
break the particle under impact conditions, and the
relationship between the energy applied and the size
distribution of the daughter products.

264
 529

Bond Grindability Method (Ball and Rod)

Work Input Determination
Ball mill circulating load 2.5: Rod mills CL 1.0.

Bond derived a formula for the calculation of the required

energy to reduce particles from a feed 80% passing size
(F80) to a product 80% (P80).
1 1
W 10 Wi
P80 F80

Where: W = work input in kWh/t

Wi = Bond Work Index in kWh/t
Multiplying the new feed (t/h) by W gives the power
requirement (kW).

530

Bond Efficiency Factors

Efficiency Factors are applied to W to derive the corrected power
requirement, based on empirical experience:

W Corrected = WBond* EF1 *EF2 *EF3 *EF4 *EF5 *EF6 *EF7 *EF8 *EF9

EF1 Dry Grinding

EF2 Open Circuit Grinding
EF3 Diameter Efficiency Factor
EF4 Oversized Feed Factor
EF5 Fineness of Grind
EF6 High/Low Ratio of Reduction Rod Milling
EF7 Low Ratio of Reduction Ball Milling Factor
EF8 Rod Mill Feed
EF9 Rubber Liners Factor

265
 531

Bond Efficiency Factors

EF1 (Dry Grinding)

With most materials, for the same range of work, dry

grinding requires 1.3 times as much power as wet
grinding. In some special cases, this correction factor
can be as low as 1.1 or great as 2.0.

532

Bond Efficiency Factors

EF2 (Open Circuit
Product Size EF2
Grinding) Control Reference
% Passing
For ball milling, EF2 is 50 1.035
a function of the 60 1.05
degree of control 70 1.10
required on the circuit 80 1.20
product. Open circuit 90 1.40
inefficiency factors are
92 1.46
as follows:
95 1.57
98 1.70

266
 533

Bond Efficiency Factors

EF3 (Diameter Efficiency Factor)

Using a base diameter of 2.44 m (8) inside liners the

correction for other diameters (in meters) is given by:

0. 2
2.44
EF 3
D

The minimum value applied for EF3 is 0.914 for practical

design purposes.

534

Bond Efficiency Factors

EF4 (Oversized Feed Factor) F Fo
Rr (Wi 7)
Fo
EF 4
Rr
Rr = Reduction Ratio, F80/P80
Wi = Rod Mill or Ball Mill Work Index in kWh/st
Fo = Optimum feed size = Zf * (13/RWi)0.5
Zf = A constant, where: rod milling = 16,000
ball milling = 4,000
RWi = Rod Mill Work Index in kWh/st
The influence of Rr should be assessed with caution in the
first stage of a two-stage circuit.
Do not use EF4 for rod mill prepared feed to a ball mill and
do not apply if EF4 < 1.0
In two-stage ball milling, use EF4 = 1.2

267
 535

Bond Efficiency Factors

EF5 (Fineness of Grind)

EF 5
P80 10.3
1.145 P80
Apply this factor only when P80 < 75m (200 mesh).

536

Bond Efficiency Factors

EF6 (High/Low Ratio of Reduction - Rod Milling)

Do not use EF6 if (Rr-Ro) is between -2 and +2:

EF 6 1
Rr Ro
2

150
Where:
5L
Ro 8
D
D = inside liner diameter of rod mill (meters)
L = length of rods (meters) = Rod Mill Inside L 0.15

268
 537

Bond Efficiency Factors

EF7 (Low Ratio of Reduction Ball Milling Factor)
If the Rr ,or reduction ratio, of the ball feed to product
drops below 6, use the EF7 correction factor. The lower
the Rr the more power required.

EF 7
2 ( Rr 1.35) 0.26
2 ( Rr 1.35)
Note: Do not apply an EF7 factor greater than 2.0 without
conducting continuous test work.

538

Bond Efficiency Factors

EF8 (Rod Mill Feed)
When calculating rod mill power for rod milling only, an
EF8 value of 1.4 is used when the feed is prepared by
open circuit crushing and 1.2 in closed circuit.

For Rod/Ball circuits 1.2 is used for the rod milling stage
only, if the feed is prepared in open circuit.

Do not use with Rod/Ball circuits with closed crushing

circuits.

269
 539

Bond Efficiency Factors

EF9 (Rubber Liners Factor)
With respect to wear resistance, rubber liners are best
suited for ball diameters up to 80 mm.
Steel liners are best suited to primary ball milling
applications requiring larger than 80 mm balls and rod
mills, and ball mills larger than 16.5 ft in diameter.
EF9 is applied to mills with rubber lifters, as they tend to
be somewhat bulkier than the equivalent steel
configuration, reducing the available grinding space.
Rubber liners also absorb a portion of the impact energy
of the steel media, reducing efficiency.
An EF9 of about 1.07 is typically assigned for rubber lined
mills.

540

Fines Correction
The product from the first stage of grinding (AG, SAG, or rod
mill) typically has a higher fines content than a crushing circuit
product.
To predict the ball mill size required in a secondary milling
application, the mill feed size is modified by removing finished
product from it.
The next slide shows the size distributions for a crushed vs.
ground feed, with different fines but the same P80.
A partition curve is typically applied to the SAG product at the
final product separation size.
The result is that only a fraction of the SAG product requires
secondary grinding, and this daughter product exhibits a
coarser size distribution than its parent.

270
 541

Fines Correction
The product from the first stage of grinding i.e. an AG mill,
SAG mill, or rod mill, usually has a different size distribution
than that produced by crushing to prepare ball mill feed.

542

Bond Method Limitations

The method is designed to predict power in a wet grinding
circuit at a 250% circulating load. Moving away from this
condition reduces the accuracy of the test.
It does not predict the behavior of large rocks in grinding
circuit where the mode of breakage is impact dominated
versus attrition and abrasion in ball mills (SAG/AG Mills).
The Bond Work Index is based on the energy per unit
mass required to reduce a particle from infinite size to
80% passing 100 m. If the P80 is less than 100 m,
serious discrepancies can occur. The closing screen in
the Bond test must reflect the size to which the particle is
to be ground.
If P80 10 m, do not apply Bond predicators of power.

271
 543

Bond Method Limitations

The shape of the size distribution generated by a two-
stage grinding operation may differ significantly from the
shape obtained by crushing.
The F80 may be the same, but the amount of fines at say
F30 or F20 may be markedly different.
A size distribution correction may be necessary to better
predict 2nd stage power requirements for:
SAG/Ball milling and 2-stage ball milling
Other unnatural or scalped feed distributions.
These conditions require additional grinding energy based
upon the variation from a more standard feed distribution.

544

Mill Speed - Critical Speed

Grinding mill is usually shown as a percentage of critical
speed, Nc.
42.31 76.63
Nc Nc
D D
(D in meters) (D in feet)
Normal mill speeds range from 60 to 90% of Nc, dictated
by operational and economic considerations.
Power drawn is proportional to mill speed, suggesting that
mills should be run as fast as possible.
However, the useful work done by the grinding charge is
related to the mode of breakage induced, which is in turn
influenced by the liner design and charge level.
Higher speeds lead to higher rod, ball and liner wear.

272
 545

Mill Speed
Effect of Mill Speed on Load Trajectory
Figure 1 illustrates the effect on the trajectory of the outer
envelope of the charge at increasing speeds for the same ball
size with two lifter designs.

546

Mill Speed
Speed Guidelines
Studies such as on the previous slide have produced the
following general guidelines:

AG Mills - An impact mode of breakage is usually sought,

and with no steel media in the mill it is possible to run at
speeds in the range 80-90% Nc.

SAG Mills - Typical operating speeds are around 75% Nc.

Liner damage will occur if the balls are allowed to impact
them directly, and SAG mills usually have variable speed
drives.

273
 547

Rod Mill Speed

Rod Mills operate at a lower speed than ball mills to
ensure that there is no cataracting of the rods. Typical
speeds related to the inside shell diameter are:

Diameter % Nc
(m) Inside Shell
2 68.0

3 65.0

4 64.0

4.57* 62.6

* max. recommended diameter

548

Ball Mill Speed

Smaller mills can be run at high speeds up to 85% Nc,

medium diameter mills at lower speed 70-72% Nc. There is
an emerging trend of operating very large mills (>5 m dia.) at
higher speeds typically 76% Nc in an attempt to overcome
an inactive kidney problem.

Typically for Ball Mills D < 5m:

% Nc = 83.5 [D] 0.108

274
 549

Liner Profile and Speed Effects

Figure 1 also shows the effect of differing lifters

on the trajectory of balls in the ball mill.

550

Liner Profile and Speed Effects

Fine Grinding:
For fine grinding, it is desirable to have the charge
cascading rather than cataracting. This is achieved by
selecting a lower mill speed and/or using a wave liner
profile.
Impact Breakage
For breakage of larger feed particles, the grinding balls
should strike the charge close to the toe. Higher lifter bars
and mill speeds will assist.

275
 551

Classification and Circulating Load

Efficient classification is key to any closed grinding circuit.
Typical equipment include screens, classifiers or
hydrocyclones.
Typical SAG Circulating Load Ratio (CLR): 50-150%
Typical CLR for Ball Mills 250-350%.
Ensure that the classifier is performing well by analyzing its
behavior on a regular basis.
CLR is best measured by mass flow to the cyclones.
There is also a standard method which uses the size
distributions of the streams to derive a mass balance.
Use these techniques to check that the mill is grinding the
optimum tonnage by maintaining the target CLR.

552

Summary
A key aspect for sizing and selecting grinding mills is to
determine the power required.
Bond's equation works well, but must be modified with
efficiency factors.
Mill Power is also influenced by mill speed and liner
profiles.
Efficient classification is critical to an effective circuit.

276
 553

554

Grinding & Classification

Circuit Design Mill Sizing

277
 555

Sessions Outline

1. Types of Mill Equipment and Circuits

2. Factors In Grinding Circuit Selection
3. Ore Testing
4. Mill Power
5. Sizing Grinding Mills
6. Grinding Mill Design and Operation

556

Grinding Mill Sizing and Design

Introduction

Mill Sizing
Factors influencing mill power
Tumbling Mill Power Calculation and Sizing
Calculating Grinding Media Size and Consumption

General Mill Design Considerations

Drive Selection
Motor Selection
Mill Discharge and Feed SystemReference
Selection
Paper
Selection of Rod Mills, Ball
Mills and Regrind Mills,
Chester Rowland

278
 557

Introduction

Objective: to describe methods of sizing ball and rod mills

once the grinding power requirements for these have been
determined.

The approach to sizing SAG mills is fundamentally similar to

ball mills with modification for the effect of grates on the
charge, aspect ratio and pebble crushing.

558

Factors Influencing Mill Sizing

a) Mill Speed
b) Mill Diameter and Length
c) Mill Discharge Opening Size
d) Type of Discharge Mill Head
e) Amount/Size of Grinding Media
f) Feed Size
g) Feed SG
h) Ore Hardness
i) Feed Rate
j) Water addition (viscosity)

279
 559

Mill Power Factors: Mill Speed

Ps = KT
Where:
Ps = Power transmitted
through shaft from motor
K = Constant HP approximately
T = Torque proportional to speed
over wide range
= RPM

Net HP % of Critical Speed 100

560

Mill Power Factors: Mill Dimensions

Mill Diameter & Length

P D2.5 PL
Log (net HP per unit length)

HP proportional to
Average length
Slope = 2.5
Net HP

Small mills ~ 2.4

Large mills ~ 2.6

Log (Mill Diameter) Mill Length

280
 561

Mill Power Factors:

Discharge Opening Size
Mill Discharge Opening Size
Distance Between Load Centroid & Mill Center
Discharge Opening Rotation Direction

Distance from mill center to

centroid of load

As the discharge opening becomes smaller, the distance from

mill center to the centroid of the load becomes smaller, due to Discharge
shift in center of gravity of load. So, HP goes down (despite small Opening
increase in load). Increases
Net HP

Spread in curves
is exaggerated

% of Critical Speed

562

Mill Power Factors:

Discharge Head Type
Grate Discharge draw more power than Overflow
due to the distance from Centroid to Mill Center
Load N centroid is closer to mill center than Load P centroid

N = center of gravity of load for high discharge mill

N
P = center of gravity of load for diaphragm (grate) discharge mill
P
Mass of Solids at Position X

Load N

Load P
Area under curve is proportional to mass of load

Feed End Distance X Along Mill Length Discharge End

281
 563

Mill Power Factors:

Discharge Head Type
Overflow Discharge

Grate Discharge

Drum Feeder

Spout Feeder

564

Mill Power Factors: Amount/Size Grinding Media

Critical Speed is the speed at which a ball (m(R-r)w)w

or rod will be centrifuged in the mill. r

From force balance, Wc (rpm): mg

Radius R

76.63 Angular

Wc Dd
velocity, w
D = 2R
d = 2r

Due to change in mass Due to shift in centroid of

load and mass
Net HP

Net HP

60 70 80
% of Critical Speed % Media Load, by volume

282
 565

Mill Power Factors

Feed Size and Rate
From Bond Equation
P = KT[k 1F80]
K, k are constant and T is feed rate
As feed rate increases, P will increase and then level off

Feed Specific Gravity

The higher the specific gravity, the higher the power draw
SG = Power Draw

566

Mill Power Factors: Ore

From Bond Equation Hardness
P = KWi
Log (HP/T)

Slope of about 1/2

Work Index
Increases

Log P80

283
 567

Mill Power Factors: Feed Rate

- Startup

Centroid Shift causes less HP

Due to build up of draw
rock in the mill
Less power from slippage
Net HP

Power drawn with steel balls Steady State

and water only. Reached
At time = 0, cut in fresh feed solids

0 Time

-Steady State

Rod Mills and Ball Autogenous Mills

Mills
Net

Net
HP

HP

Feed Rate Feed Rate

568

Mill Power Factors: Water Addition Rate, Pulp Viscosity

Overflow Discharge: - Rod Mills : 80% solids

- Ball Mills 76% solids

Water Addition Rate Viscosity, Flow and Power Draw

Water Addition Rate Flushing Fines, Power Draw and

Wear
Ore containing clays can be
Pulp Viscosity,

Ice Cream Discharge

excessively viscous. The viscosity
can be reduced by adding:
Sausage
1. Water Discharge
cp

2. Polyacrylic acids
3. Calgon (phosphate dispersant) Pulp Density, gm per cc

284
 569

Mill Grinding Power and Sizing Calculation: Work Input

The mill work input to grind a tonne of feed of 80 %

passing size (F80) to a product passing size of 80 % (P80)
is calculated by the Bond equation:
W = 10 Wi [1/P80 - 1/F80]
where:
W = work input in kWh/t
Wi = Bond Work Index in kWh/t
Efficiency factors EF are applied to W to derive the
corrected power requirement WCOR
WCOR is multiplied by the new mill feed tonnage T to give
the mill power requirement P = T * WCOR
This is the power that must be applied at the mill drive in
order to grind the feed tonnage T from one size
distribution F80 to a finer product size distribution P80.

570

Mill Grinding Power and Sizing Calculation:

Matching Mill Size to Power Requirement

Once the mill power is determined the mill size to draw the
required power must be calculated.
Power draw theory is based upon a charge load in
equilibrium, and relates to its center of gravity.
The centroid of the charge is maintained in dynamic
equilibrium at an angle of repose A to the vertical by a
mechanical lever arm force balance between the mill drive
and charge weight.

285
 571

Mill Grinding Power and Sizing Calculation:

Matching Mill Size to Power Requirement
The figure below shows a section of a mill charge load in
equilibrium in relation to its center of gravity.
W = weight of charge
N D = Diameter
C = distance of center of
D gravity of charge from
center of mill in feet
C A = dynamic angle of
repose of the charge
N = mill speed in rpm
W
A

572

Matching Mill Size to Power Requirement:

DuPont Power Model
An early theoretical power model of DuPont (1900s) shows
the effect of charge weight, mill diameter and mill speed on
the power draw per unit of mill length (P/L) :

1. Charge Weight: P/L Mass W D2

2. Mill Diameter: P/L Lever Arm Length C (Centroid to Mill
Center) D
3. Speed: P/L Speed 1/ D 2

Therefore P/L D2 * D * 1/ D = D 2.5

or P D2.5 * L

286
 573

Matching Mill Size to Power Requirement:

DuPont Power Model
P D2.5 *L

This illustrates that:

- mill power is more sensitive to diameter than length
- the relationship between length and power is linear
- diameter affects power draw exponentially
- incremental changes in diameter provide step changes
in power draw

Therefore the selection of larger diameter (and fewer)

mills can significantly reduce the number of mills required
in an application.

574

Matching Mill Size to Power Requirement:

DuPont Power Model
Capital cost climbs steeply with diameter due to:
manufacturing methods
greater load on mill structure
more expensive drive systems

Therefore the number of mills required becomes a trade-off

between capacity and capital. In general, the larger the mill
selected the lower the overall installed capital.

There are practical limitations. Currently the largest mills are

about:
Rod Mill 15 ft diameter by 24 ft long ( 2,625 hp)
Ball Mill 26 ft diameter by 38 ft long (20,770 hp)
SAG Mill 42 ft diameter by 26 ft long (22,000 hp)

287
 575

Matching Mill Size to Power Requirement:

Nordberg Power Model- Theoretical Approach

Nordberg used a mechanical torque arm force balance

analysis to show that the theoretical power input (hp)
required to maintain the centroid of a mill charge in
equilibrium at an angle of repose A to the vertical is:

hp = K * (W) * (C) * Sin A * 2 * N

where:
K =1/33,000
W= weight of charge
C = distance of center of gravity of charge from center of
mill in feet
A = dynamic angle of repose of the charge
N = mill speed in rpm

576

Matching Mill Size to Power Requirement:

Nordberg Power Model
hp = K * (W) * (C) * Sin A * 2 * N
The model is based on the availability of data from similar
installations. If the value of the angle A can be found then
the power demand of mills with various diameters at the
same speed can be calculated.
However the value of angle A varies with:
the type of discharge
percent of critical speed
grinding condition.

Thus direct comparison can only be made between mills

with a similar type of discharge.

288
 577

Matching Mill Size to Power Requirement:

Nordberg Power Model
If various types of discharge are to be used, the following
factors must be applied for mills of the same size and speed:
Dry diaphragm = 1.0
Wet diaphragm = 0.9
Wet overflow = 0.8

In order to use the preceding Nordberg Equation, it is

necessary to have considerable data on existing installations.
Therefore, this approach has been simplified.

578

Matching Mill Size to Power Requirement:

Simplified Nordberg Power Model
The five basic conditions that determine the horsepower
drawn by a mill are:
1. Diameter
2. Length
3. Charge (% Loading)
4. Speed
5. Mill type
Nordberg incorporated these conditions into four factors
A,B,C & L to allow the calculation of the approximate
horsepower of a mill at the pinion drive shaft as follows:
HP = A * B * C * L

289
 579

Matching Mill Size to Power Requirement:

Simplified Nordberg Power Model
The Nordberg power factors for calculating rod and ball mill power are
on charts on pg 9 of the Nordberg Grinding Catalog, where:
A = factor for diameter inside the shell liners = D^2.5/5.6442
B = factor for mill type and charge volume (% loading) steel grinding
media
C = factor for mill speed expressed as a percentage of mill critical
speed
L = length in feet of grinding chamber measured between head liners
at the junction of the shell and head liners (Equivalent Grinding
Length EGL) in most cases subtract 6 from the length inside the
mill shell.

B factor is based on steel grinding media at 315 lbs per cubic ft. The
B factor must be adjusted by the ratio of the actual charge density or,
Factor =B x charge density/315.

580

Summary
Tumbling mills are sized to deliver the power required
to achieve the desired grind size.
Several factors influence power draw, including mill
speed, dimensions, type, feed size, type and rate.
Power draw theory is based upon a charge load in
equilibrium, and relates to its center of gravity
Mill power is more sensitive to diameter than length
Mill size can be estimated by the simplified Nordberg
Power Model.

290
 581

Grinding & Classification

Circuit Design Mill Sizing
Example

582

Nordberg Mill Sizing Example

- Calculation
Exercise: Size a single stage ball mill (overflow) in closed
circuit with a cyclone with the following parameters:

Feedrate = 500 tonne/h

F80 = 9,400 micrometers
P80 = 175 micrometers
RWI = 13.2 kWh/st
BWI = 11.7 kWh/st
CL = 250% Circulating Load
Cdensity = 340 lb per cubic ft
SG = 2.7
Ai = 0.25

291
 583

Nordberg Mill Sizing Example

Calculation
The feed to a standard Bond ball mill grindability test is
minus 6 mesh (3360m)

However the coarser fraction of a minus single-stage

ball mill feed is not included in the feed to the grindability
test mill

If RWI is different than BWI, then particularly if the former

is higher, a two step calculation should be used to
determine the grinding power input, using 2100 m to
divide the calculations.

584

Nordberg Mill Sizing Example Calculation

Calculate Uncorrected Grinding Power Input
Step 1:
W = 10 * (13.2 - 13.2 )= 1.52 kWh/st
2,100 9,400

Step 2:
W = 10 *(11.7 - 11.7) = 6.29 kWh/st
175 2,100

292
 585

Nordberg Mill Sizing Example Calculation

Calculate Uncorrected Grinding Power Input
Step 3:
Total = 1.52 + 6.29 = 7.81 kWh/st
= 7.81 * 1.102* 1.341* 500
= 5766 HP, uncorrected

Where:
Power (HP) = Power (kW) x 1.341
1 tonne = 1.102 short ton

586

Nordberg Mill Sizing Example - Calculation

Step 4: ApplyEfficiencyFactors

HPcorrected = HPuncorrected*Relevant EF1 to EF8 factors

EF1: Dry grinding. Does not apply.

EF2: Open circuit grinding. Does not apply.

EF3: Diameter Efficiency = (2.44/D)0.2 .

Mill will be larger than 3.81 m (12.5) in diameter so use
0.914.

293
 587

Nordberg Mill Sizing Example - Calculation

EF4: Oversize Feed = F Fo

Rr (Wi 7)
Fo
EF 4
Rr

Rr = Reduction Ratio = F80/P80 = 9400/175 = 53.7

Fo = Optimum feed size = Zf*(13/RWI)0.5
Zf = 4000 (ball milling)
Fo = 4000 * (13/13.2)0.5 = 3970
EF4 = 53.7 + (11.7-7) * [9400-3970)/3970] = 1.12
53.7

588

Nordberg Mill Sizing Example - Calculation

EF5: Fineness of Grind P80 > 75m, Does not apply.

EF6: High/Low Rr - Rod Milling, Does not apply.
EF7: Low Rr Ball Milling Factor, Does not apply.
EF8: Rod milling factor, Does not apply.
EF9: Rubber Liners Factor, Does not apply.

HP,corrected = 5,766 * 0.914 * 1.12 = 5903 HP

294
 589

Nordberg Mill Sizing Example - Calculation

Step 5: Select # of Mills
Assume use 1 mill: 5,903/1 = 5,903 HP Mill

Step 6: Size Mill Shell

HP = A * B * C * L

We dont know D so substitute factor A with D^2.5/5.6442

Charge density is 340 lb/cubic ft so multiply B by ratio of

340/315

5,903 = D2.5/5.6442 * B * (340/315) * C * L

590

Nordberg Mill Sizing Example - Calculation

Step 7: Determine factor B:

Most overflow discharge ball mills operate with a charge

volume that occupies 35% to 45% of the mill volume.
> 20 ft use 35%
< 20 ft > 16.5 ft use 40%
< 16.5 ft use 45%
An average value would be 40%. From tables,
Wet Overflow Ball Mills @ 40% Loading B = 5.02
The B Factor must be adjusted to the steel density of the
balls (340/315).
5,903 = D2.5/5.6442 * 5.02 * (340/315) * C * L

295
 591

592

Nordberg Mill Sizing Example - Calculation

Step 8: Determine factor C:

Critical Speed
The percent of critical speed (peripheral speed at which
charge centrifuges) is one of the major factors in
determining the power that a grinding mill draws.

To relate critical speed and peripheral speed as mill

diameter increases, the average recommended speed as
% of critical speed is shown in the table on the following
slide.

296
 593

Nordberg Mill Sizing Example - Calculation

Step 8: -cont-
Through an iterative Mill Diameter % of Critical
process, you can find the Inside Liners Speed

mill diameter will be >15 ft, Meters Feet Rod Ball

Mills Mills
therefore use a speed of
0.91-1.83 3-6 76-73 80-78
68% of .
1.83-2.74 6-9 73-70 78-75
2.74-3.66 9-12 70-67 75-72
From Nordberg tables
3.66-4.57 12-15 67-64 72-70
Factor C will be = 0.1583
> 4.57 >15 - 70-68
AND by substitution

5,903 = D2.5/5.6442 * 5.02 * (340/315) * 0.1583 * L

594

Factor C Ball Mill Sizing

297
 595

Nordberg Mill Sizing Example -

Calculation
Step 9: Estimate Mill L/D Ratio.

Recommended rod mill length to diameter inside liners is

1.4 to 1.6.

Being free from the limits imposed on rod mills, ball mills
have more variation in length to diameter ratios, ranging
from 1:1 to 2:1. The ratio used varies with
- the circuit type (type of grinding)
- size of the feed
- the ratio of reduction and specified fineness of grind
In general, as desired fineness of grind L/D

596

Nordberg Mill Sizing Example - Calculation

Ball Mill L/D Ratio General Application Guidelines

Type of Feed F80 Top Ball Size L/D Ratio

Grinding Microns mm in.

Wet 5,000 10,000 60 90 2.5 3.5 1:1 to 1.25:1

Wet 900 4,000 40 50 1.8 2.0 1.25:1 to 1.75:1

Wet or Dry Fine Feed Regrind 20 30 - 1 1.5:1 to 2.5:1

Wet or Dry Fine Feed- Open 20 50 - 2.0 2.0:1 to 3.0:1

Dry 5,000 10,000 60 90 2.5 3.5 1.3:1 to 2:1

Dry 900 4,000 40 -50 1.8 2.0 1.5:1 to 2:1

298
 597

Nordberg Mill Sizing Example - Calculation

Step 9: -cont-

Based on the preceding table, wet milling and a F80 of

9,400 mm and P80 of 175 microns (not fine regrind) a L/D
ratio of 1.25 is selected.

By substitution into equation:

5,903 = D2.5/5.6442 * 5.02 * (340/315) * 0.1583 * 1.25D

598

Nordberg Mill Sizing Example - Calculation

Step 10:
Calculate Mill Diameter (Inside liners):
Power Equation
5,903 = D2.5/5.6442 * 5.02 * (340/315) * 0.1583 * 1.25D

Rearranging:
31,075 = D3.5
D = 19.2 feet (inside liners)

L = 1.25 * 19.2
L = 24.0 feet

299
 599

Nordberg Mill Sizing Example - Calculation

Step 11:
Select Mill Shell Size:
Add new shell steel liner thickness (0.6 ft) to calculated
diameter
= 0.6ft + 19.2 ft = 19.8 ft (Closest standard is 20.0 ft)

Select a 20.0 ft inside shell diameter by 24.0 ft long overflow

ball mill @ 40% ball charge and running at 5,903 hp.

600

Nordberg Mill Sizing Example - Calculation

Step 12:
Calculate Maximum Ball Size for Mill Charge:

The equations for selecting the largest diameter (in inches)

rod (R) or ball (B) in the initial mill charge, or make-up
charge, for a mill of diameter D is:

R = (F800.75 /160) * [(SG * RWI) / (100 * Cs * D 0.5)]0.5

B = (F80 /K)0.5 * [( SG * BWI)/ (100 * Cs * D 0.5)]0.34
Mill Type Steel or C.I. Balls K
Ball Mill K Factor:
Wet Overflow 350
Wet-Diaphragm 330
Dry- Diaphragm 335

300
 601

Nordberg Mill Sizing Example - Calculation

Step 12: (cont.)

F80 = 9.4 mm, convert to microns
K = 350
SG = 2.7
Cs = 68%
BWI = 11.7
D = 19.8 ft
by substitution:
B = (9400/350)0.5*[( 2.7* 11.7)/(100*0.68*19.80.5)]0.34
B = 2.4 in. Closest standard is 2.5 in.

602

Nordberg Mill Sizing Example - Calculation

Step 13: Calculate rod/ball and shell liner consumption:

The following empirical equations use the abrasion index Ai
to estimate rod, ball, and liner wear rates.

Wet Rod Mills:

Rods kg/kw-hr = 0.1590 * (Ai 0.020)0.2
Liners kg/kw-hr = 0.0159 * (Ai 0.015)0.3

Wet Ball Mills:

Balls kg/kw-hr = 0.1590 * (Ai 0.015)0.34
Liners kg/kw-hr = 0.0118 * (Ai 0.015)0.3

301
 603

Nordberg Mill Sizing Example - Calculation

Step 13: (cont.)

Importance of Mill Liners

- Shell Liner Protect the Mill Shell

- Lifters attached to Liners which help distribute load for
grinding
- Lifter wear leads to loss of power
- SAG/AG Mills have white metal liners
- Ball Mill > 18 ft have white metal liners
- Ball Mill< 18 ft can have rubber liners (but note EF8)

604

Various liner materials and arrangements

302
 605

Typical Abrasion Index Values

Material Abrasion g Ai
Dolomite 0.0160
Schist-biotite 0.1116
Copper Ore 0.1472
Hematite 0.1647
Heavy Sulphides 0.1284
Magnetite 0.2517
Gravel 0.3051
Granite 0.3937
Quartzite 0.7751

606

Nordberg Mill Sizing Example - Calculation

Step 13: (cont.)

Substitute Ai = 0.25 into ball mill equations
Balls = 0.1590 * (0.25 0.015)0.34 = 0.0972 kg/kw-hr
Liners = 0.0118 * (0.25 0.015)0.3 = 0.0072 kg/kw-hr
Liner consumption typically ~10% of media consumption.

Multiply by the power draw (kw) and divide by feed rate of

500 t/h to give consumption in kg/t.

Balls = 0.0972 * (5903 *0.75) / 500 = 0.86 kg/t

Liners = 0.0072 * (5903 *0.75) / 500 = 0.064 kg/t

303
 607

Mill Discharge and

Feed Type Selection (Ball Mill )

Overflow Discharge: Grate Discharge:

Suitable for almost all 15-20% higher capacity
applications per unit volume
Simple and trouble-free. Coarser product with high
The discharge trunnion circulating load producing
can be furnished with a little extreme fines.
trommel screen. Can have a trommel
screen as well.

608

Mill Discharge and

Feed Type Selection (Rod Mill)

Overflow: Common for wet mill rod milling. Diameter of

discharge trunnion is larger than feed to promote flow.
End Peripheral Discharge: Used when a coarse product is
required.
Center Peripheral Discharge: Suitable for dry grinding at
extremely high capacities and coarse grinding, wet or dry.
Also applicable for viscous material and moisture content 3-
15% by mass.

304
 609

Example of Grate Discharge with Spout

Feeder Rod Mill

Spout Feeder

610

Nordberg Mill Sizing Example -

Calculation
Spout feeders normally feed rod mills
Spout feeders require at least 5 ft head between mill
center line and feed hopper for proper flow
Spout feeders are normally fed from ball mill cyclone
underflow box, requiring higher pumping heads relative to
a scoop or drum feeder.
There is a trade-off with scoop/drum drive power and
higher maintenance, such that in modern large mills the
scoop/drum feeder is rarely used.
Based on the exercise:
Select an overflow ball mill with a spout feeder.

305
 611

Summary
Mill sizing led to the selection of:
A 20.0 ft inside shell diameter by 24.0 ft long overflow
ball mill @ 40% ball charge and running at 5,903 hp.
Ball size: 2.5 in.
Ball wear estimate: 0.86 kg/t
Liner wear estimate: 0.064 kg/t
Feeder/Discharge Arrangement:
Overflow mill with a spout feeder

612

Motor Selection
Drive and Motor Types and Efficiencies
Motor Fixed Variable Speed

Drive LSS WR LCI CCV WR PWM

Single Pinion 4.5-10MW 0.950 0.932 0.922 0.916 0.899 0.924

Dual Pinion 9-20MW 0.950 0.912 0.922 0.916 0.899 0.924

Gearless 9-30MW NA NA NA 0.915 NA 0.923

LSS Low Speed Synchronous

Motor Key:
WR Wound Rotor
LCI Load Commutated Inverter
CCV Cycloconverter
PWM Pulse Width Modulated

306
 613

Motor Selection - Summary

In general the mills should be driven by a large enough motor to allow the
mill:

1) to operate with a 45% by mill volume charge with new liners and to
2) to overcome the drive train and motor efficiency

Based on the example calculation in the last section:

Select a single-pinion fixed speed drive with a low speed synchronous

motor.

Power required = 5903 hp at the mill pinion at 40% mill volume charge.

Based on previous Table: Drive efficiency = 0.95

Select motor size of 5903 * 45/40 * 1/ 0.95 = 6,990 HP = 5250 KW

BALL & ROD MILL SIZING

OLAV MEIJO

HATCH ENGINEERING

May 2013

307
 Presentation Outline

1. BriefintroductiontoBondstheory
2. LabWorkindex Equipment Procedure
3. Workindexcalculation
4. CorrectionFactorsappliedtotheLabworkindex
5. Calculationofthepowerrequiredforgrinding
6. Calculationofthemillpowerdraw
7. Allcalculationstogether
8. FAQs
9. SAGdesigntestmethods

61
5

Introduction

In 1930 Allis-Chalmers hired Fred Bond to carry out

research on size reduction of ores and grains.

Bond first task was to determine if the two existing

theories of comminution were right.

Bond found that Kick and Rittinger theories were wrong

and he proposed the third theory of comminution.

61
6

308
Introduction
Bonds second task was to develop a relationship
between ball mill operating data and grindability test
data.
Bond developed a grindability method to determine the
work index Wi test.

61
7

The Bond Work Index Wi

Theequationshownbelowisusedtodeterminethevalueofthe
workindexWibasedonthestandardBondgrindabilitylabtest.

61
8

309
 The Bond Work Index Wi

The feed for the Ball mill grindability test is 100% - 3350 microns and
80% -2100 microns

The feed for the rod mill grindability test is 100% -13200 microns.

The Wi Test corresponds to the motor output power Bond correlated to

an overflow discharge ball mill of 2.44 m(8 foot) internal diameter in
wet grinding conditions, closed circuit at 250% circulating load.

61
9

The Bond Work Index Wi

Doesitreallywork?

62
0

310
 Validity of the Bond theory

RelationshipbetweenEnergyconsumptionandparticle
size,R.T.Hukky1961(TakenfromthehistoryofgrindingbyAlbanJ.Lynch
andChesterA.Rowland)

62
1

Factors affecting the Bond work index

There are eight efficiency factors to be applied to the lab test work
index. The result obtained is the corrected work index:

Wi corrected = Wi test x EF1 x EF2 x.EF8

These factors are applied to take into account conditions observed in

real applications that differ from the bond lab test conditions.
The efficiency factors are:

62
2

311
 Factors affecting the Bond work index

EF1 Dry grinding

EF2 Open circuit ball mill
EF3 Diameter efficiency factor
EF4 Oversized feed
EF5 Fine grinding in ball mill product P80 less than 75 microns
EF6 High or low ratio of reduction rod mill
EF7 Low ratio of reduction ball milling
EF8 Rod milling

62
3

Efficiency Factors

EF1 : This factor is applied for dry grinding. The value is 1.3
EF2 : Open circuit grinding requires more energy than closed circuit
grinding and is a function of the product size. The table below shows
the values

(Chester A. Rowland and David M. Kjos)

62
4

312
 Efficiency Factors

EF3Diameterefficiencyfactoriscalculatedbasedontheballmilinside
diameterusedbyBond.

EF3=(2.44/D)0.2 andEF3=0.914whenD>3.81meters(12.5)

EF4Optimunfeedsizeisappliedwhenthefeedsizetoaball/rodmillis
coarserthattheoptimunsizeFo.

EF4=(R+(Wi7)(FFo)/Fo)/R

RatioofreductionR=F80/P80,

Fo=4000(13/Wi)0.5Forballmills

Fo=16000(13/Wi)0.5ForRodmills

62
5

Efficiency Factors

EF5:ThisfactorisapplywhentheP80isfinerthan75microns.Thisfactoris
calculatedusingtheequation:

EF5=(P80+10.3)/(1.145xP80)

EF6:ThisfactorisappliedforrodmillswhentheratioofreductionRishighor
lowoutsidetherangeRo=+/ 2:

EF6=1+(RRo)2/150
Ro=8+5L/D,
L:RodlengthD:internmilldiameter

EF7:ThelowratioofreductionfactorisappliedwhenRislessthan6

EF7=(2(R1.35)+0.26)/(2(R1.35)

62
6

313
 Efficiency Factors

EF8: This factor is apply to rod milling only. There are only
recommended values:

EF8=1.4 for open circuit crushing, rod milling only

EF8=1.2 for closed circuit crushing, rod milling only

EF8=1.2 for open circuit crushing and Rod mill-ball mill circuit

EF8=1.0 for closed circuit crushing and Rod mill-ball mill circuit and
rod mill F80 is less than 12 mm

62
7

Mill Power Draw

Bond developed an empirical correlation between power and ball

charge to determine power required for a wet grinding overflow ball
mill. The correlation was later modified by Rowland and Kjos. The
final equations is:

kW/st = 3.1 D0.3 (3.2 -3Vp) Cfs (1-(0.1/2(9-10Cfs)) + Ss

D = Mill diameter Inside liners in ft.

Vp = Mill volume fraction of balls
Cfs = Fraction of critical speed
Ss = Ball size factor
*For low level grate discharge mills applied a factor of 1.16 to the above
calculation.

62
8

314
 Mill Power Draw

Ss=(B3D/20)/2

B =Ballsizeininches
D =Milldiameterinsidelinersinfeet
Ss =Powerpershorttonofball

62
9

Mill Power Draw

63
0

315
Ball Mill Sizing All together

63
1

Ball Mill Sizing All together

63
2

316
 Ball Mill Sizing FAQs

CanIusetheBondballmillworkindexforsizingcrushers?
No.
TheBondworkindexisnotthesameasthebondcrusher
workindex.

Bondequationisinappropriatetodeterminetheenergy
requiredforcrushing.

Bondcrushingindexunderestimatesthepowerrequiredfor
crushing.

63
3

Ball Mill Sizing FAQs

Imsizingaballmill,whydoIneedtherodmillworkindex?

Energyrequiredforgrindingfrom13200micronstoaP802100
micronsiscalculatedbyusingtherodmillworkindexandthen
addedtotheballmillenergyrequiredfrom2100micronstothe
targetP80.Allballmillefficiencyfactorsshouldalsobeapplied.

63
4

317
 Ball Mill Sizing FAQs

WhydontusebigRodMills?

Rodmillsarelimitedincapacitybythemaximumrodlength.

63
5

Ball Mill Sizing FAQs

WhatstherelationshipbetweenRodmillworkindex(RWi),
Bondmillworkindex(BWi)andtheamountofpebbles
producedinaSAGmill?

RWi>BWi Specificenergyrequirementishigherinthe
largefractionthanthespecificenergyinthesmallfraction
thustheprobabilityofproducingpebblesishigher(to
2pebbles).

BWi>RWi Specificenergytogrindcoarseparticlesis
lowerthanthespecificenergyrequiretogrindfineparticles
thusthelikelihoodofformingpebblesislow.

63
6

318
 Ball Mill Sizing FAQs

ImusingtheBondequationtocalculatethemillpowerbut
itsnoevenclosetowhatthevendorsproposed.Whats
wrong?\

Doublecheckthatthecorrectionfactorsyouareusingare
right.IfstillItsnotclosetovendorscalculation,thevendor
iswrong!

63
7

JK drop weight test-JKSimMet

63
8

319
 JK drop weight test-JKSimMet

TheJKMethodusestwomethodstocharacterize
orebreakageatdifferentenergylevels.
1. Highenergylevelsarecharacterizedbyanimpact
breakagetestusingadropweightdevice
2. Lowenergylevelsarecharacterizedanabrasiontest
usingatumblingtest.TheabrasionparameterisTa
determinedbytheabrasiontest.
3. T10isthepercentagepassing1/10oftheoriginalsize.
Ecs isthespecificcomminution energy

63
9

JK drop weight test-JKSimMet

Tousetheresultsoftesting,theoretypeparametersAandb
andta areenteredintotheSAG/autogenous millmodelin
JKSimMet,
Thesimulationpredictsproductsizeandmillloadusing
appropriatebreakagerates.Thesimulatorcanthenalsobe
usedtopredictmillperformancewithvariationsinscreenand
classifierconfigurationsorevenwithrecyclecrushing.
PhantomCycloneinJKSimMet
Thephantomoverflowrepresentsthefinishedproduct
producedbytheSAGmillwhichwillrequirenoworkbytheball
millcircuit,asitwillreportdirectlytotheactualballmill
cycloneoverflow.Thephantomunderflowhowever,
representstheactualtonnageandf80ofmaterialonwhichthe
ballmillwillperformwork

64
0

320
 SMC Test:Steve Morrells Approach to Mill Sizing (1)

TheSMCTestgeneratesarelationshipbetweenspecificinputenergy(kWh/t)
andthepercentofbrokenproductpassingaspecifiedsievesize.
Theresultsareusedtodeterminethedropweightindex(DWi),whichisa
measureofthestrengthoftherockwhenbrokenunderimpactconditions.
TheDWi isdirectlyrelatedtotheJKrockbreakage parametersAandband
hencecanbeusedtoestimatethevaluesoftheseparameters

64
1

SMC Test:Steve Morrells Approach to Mill Sizing (1)

64
2

321
 SMC Test:Steve Morrells Approach to Mill Sizing (1)

64
3

Steve Morrell Approach to Mill Sizing (1)

Totat specific energy to reduce crusher product to final product size :

WT W a W b
where
W a specific energy to grind coarse particles
Wb specific energy to grind fine particles
The general size reduction equation is

The general size reduction equation

is as follows :
f ( x2 ) f ( x1 )
Wi KM i 4( x 2 x1
where :
Wi Specific comminutio n energy at a pinion (kWh/tonne )
K 1 for all fine circuits and all coarse circuits without a recycle pebble crusher and 0.95 where circuits do have a pebble crusher.
M i Work index related to breakage property of an ore (kWh/tonne );
x 2 80% passing size for the product ( m)
x 1 80% passing size for the feed ( m)
f(xj) - (0.295 x j / 1000000 ) (Morrel, 2006)
M ia is obtained directly from SMC test
18 .18
M ib 0.295 f ( p 80 ) f ( f 80 )
P1 (Gbp )( p 80 f 80 )

P1 closing screen size.

64
4

322
 Steve Morrell Approach to Mill Sizing (1)

Coarse particle breakage work index (Mia):

Mia = 19.5 kWh/t (from SMC test)

Fine particle breakage work index (Mib)

From the BBWI test results the ff values were obtained:

Gbps = 1.3 gr/rev

F80(um) = 2250
P80(um) = 78
P1(um) = 106
f(xj) - (0.295 x j / 1000000 )

f(p80) = -0.2951

f(f80) = -0.2973

18 .18
M ib 0.295 f ( p 80 ) f ( f 80 )
P1 (Gbp )( p80 f 80 )

Mib (kWh/t)= = 20.1136

64
5

Steve Morrell Approach to Mill Sizing (1)

Primary crusher product P80 (um) = 1500000

Final Product P80 (um) = 75

Coarse and fine particle comminution specific energy

f ( x2 ) f ( x1 )
Wi KM i 4( x2 x1

K 1
Coarse particle comminution specific energy

x1(um) 100000 Crusher Product

x2(um) 750 Definition
fx1 -0.395
fx2 -0.2958

Wa (kWh/t) = 10.184

Fine particle comminution specific energy

x1(um) 750
x2(um) 75
fx1 -0.2958
fx2 -0.2951

Wb (kWh/t) = 11.1477

Total comminution specific energy

WT (kWh/t) = 21.3317 64
6

323
 SAG design Test
(1)

TheSAGDesign testmeasuresthepinionenergytogrindorefrom80%
passing152mmto80%passing1.7mm(WSAG).
The2nd stageofthetestmeasurestheBondBallMillWorkIndexonSAG
groundore,SdBWI.
SAGDesign orefeedispreparedfromaminimumof10kgpersampleof
splitorwholediamonddrillcorepiecesbystagecrushingtheoreinajaw
crusherto80%productpassing19mm.
ThecrushedoreisthengroundinaSAGDesign SAGmill(489mminside
diameterx163mmEGL),seenhere,thatoperateswithparameterssimilar
tocommercialSAGmills(26%totalcharge;11%steelload,15%oreload;
androtationat76%ofcriticalspeed)

(1)http://sagdesign.com/home/productsandservices/sagdesigntest

64
7

SAG design Test

(1)

64
8

324
 SAG design Test
(1)

64
9

Plant Survey

Theobjectiveofaplantsurveycanbe:
1. toincreasethroughput
2. toincreasethefiness ofgrind
3. reducingthecostsassociatedtoenergyexpenditure
4. replacementofgrindingmedia
5. replacementoflifter

65
0

325
 Plant Survey

Thefinalrecommendationsmadetotheclientwillbetheoptimum
conditionsproposedfor:
1. Millfeedsize
2. Feedrate
3. Ballloadandsize
4. Percentagesolids
5. Dischargemechanism
6. Recyclecrushing
7. Millcirculatingload
8. Operationofcyclones
9. Milllinertypeandprofile
10. Throughput
65
1

Plant Survey

Streamstobesampled

1. ROMfeed(beltcut)TPH,PSD,%Sol
2. Cycl.Feed%Sol,PSD
3. Cycl.OF
4. Cycl.UF

65
2

326
 Application of Bonds Correction Factors in Trade-off
studies

Alternative 1 Alternative 2 Alternative 3 Alternative 4

Qty 2 new + 1 future 1 new + 1 future 2 new 1 new + 1 future
Equipment Dimensions 13x19 16.5x21 15x19 16.5x21
Equipment Arrangement Parallel Parallel Parallel Series
Project current capacity
Project future capacity
Required power @
2640 2520 3680 2520
current capacity [kW]
Required power @ future
3960 2520 N/A 4180
capacity [kW]
Immediate Equipment
Installation Cost @ R$ 42.9 M R$ 32.2 M R$ 53.8 M R$ 32.2 M
Project current capacity
Additional Equipment
Installation Cost for future R$ 23.0 M R$ 22.9 M N/A R$ 32.2 M
capacity
Final Equipment
Installation Cost after R$ 65.9 M R$ 55.1 M R$ 53.8 M R$ 64.4 M
future expansion

65
3

Application of Bonds Correction Factors in Trade-off

studies

65
4

327
 Application of Bonds Correction Factors in Trade-off
studies

65
5

SAG Mill Power Draw

Primary SAG Mill

Mill Operating Parameters and Power Required:

Daily Feed Tonnage 35000 tpd

Mill Availability 92 %
Mill Feed Rate 1585 tph
Feed Size F80 150000 um
Product Size P80 2500 um
SAG Mill Work Index 15.5 kWh/t
SAG Efficiency Factor 1.5
Transmission Loss Factor 1.05
Unit Power Consumption 4.25 kWh/t
Mill Power Required 6740 kW
Mill Power Required 9039 HP
SAG Mill Power Installed 10000 HP

Fit of Mill Size to Motor Size:

Number of SAG Mills 1

Power Installed per SAG Mill 10000 HP
Mill Outside Diameter 32 ft
Mill Length-EGL 16 ft
Percent of Critical Speed (VS) 76 %
Mill Speed, rpm 10.38
Percent Volume Total Charge 28 %
Percent Volume Steel Charge 8 %
Tons of Steel Charge 149.27
Ore Specific Gravity 2.8
Slurry Pulp Density 70 % sol
Slurry Specific Gravity 1.82
Charge Specific Gravity 3.40
Charge Density, lb/ft 212.03
Mill Power Draw 7920 kW 65
Mill Power Draw 10621 HP
6

328
 SAG Mill Power Draw

65
7

SAG Mill Power Draw

SAG MILL PARAMETERS

Liner thickness m 0.10-0.15 m
Fraction of crit. speed 0.72 to 0.75
Ball volume nominal design % 10-15%
Ball volume max operating % Max operating
Ball volume structural design % 2-5% above max operating
Total filling nominal design % 26% for SAG mill, 28% for AG mill
Total filling max operating % 30-35%
Discharge slurry % solids % solids 65% to 78%, typically 72%
Discharge mechanism Grate discharge for SAG or AG mill

BALL MILL PARAMETERS

Liner thickness m 0.075 m
Fraction of crit. speed 0.72 to 0.76
Ball volume nominal design % 38% to 40%
Ball volume max operating % 38% to 40%
Ball volume structural design % 40%
Discharge slurry % solids % solids 65% to 78%, typically 70%
Discharge mechanism Overflow for ball mill
Discharge screen Trommel

65
8

329
 65
9

660

Size Classification

Primary Sources: B Klein, AJ Gunson

330
 661

Classification - Outline
1. Introduction
2. Wet Size Classification
Principles
Types of classifiers
Factors affecting performance Reference Paper
Separation efficiencies Hydrocyclone Selection
for Plant Design
3. Hydrocyclone Classifiers Timothy Olson and
Hydrocyclone description Patrick Turner

Geometry variables
Process variables
Cyclone sizing & selection criteria
Example calculation
4. Maintenance and Optimization

662

Size Classification Introduction

Size classifiers (water or air)
separate particles of various
sizes, shapes and specific
gravities under the influence of
gravitational or centrifugal
forces.
Size classifiers enhance the
effect of particle size over other
properties to produce a size split.
Size classification is critical to
achieving the target particle size
in order to ensure efficient
valuable mineral recovery.

331
 663

Size Classification Introduction

Factors that influence size separation:
Small particles settle slower than large particles.
In free vortex motion, centrifugal forces affect movement
of large particles more than small ones.
Small particles have less inertia and therefore flow with
liquid or suspending medium.
Large particles require higher conveying velocity.
Collision Frequency Increases with particle size.

664

Size Classifier Categories

Physical - Screens
Wet Classifiers (Water)
Mechanical
Spiral Classifiers
Rake Classifiers
Non-Mechanical
Cones
Hydraulic Classifiers - jigs
Hydrocyclones
Pneumatic (Dry) Classifiers
Cyclones

332
 665

2. Classification Principles
Cut Size (separation size) has many definitions
Size which passes 95% of the overflow
Size at which cumulative percent passing in the
overflow equals the cumulative percent coarse in the
underflow
X50 as determined from fractional recovery curve.
X50C as determined from corrected fractional
recovery curve

666

2. Classification Principles
Fractional Recovery to the underflow stream
Ri = Uui/Ffi
Where U = tph of dry solids in underflow
Ui = weight fraction retained in size interval i in
underflow
F = tph of dry solids in feed
Fi = weight fraction retained in size interval i in feed

333
 667

Classifier Performance

668

Classifier Performance
Classifier efficiency is measured by imperfection of
separation, I

d75 = Size at which 75% passes to U/F d 75 d 25

d50 = Size at which 50% passes to U/F I
d25 = Size at which 25% passes to U/F
2d 50

I = 0 means perfect separation

334
 669

Classifier Performance

670

Classifier Performance
To correct a partition curve

yi R f
yi '
1 Rf
Where:
yi = Corrected recovery of i
yi = Uncorrected recovery of i
Rf = Recovery of water to coarse fraction

335
 671

Do I have the correct curve?

672

Rake & Spiral Classifiers

length

feed Classification
variables:
Fluid velocity
-Feed rate

Wier
-Particle size, shape,
height SG
Coarse
Bottom material
-Tank geometry
slope
Rake or (length, slope
Spiral
Fine freeboard)
material
- Rake/spiral velocity
(2-10 rpm)

336
 673

Rake & Spiral Classifiers

Adjust rake travel and frequency, spiral rpm
Balance transport velocity against turbulent environment
Adjust weir height to achieve correct cut point
Can use wash water sprays to clean coarse fraction
Dont feed into pool agitation

674

Rake & Spiral Classifiers

Longer spiral for dewatering applications
Spirals classifiers can be
steeper than rake classifiers
100 1000 um

337
 675

Settling Cones

Used in desliming or dewatering applications

676

Jig Classifiers

Finer material shorter strokes, greater frequency

Coarser material longer strokes, lower frequency
Better suited to density classification

338
 677

3. Hydrocyclones

678

Hydrocyclones - Introduction
Hydrocyclones are mainly used in mineral processing
classification flowsheets.

1.Hydrocyclone Description
2.Process and Geometry Variables
3.Efficiency and Performance
4.Cyclone Selection Criteria
5.Example Calculation

339
 679

Why use hydrocyclones?

Small footprint
Low capital expenditure
No moving parts
Reliable
Efficient
Can achieve fairly dense underflow

Often abused in mineral processing plants!

Typically a good place to begin optimisation

680

i) Hydrocyclone Description

340
 681

Hydrocyclone Description

682

ii) Hydrocyclone Classification

Geometry Variables:
Inlet Area
Determines entrance
velocity and affects
tangential velocity
profile. Rectangular
are most common.
Increased area
requires increased
flowrate to maintain
tangential velocity.
Inlet Area is typically 6
to 8% of cross-
sectional area of feed
chamber.

341
 683

Variables Affecting Cut Point

Parameter Change (Increase) Cut Point Change

(coarseness of U/F)

Cyclone Diameter
Vortex Finder Diameter
Apex Diameter
Barrel Length
Cone Angle
Inlet Pressure
F80
Feed SG
Fluid Viscosity

684

Variables Affecting Capacity

Parameter Change (Increase) Capacity Change

Cyclone Diameter
Vortex Finder Diameter
Apex Diameter
Inlet Pressure
Inlet Area

342
 685

iii) Cyclone Selection Criteria

Based on experimental studies and field work, the relationship
for cyclone diameter is as follows:

D = 0.02338(1-V/Vm)2.167 (x50c)1.515 (P)0.4242 (s - l)0.7576

where
D is cyclone diameter in cm
V is the volume percent solids in the Feed
Vm is maximum percent solids = 53%
x50c is cut size in m
P is the inlet pressure in kilopascals (100Kpa = 14.5 PSI)
s & l are specific gravity of solid and liquid

686

Cyclone Selection Criteria

The cut size can be estimated from the equation:

X50c = 3.14 (dy) Ln(119.12/yd)

where
yd is the cumulative % finer than size dy (m)
Example: If target P80 is 150 m, yd = 80, dy = 150 m
We require:
1. Water and Solids Balance on Weight and Volume Basis
2. Determine Cyclone Diameter
3. Determine Number of Cyclones
4. Estimate Inlet Area
5. Estimate Vortex Finder Diameter
6. Estimate Apex Diameter (Spigot Size)

343
 687

Cyclone Design Rules of Thumb

Inlet AI = 0.05 Dc2
Vortex Finder Do = 0.35 Dc
(can be 0.2 0.45 Dc)
Apex Du 0.2 Dc
Du/Do < 0.45 Rope
0.45 < Du/Do < 0.56 Rope or
Spray
0.56 < Du/Do < 0.90 Spray
Cone Angle = 10o 20o

688

iv) Example Calculation

Problem: Select cyclones for the following circuit
Po, O, Wo Po = 36.5%
80% passing 150 micron in O
(U/F) = 4

CYCLONES
Pt
Solids SG = 3.2
T
Rod Mill Feed, F = 250 stph
Pu
Wt
U
Pt must be greater than
55% solids by weight Wu

ROD MILL PUMP BALL

BOX MILL

Water

344
 689

Balance Across Cyclone

Overflow
250 stph solids
36.3% solids
Feed
P80 = 150 um
? stph
> 55% solids
F80 = Determine
unknowns for
solids and water
balance
Underflow
? % solids
? stph
P80?

690

Task #1: Water and Solids Balance

Step 1: Select U/F solid content to prevent roping.
Roping When too high a density of solids reports to the underflow plugging the
apex. This results in coarse material reporting to the overflow

100
90 From graph, for O/F
Overflow (O/F) % Solids by Weight

80
70
solids of 36.5%, the
60 maximum U/F solids
50 is approximately 81%
40

30 Select U/F Solids =

80% by weight
20
Roping is probable
to the right of each
15
curve

10
75 76 77 78 79 80 81 82 83 84 85 86 87 88
Underflow (U/F) % Solids by Weight

345
 691

Water and Solids Balance

Classifying Cyclone Mass and Volume Balance:
Circulating Load: 400%
Solid S.G.: 3.2, Water SG.: 1.0
Feed = O/F
Cyclone Feed = O/F + U/F
Weight % Solids = 100 x STPH Solids / STPH Slurry
Volume % Solids = 100 x USGPM Solids/ USGPM Slurry

Product Solid Liquid Slurry Solids (%)

STPH USGPM STPH USGPM STPH USGPM STPH USGPM
Feed 250
O/F 250 313 434.9 1740 685 2052 36.5 15.2
U/F 1000 1250 250 1000 1250 2250 80.0 55.6
Cycl. 1250 1563 684.9 2740 1935 4302 64.6 36.3
Feed

Note: USGPM = STPH (4/SG)

692

Balance Across Cyclone

Overflow
250 stph
Feed 2052 USGPM
1250 stph 36.3% solids
4302 USGPM F80 = 150 um
64.6% solids

Underflow
1000 stph
2250 USGPM
80% solids

346
 693

Task #2: Select Cyclones

Step 1: Determine Cyclone Diameter
D = 0.02338 (1V/Vm)2.167 (X50c)1.515 (P)0.4242 (t)0.7576

V = 36.3%
Vm = 53%
X50c = 3.14(dy) ln(119.12/yd)
where dy = 150 m
yd = 80%
X50c = 187.5 m
P = 8 psi = 8 x (100/14.5) = 55.17 kPa
(t) = 3.2 1.0 = 2.2

D = 52.8 cm = 20.8 inches

Since 20 inch is a standard size, select as cyclone
diameter

694

Determine Number of Cyclones

Step 2: Determine Number of Cyclones
Total flow to cyclones, V = 4,303 USGPM
The estimated capacity for a single cyclone based on water flow is:

Q = 0.7071 D2P
= 0.7071 x (20)2 x 8
= 800 USGPM
Number of Cyclones = V/Q = 5.38
Say 6 cyclones. For extra capacity, select 7 cyclones

Step 3: Estimate Inlet Area

Inlet area = 0.05 D2
= 20 square inches

Step 4: Estimate Vortex Finder Diameter

Vortex Finder Diameter = 0.35 D = 0.35 x 20
= 7 inches

347
 695

Cyclone Selection
Step 5: Estimate Apex Diameter (Spigot Size)
Minimum diameter below which roping will occur:

S = 4.16 16.43 / [2.65 - + (100/Pu)] + 1.10 ln(U/)

where = 3.2
Pu = 80%
U = 1000 tph / 6 cyclones (ie. assuming 6 cyclones)
S = 3.29 inches
Therefore use a spigot with a diameter of 3.29 inches or greater
(say 4)

Summary
Select 7 x 20 inch diameter cyclones for cut size of 187.5um at Feed
of
4303 USGPM with 55% solids:

- Inlet area of 20 square inches

- Vortex finder diameter of 7 inches
- Apex diameter of at least 4 inches

696

Operational Aspects
Correct underflow fan, 20o and hollow centre
Low % solids in feed, high % solids in underflow
Minimise underflow tail (fines in underflow)
Maintain correct spigot size
Maintain correct feed pressure
Low % solids in feed

Roping Conditions
Du/Do < 0.45 Rope
0.45 < Du/Do < 0.56 Rope or Spray
0.56 < Du/Do < 0.90 Spray

348
 697

Optimization
Number of Cyclones Pressure d50c
Apex smaller Du = larger d50c & lower water recovery to UF
Vortex Finder Larger Do = larger d50c & lower water
recovery to UF
Feed Water Diluting feed slurry reduces fines tail and may
improve efficiency (particle-particle interactions)
Cyclone Diameter larger Dc = Larger d50c (& lower wear &
pumping costs)

698

Maintenance Aspects
Right size apex and vortex finder
Good liner condition
No odd wear patterns, not worn through
Liners correctly installed, no steps
Functional distributor
Unbiased flow patterns, clean pressure ports
Pump well maintained
Be careful when changing apex / liners
Dont drop parts into the launder!

349
 699

Reference
Texts:
Wills, Barry, 1997, Mineral Processing Technology, 6th
Ed
Napier-Munn, T., Morrell, S., Morrison, R., Kojovic, T.,
1996, Mineral Comminution Circuits: Their Operation
and Optimization
Papers:
Timothy Olson and Patrick Turner, Hydrocyclone
Selection for Plant Design,
http://www.krebs.com/literature.php/hardrock_mining/
Richard Arterburn, The Sizing and Selection of
Hydrocyclones,
http://www.krebs.com/literature.php/hardrock_mining/

Fine Grinding

350
 References:
Mark Adams, Mine 331, ubc, presented on Nov., 2012
www.outotec.com
www.isamill.com
www.metso.com
www.flsmidth.com
Burford and Clark, 2007. IsaMillTM technology used in efficient grinding circuit.
Gao and Forssberg, 1995. Prediction of product size distribution for a stirred ball mill
Kelly and Spottiswood, 1982. Introduction to mineral processing
Hogg and Cho, 2000. A review of breakage behavior in fine grinding by stirred-media milling
Jankovic, 2003. Variables affecting the fine grinding of minerals using stirred mills
Larson, Anderson, Morrison and Young. Regrind mills: challenges of scaleup www.isamill.com
He, Wang, Forssberg, 2004. Slurry rheology in wet ultrafine grinding of industrial minerals: a
review
Parry, 2006. Ultrafine grinding for improved mineral liberation in flotation concentrates
Tong, Klein, Zanin, Skinner, and Robinson, 2012. Stirred milling of siliceous goethitic nickel
laterite batch grinding study
Drozdiak, Klein, Nadolski, and Bamber, 2011. A pilot-scale examination of a high pressure
grinding roll/stirred mill comminution circuit
Wang, Nadolski, Mejia, Drozdiak, and Klein, 2013. Energy and cost comparisons of HPGR
based circuits with the SABC circuit installed at the Huckleberry mine
Roufail, Klein, and Radziszewski, 2012. Morphological features and discrete element method
(DEM) forces produced in high speed stirred mill

Outline
Introduction
Fine Grinding Technologies
Fine Grinding Flowsheet (IsaMill)
Grinding Mechanisms and Conditions
Case Studies
Sizing and Scale-up
Selection Criteria
Conclusion

351
 Introduction
Emergence of fine grinding
Most of the worlds high-grade, coarse-grained deposits have been
depleted
Especially in the latter half of the 20th century, attention has turned to
the mining of low-grade, fine-grained deposits
These fine-grained deposits have necessitated fine grinding to
produce the liberation grind sizes required for downstream processes
to succeed and to do so efficiently enough to make the process
economically viable

Introduction
Example Necessity Breads Innovation
In the 1980s, Mt Isa Mines (now Xstrata) owned the McArthur River Pb-
Zn ore body, which required a 7 m grind for liberation
Existing grinding technologies were tested, but resulted in the orebody
being uneconomical
- Power consumption too high
- Generally ineffective below 20 m
- Poor flotation due to negative influence of steel grinding media
MIM looked outside of mining for a solution and partnered with Netzsch,
who manufactured small stirred mills for other industries
Once scaled-up, this technology was known as the IsaMill and became
enabling technology for start-up of McArthur River Mine

352
 Introduction
Goal of Grinding
The goal of a grinding machine is to use electrical energy to do work on
ore as efficiently as possible (i.e., with as few losses as possible)
Electrical Energy = Mechanical Energy
+ Sound, Thermal, etc energy Losses
There is no difference in fine grinding, except that more energy is
generally required to break finer ore

Introduction
Grinding Energy Curve

353
 Introduction
History of Fine Grinding
1870s: Ball mills are first used for grinding on industrial-scale
1900s: Fine grinding is practiced extensively in ceramic, paint and pharmaceutical
industries using different small-scale mills
1953: Tower Mill is developed by Nichitsu Mining Industry in Japan
1960s: Stirred Media Detritor (SMD) is developed by English China Clays in UK
1980: First Tower Mill is installed in a mining application
1980s: Mt Isa Mines (now Xstrata) partners with Netszch in Germany to scale-up their
horizontal stirred mill for mining applications
1991: Tower Mill license acquired by Svedala (now Metso) and renamed the Vertimill
1994: First production-scale horizontal stirred mill installed by Xstrata and renamed the
IsaMill
1996: First SMD is installed in a mining application when license is acquired by Svedala
(now Metso)
2000s: Other fine grinding mills are acquired/licensed to FLSmidth and Outotec who bring
them into mining industry

Fine Grinding Technologies

Technologies and Typical Grinding Range

Technology Type Typical Grinding Range: m

Ball Mill 50-10000

Vertimill 20-6000

IsaMill 5-400

SMD 5-100

HIGmill Under development

VXPmill Under development

354
 Fine Grinding Technologies
Ball Mill
First used in mining in the 1870s
Multiple manufacturers around the world
Horizontal configuration
Normally closed-circuit with cyclones
Cyclone inefficiency (fines bypass) often
leads to overgrinding
Steel media (25-90 mm or 1-3.5) up to
45% full
Power intensity: 20 kW/m3
Generally accepted as less efficient than
stirred milling below 100 m product sizes ( http://www.flsmidth.com
and ineffective below 20 m)

Fine Grinding Technologies

Vertimill
First used in mining in 1980

Metso has license to market to mining

Vertical configuration open to atmosphere

Screw agitated

Top fed, bottom discharge

Open or closed circuit with cyclones

Steel media (12-37 mm or 0.5-1.5)

Power intensity: 40 kW/m3

Operating speed: 3 m/s

Generally considered inefficient below 20 m

Obtained from http://www.metso.com

355
 Fine Grinding Technologies

Stirred Media Detritor (SMD)

First used in mining in 1996

Metso has license to market to mining

Vertical configuration open to atmosphere

Pin agitated

Screens to retain media

Top fed, top discharge

Open or closed circuit with cyclones

Sand or ceramic media (2-5 mm)

Power intensity: 60 kW/m3

Operating speed: 3 m/s

http://www.metso.com

Fine Grinding Technologies

VXPmill
First used in mining in 2006

Previously named the Deswik Mill

Manufactured by FLSmidth

Vertical configuration open to

atmosphere

Disc agitated

Disc spacing and number variable (up to

16 discs)

Bottom fed, top discharge

Ceramic media (2-2.5 mm) up to 80%

full

Operating speed: 10 m/s

www.flsmidth.com Deswik Mill at UBC

356
 Fine Grinding Technologies

HIGmill
New to mining industry as of 2012

Outotec has license to market to mining

Vertical configuration open to atmosphere

Disc agitated

Disc spacing and number variable (up to 30 discs)

Bottom fed, top discharge

Normally open circuit with cyclones

Ceramic media up to 70% full

http://www.outotec.com

Fine Grinding Technologies

IsaMill
First used in mining in 1994

Xstrata Technology has license to market to mining

Horizontal configuration operating under pressure

Disc agitated

Internal classifying system produces steep particle

size distribution and less overgrinding than others

Normally open circuit with densifying cyclones

(operates at 40-60% solids)

Ceramic or sand media (1-6 mm) up to 75% full

Power intensity: 300 kW/m3

Operating speed: 20 m/s

http://www.isamill.com

357
 Fine Grinding Flowsheet
McArthur River Zinc/lead mine, M3000 IsaMill, Feed: P80 70 m, Product: P80 7 m

Burford and Clark, 2007

Fine Grinding Flowsheet

Simplified Potgietersrust Platinum mine C-Section (Anglo Platinum)
Flowsheet with a M10,000 IsaMill

Media: 3.5 mm MT1, Feed: P80 75 m,

Product, P80 < 53 m, Energy
consumption: 9 kWh/t
Burford and Clark, 2007

358
 IsaMill Grinding Mechanism

Burford and Clark, 2007

Product Size vs. Energy Usage

Jankovic, 2003

359
 Size Reduction Mechanisms

Kelly and Spottiswood, 1982,

Hogg and Cho, 2000
Gao and Forssberg, 1995

Grinding Conditions
IsaMill
Media Size
Media Fill
Stirrer Speed
Solid Content
Feed Size
Flow Rate
pH Control
Additive Addition
M20 Stirred mill at the NBK Institute of Mining

360
 Grinding Conditions
Grinding Media
The goal of a grinding machine is to use electrical energy to do work on
ore as efficiently as possible (i.e., with as few losses as possible)
Grinding medias job is to transfer energy from a grinding machine
to the ore for breakage
The majority of energy losses in grinding occur in the transfer of energy
from the machine to the ore
Since grinding media is the conduit for energy to get from the machine
to the ore, it is vitally important

Grinding Conditions
Grinding Media
Energy Transfer in a Grinding Media
Medias Energy
What makes up medias energy?

Examples of Grinding Media

Burford and Clark, 2007

361
 Grinding Conditions
Cost of Grinding Media
Grinding media is often the 3rd highest cost in processing behind energy
and labour
Proper media selection can improve economics by:
- Reducing its own cost through price and wear improvements
- Reducing energy usage through more efficient energy transfer from
grinding machine to ore

Media Selection
Type, Size, Supplier and Model, Price

Grinding Conditions
Slurry Rheology
Slurry rheology significantly influences
the grindability of industrial minerals in
wet ultrafine grinding
Parameters: mineralogy, solid
concentration, particle size and
distribution, particle shape, temperature,
rotation, pH, and dispersants
Rheology optimization to increase
throughput, energy efficiency and
product size

He et al., 2004

362
 Case Study - 1

Ultrafine grinding for Improved Mineral Liberation in

Flotation Concentrates
Parry, 2006
Objectives: Effect of stress intensity on breakage rates for minerals of
different hardness; Effect of mill type on grinding energy requirements;
Effect of stirred milling on downstream processing in terms of particle size
distribution and mineral liberation
Results: It is possible to target either hard or soft minerals for liberation in
stirred milling; Mineral liberation behavior was similar for the horizontal
and vertical high-speed stirred mills. The greatest benefit of regrinding
using high-speed stirred mills was improved quartz liberation.

Case Study - 1

Netzsch LME4 stirred mill at UBC Laboratory 1.5 L batch SMD at UBC

363
 Case Study - 1
By varying the stress intensity it is possible to target either hard or soft minerals
for liberation Selective comminution was suggested in stirred milling

Effect of Stress
Intensity

Case Study - 1
The greatest benefit of regrinding using high-speed stirred mills was improved
quartz liberation

Netzsch mill
products

364
 Case Study - 2
Stirred Milling of Siliceous Goethitic Nickel Laterite to
Upgrade Ni
Tong, Klein, Zanin, Skinner, and Robinson, 2012
Based on the differences in the mechanical properties of mineral
components in ores, selective grinding was investigated to update
valuable minerals --- properties of mineral
Previous study indicates an opportunity for selective size reduction of
particles of differing hardnesss using a stirred mill --- mill
At low stirrer speed, soft minerals break faster than hard ones. Breakage
of the softer or harder components in an ore can be targeted by adjusting
the stress intensity in stirred mills --- grinding conditions
Results: The breakage rates with respect to sample mass for Ni, Mg, and
Si indicate that: Mg>Ni>Mass>Si. The optimum grinding time for the
highest Ni upgrade was 0.25 min. The Ni grade increased from 0.88% to
1.35%, with 24% Ni recovery

Case Study - 2

Batch grinding tests:

Netzsch LME4 stirred mill at UBC Feed size: -2000 m
Product: 38 m

365
 Case Study - 2
Effect of grinding time on the breakage of +38-2000 m siliceous goethitic
nickel laterite particles: 20 wt% solid, 1000 rpm, 50% charge volume
Time 0-2000 All - 38 m All + 38 m - 38 m particles
min m product product from milling +38
Grade, % m feed
wt% Grade, wt% Grade, wt% Grade,
% % %
0 1.14 46.0 1.44 54.0 0.88 0 0
0.25 1.14 54.3 1.43 45.7 0.79 8.3 1.35
0.5 1.14 57.8 1.43 42.2 0.74 11.8 1.37
1.0 1.14 63.6 1.39 36.4 0.70 17.6 1.25
2.0 1.14 70.4 1.36 29.6 0.62 24.4 1.19
3.0 1.14 74.8 1.33 25.2 0.59 28.8 1.12
4.0 1.17 78.6 1.33 21.4 0.58 32.6 1.07

Case Study - 2

Effect of grinding time on the weight fraction remaining on 400 mesh screen and the
specific rate of breakage: 20 wt% solid, 1000 rpm, 50% charge volume, siliceous
goethitic nickel laterite (38-2000 m)

366
 Case Study - 2

Effect of grinding time on the grade changes and recovery with respect to elements: 20
wt% solid, 1000 rpm, 50% charge volume, siliceous goethitic nickel laterite (38-2000 m)

Case Study - 3
A Pilot-Scale Examination of a High Pressure Grinding
Roll / Stirred Mill Comminution Circuit
Drozdiak, Klein, Nadolski, and Bamber, 2011
Cone crusher / ball mill, HPGR / ball mill, HPGR / stirred mill circuits were
examined on Mesaba copper-nickel deposit, feed size: F80: 21 mm, P80:
75 m
Results: based solely on the specific energy requirements for
comminution, the HPGR / stirred mill circuit achieved a reduction of 9.2%
and 16.7% over the HPGR / ball mill and core crusher / ball mill circuits,
respectively

367
 Case Study - 3

Pilot-scale HPGR installation at UBC M20 stirred mill at UBC

Case Study - 3

HPGR / Stirred Mill

flowsheet A

HPGR / Stirred Mill

flowsheet B

368
 Case Study - 3
Summary of results for the first-stage HPGR operating in open (Circuit A) and closed
(Circuit B) circuit

Case Study - 3
Summary of stirred mill operating conditions

Stirred mill signature

plot results

369
 Case Study - 3
Summary of specific energy consumption for each circuit

Case Study - 3
Proposed layout for an HPGR / stirred mill circuit

370
 Case Study - 4
Energy and Cost Comparisons of HPGR Circuits with the
SABC Circuit Installed at the Huckleberry Mine
Wang, Nadolski, Mejia, Drozdiak, and Klein, 2013
To summarize a comprehensive energy and cost study comparing an
existing SAG-ball mill circuit with HPGR-ball mill and HPGR-stirred mill
circuit
Results:
The HPGR-ball mill circuit achieved a 21% reduction in energy
consumption over the existing SAG-ball mill circuit at the same P80 grind
size of 160 m
At a grind of 80% passing 75 m, the HPGR-stirred mill circuit showed a
34% reduction in energy compared to the base case
The energy reduction for the new flowsheets significantly improved the
economics of the Huckleberry comminution duty

Case Study - 4

371
 Case Study - 4

Huckleberry SABC circuit base case

Case Study - 4

HPGR ball mill

circuit

HPGR
stirred mill
circuit

372
 Case Study - 4
Summary of stirred mill test conditions and results

Stirred mill signature

plots

Case Study - 5
Morphological Features and Discrete Element Method
(DEM) Forces Produced in High Speed Stirred Mill
Roufail, Klein, and Radziszewski, 2012
The effect of different operating parameters of high speed stirred mill on
the particle breakage mode is addressed. Morphological features of
broken particles is investigated and related to the types of forces
generated in the mill using Discrete Element Method (DEM)
Results: About 60% of total particles that are morphologically analyzed
for quartz and galena were rough particles. Such an observation indicates
that the fine product are broken via fracture. The agitator speed, mineral
type, and residence time would dictate the type and mode of particle
breakage

373
 Case Study - 5
Morphology Roughness Level Definitions and Illustration

Breakage Mode versus Roughness Level

Roughness
Breakage Mode
Level
- Started Abrasion
R1 (Transgranular)
Hammered - Then Exposed to Impact
(Indents on Surface)
R2 Abrasion
Smoothest (Transgranular)
R3 Exposed to both Abrasion and Fracture
Semi-Rough (Transgranular and Intergranular)
R4 Fracture
Rougher (Intergranular)
R5 Fracture
Roughest (Intergranular)

Case Study - 5

The smooth particles increased, and the rough particles decreased with time
The fracture breakage may be the predominant breakage mechanism
For coarse particles, attrition was the main mechanism as residence time
increased

374
 Case Study - 5
Quantitative Morphological Statistical Analysis (by Clemex)
Morphological Feature
Initial (P1)
Agitator Speed (Skewness Value)
Mineral Type
(RPM) Residence
Sphericity Elongation Roughness
Time (P4/P5)
P1 -0.45 -0.77 -1.26
1000
Galena P5 -0.69 -0.78 -1.53
Concentrate P1 0.00 -0.61 -0.55
2000
P4 -0.43 -1.22 -1.06
P1 0.79 -0.13 -0.9
1000
P5 -0.18 -0.36 -1.49
Quartz
P1 -0.20 -0.61 -2.00
2000
P5 -0.61 -0.91 -3.06
P1 -0.14 -0.55 -1.33
Mixed Quartz & 1000
P5 -0.30 -0.60 -0.60
Galena
P1 -0.14 -0.50 -1.36
Concentrate 2000
P5 -0.29 -0.77 -1.58

Case Study - 5

Initial Setting of Particles in the 3 Sections (A, B, and C)

at Time Zero

375
 Case Study - 5
Maximum Normal and Tangential Forces
Maximum Normal Forces Maximum Tangential
Agitator
(N) Forces (N)
Speed (RPM)
A B C A B C
1000 2.0 1.6 0.8 0.3 0.2 0.1
1500 2.3 1.9 1.2 0.5 0.4 0.2
2000 2.7 2.3 1.7 0.4 0.3 0.2

Maximum Normal and Tangential Forces Distribution

Maximum Normal Forces Maximum Tangential
Agitator
(N) Forces (N)
Speed (RPM)
A B C A B C
1500 Media 4.4 3.5 4.7 1.1 0.9 1.1
1500 Galena 4.7 4.1 4.3 1.6 1.4 1.2
2000 Media 4.5 4.4 4.1 1.0 1.1 1.0
2000 Galena 5.4 3.7 5.0 1.8 1.1 1.5

Sizing and Scale-Up

General
Fine grinding machines are sized based on throughput (t/h) and energy
requirement (kWh/t)
Energy requirement is generally quantified based on testwork results must be
scrutinized
It is dependent on the following
- Specific ore
- Grinding machine
- Operating conditions (speed, slurry density, etc.)
- Media (type, size, density and quality)
- Feed and target product size

376
 Sizing and Scale-Up

Schematic diagram of experimental flow signature plot

study (Roufail, 2011)

Sizing and Scale-Up

The pendulum testing flow

circuit
Rahal et al., 2011

377
 Sizing and Scale-Up
IsaMill
IsaMill is sized using a Signature Plot Test
Continuous test with 15 kg of ore producing a specific energy graph
Conducted in 4 L IsaMill identically proportional to full-scale
Uses identical media to full-scale
Conducted under same operating conditions as full-scale (density, pressure,
media charge level, etc.)
Published operating vs. scale-up data shows the units scale-up well

Burford and Clark, 2007

Sizing and Scale-Up

IsaMill (continued)
Signature Plots Test: Scale-up of MRM M3000 IsaMill

Barns and Curry, 2006

Larson et al.

378
 Sizing and Scale-Up
IsaMill (continued)
Sizes presently available

Model Grinding Volume Power Max Flow rate

L kW m3/h
M100 100 75 12
M500 500 200 30
M1,000 1,000 355 or 500 90
M3,000 3,000 1120 ----
M5,000 5,000 1500 160
M10,000 10,000 3000 250
M50,000 50,000 8000 1000

Selection Criteria
General
When selecting a technology for fine grinding, the following should be
considered:
- Feed and product size required
- Capital cost
- Operating cost (media/power consumption and maintenance)
- Availability
- Operability
- Downstream process requirements (chemistry, density, etc.)
- Accuracy and reliability of sizing

379
 Conclusion
Highlights
Fine grinding is becoming more and more of a requirement in effectively
processing the fine-grained, low-grade deposits of today
There are several technologies available on the market today to accomplish
fine grinding, each having different attributes
Accurate determination of the energy requirement of a certain technology and
duty is key in its sizing
Always consider the total cost of ownership and the accuracy and reliability of
the sizing

Energy Efficiency in Mining

Bern Klein
N. Emre Altun
Andrew Bamber
Jeff Drozdiak
Stefan Nadolski
Persio Rosario
Chengtie Wang
AJ Gunson
Trent Weatherwax
Reem Roufail
Jennifer Parry
Libin Tong
Robert Hall
Malcolm Scoble
Mario Morin
Scott Dunbar

380
 Energy and Mining
Mining and non-ferrous metal smelting accounted for
7% of Canadas industrial energy consumption (2009)
6% of direct industrial GHG emissions of Canada (2009)
Mining is #2 energy consumer in British Columbia accounting for 5% of
BC Hydros power generation (primarily open pit mining)
Comminution is principal energy consumer (50-70%)
Material handling (haulage) # 2
Water/dewatering #3
Estimates for UG Mining 40% of energy goes to comminution, 40% to
ventillation
Source: A Review of Energy Consumption and Related Data: Canadian Mining and Metal Smelting and Refining Industries 1990 to 2009. Nyboeer, J., Rudd, S., March
2011, Canadian Industrial
Energy End-use Data and Analysis Centre, Simon Fraser University, Burnaby, BC, Canada

Motivation for Reducing Energy Usage

Energy usage represents a significant operating cost for mines

Cost of energy will increase in the future

Energy Conservation and GHG Reduction a priority in Canada

Canada GHG Emissions and Intensity Legislation 2013 (potential)
BC Clean Energy Act 2009 - BC Hydro is mandated to save 66% of
new load growth
TSM Initiative GHG and energy efficiency guidance document
(MAC)
CMIC Energy efficiency - The 40% Mine

381
 Motivations for Improving Energy Efficiency
Energy Intensity Indicators for Metal Mining

Source: A Review of Energy Consumption and Related Data: Canadian Mining and Metal Smelting and Refining Industries 1990 to 2009. Nyboeer, J., Rudd, S., March
2011, Canadian Industrial Energy End-use Data and Analysis Centre, Simon Fraser University, Burnaby, BC, Canada

Motivations for Improving Energy Efficiency

Energy Intensity Indicators for Non-Metal Mining

Source: A Review of Energy Consumption and Related Data: Canadian Mining and Metal Smelting and Refining Industries 1990 to 2009. Nyboeer, J., Rudd, S., March
2011, Canadian Industrial Energy End-use Data and Analysis Centre, Simon Fraser University, Burnaby, BC, Canada

382
 Motivations for Improving Energy Efficiency
Spot Crude Oil Price 1990-2011
120 150
Price Change

C h an g e in S p o t C ru d e Oil
100

S p o t C ru d e Oil P rice ($
100
80
U S /B arrel) 50

P rice (% )
60
0
40
-50
20

0 -100
1985 1990 1995 2000 2005 2010 2015
Year

Electrical energy ~ price of diesel incentive to electrify

Source: http://www.tradingeconomics.com/canada/inflation-cpi; STCA Statistics Canada

Total Energy and Potential Savings in Metal Mining

Blasting
50

Diesel Equipment
Specific Energy Requirement (kWhe/ton)

45

Drilling
40

Digging
35

Ventilation
30

Dewatering
25

Crushing
20

Grinding
15

Separation & Floatation

10

Ancillary Operations
5

0
Current Best Practice Practical Theoretical
Source: Industrial Technologies Program, USDOE, June 2007 Minimum Minimum

383
 Emerging and Enabling Technologies

Comminution (HPGR, stirred mills)

Application of sensors, pre-concentration & waste rejection
(sorting technologies)
Hydromet (Galvanox, Electrowinning)
Improved energy efficiency through optimized water usage
Energy recovery
Increasing trend of electrifying technologies

Comminution

384
Energy Efficient Comminution Technologies

High Pressure Grinding Roll (HPGR)s versus AG/SAG circuits

Stirred Mills versus Ball Mills

Novel circuits

HPGR - ISA Mill Circuit

AG HPGR circuit for high clay ores

HPGRs
Potential Benefits
Energy savings
Improved metallurgy (liberation)
Considered only for hard ores
Other Potential Applications
HPGR of pebble crusher product
High clay ores
Deposits with ores of variable hardness

385
 High Pressure Grinding Rolls (HPGR)
UBC-Koeppern HPGR

High Speed Stirred Mills

Potential Benefits
Energy savings
Selective Comminution
Considered primarilly for fine grinding
Other Potential Applications
Primary Grind

386
ISA Mill

Stirred Media Detritor

387
 Crusher Ball Mill vs HPGR Ball Mill
vs HPGR ISA Mill
(a)
(b)

(c)

J. Droizdiak MASc

Energy Comparison
Comparison of specific energy consumption for each circuit

Energy consumption in the HPGR / stirred mill circuit

Unit Operation Feed f80 Product p80 Specific Energy Consumption Specific Energy Consumption
(mm) (mm) with Dry Screening (kWh/t) with Wet Screening (kWh/t)
First Stage HPGR 21 7.68 1.54 1.54
Second Stage HPGR 7.68 0.35 2.91 3.58
Stirred Mill 0.34 0.075 9.73 9.73
TOTAL 14.18 14.85

388
 SABC Circuit versus HPGR Circuit

SABC Circuit versus HPGR Circuit

SABC Circuit Power HPGR Circuit Power
P80 = 160 um
Operation Power (kW) Operation Power (kW)
SAG Mill 7435 HPGR 3175
Crusher 149 Crusher 332
Ball Mill 8167 Ball Mill 8839
Material Handling 736 Material Handling 1090.4
Total 16487 13436.4
Energy Savings % 19

389
 HPGR ISA Mill Circuit

SABC vs HPGR vs HPGR-ISA Circuit

SAG Circuit HPGR Circuit HPGR-Stirred Mill
P80 = 75 um
Power Power Power
Operation (kW) Operation (kW) Operation (kW)
SAG Mill 7950 HPGR 3175 HPGR 7141
Crusher 87 Crusher 332 Crusher 332
Ball Mill 9079 Ball Mill 12133 Stirred Mill 4143
Material Material Material
Handling 762.4 Handling 1282.4 Handling 953.4
Total 17878.4 16922.4 12569.4

Energy
Savings % 5 30

390
 AG - HPGR Circuit
- Soft Ores Containing Clays

Crusher Feed Bin

Trommel
HPGR
Screen
Cone Crusher
Washing
Screen
Autogenous
Diverter
Mill/Scrubber

To Ball Mills
Coarse Ore

P. Rosario PhD Thesis

AG - HPGR Circuit versus SABC Circuit

AG - HPGR SABC Total
Feed Savings
Feed rate ( 1 line / 2 lines) 81,600 69,485 t/d
Availability 85% 94%
Fresh Feed / Total w. Rec Solids 4,000 3,080 t/h
F80 123 123 mm
Sub Specific Energy (Fresh/Total) 4.29 7.79 kWh/t 44.9%
Trommel&Screen 0
Aperture 12.7 15.9 mm
Total U/S - T80 4.880 5.361 mm
Bond WI 15.0 15.0 kWh/t
Cyclone O/F P80 200 1 m
HPGR Specific Energy
(Fresh/Total) 7.03 7.41 kWh/t
Sub Specific Energy (Fresh/Total) 11.32 15.21 kWh/t 25.5%

391
 Pre-concentration
and
Waste Rejection

Sensing and Sorting Technologies

Hand sorting - pre-Roman times
Automated sorting
Uranium radiometric sorting Ontario 1958
Diamonds X-Ray fluorescence W. Australia 1985
Recent large scale examples (est. 300 sorters
installations)
Nickel, Kambalda W. Australia
Platinum, Amplats, Rustenburg UG2 Section
Sensors - Surface versus Bulk Properties
Challenges Better sensors, higher throughput
machines

392
Courtesy C.
Bergman
Mintek, 2009

Sensor Technologies
Method Analysis Application

Photometric (reflection, Surface Coal, sulphides, phosphates,

brightness, grey level, oxides
RGB, IR, UV, texture)
Radiometric Bulk Uranium, gold
Conductivity, magnetic Bulk Metal sulphides, native metals,
susceptibility iron oxides
X-Ray Fluorescence Surface Diamonds, metal sulphides,
limestone, iron
X-Ray Transmission Bulk Coal, sulphides

393
Optical Sorting

CommoDas
Optical Image
MikroSort
Analyzer
Optical Sorter
at UBC

Conductivity Sorting

Conductivity Testing at UBC

PC

Sort Signal

A/D Converter:
Signal generation
and analysis

CommoDas
ROM Secondary EM
Sensing Coil 1 Sensing Coil 2 Sensing Coil 3 Conductivity Sorter

Amplifier
Bridge/
Power Supply Balancing Coil 1 Balancing Coil 2 Balancing Coil 3

394
Sudbury Operations - Energy Assessment

Sudbury Operations - Conductivity Sorting

Deposit Feed Grade (%) Conc. Mass (%) Conc. Grade (%) Recovery (%)
Ni Cu Mg Ni Cu Mg Ni Cu Mg
Craig 8112 1.16 0.47 5.54 72 1.50 0.57 5.16 93.49 87.40 67.46

Craig LGBX 2.10 0.35 2.57 83 2.43 0.37 2.39 95.85 86.70 77.07

Fraser Ni 0.81 0.36 4.21 80 0.94 0.40 3.73 92.73 89.43 70.67

Fraser Cu 0.83 11.42 1.81 41 1.65 20.92 0.68 81.12 74.89 15.42

TL Footwall 1.29 9.08 1.90 66 1.85 12.05 1.08 94.66 87.88 37.51

TL Zone 2 1.40 0.87 3.41 62 2.03 0.87 3.41 90.35 83.84 59.11

TL Zone 1 0.68 0.43 6.00 44 0.98 0.48 5.58 63.07 48.43 40.47

Montcalm East 1.66 0.56 4.61 75 2.06 0.63 4.17 93.60 85.48 68.22

Montcalm West 0.32 0.15 5.97 30 0.64 0.30 6.05 59.23 57.50 29.93

395
McCreedy East Mine U/G Sorting

McCreedy East Mine - U/G Sorting

396
 Sudbury Operations Sorting (Bamber Ph.D.)
Overall reduction in energy consumption 20%

10000.00

9000.00
Base

8000.00
Precon
7000.00

6000.00

Power (kW) 5000.00

4000.00

3000.00

2000.00

1000.00

0.00
Montcalm Thayer Fraser Fraser Craig Onaping Ni Rim S Ni Rim S
Lindsley Copper Nickel Depth F/W

Thayer Fraser
Operation Montcalm Lindsley Copper Fraser Nickel Craig Onaping Depth Ni Rim S
Hoisting $399,995 $1,319,625 $505,001 $684,364 $2,391,748 $1,891,163
Haul $786,583 $302,422 $884,600
Pre-con -$1,342,180 -$843,569 -$615,687 -$979,603 -$1,285,380 -$1,285,380 -$1,167,864
Grinding $560,607 $273,248 $236,058 $320,410 $476,930 $476,770 $418,730
Processing $1,397,813 $698,906 $436,817 $873,633 $1,310,450 $1,310,450 $1,135,723
Overall
Savings $1,402,823 $831,002 $1,376,812 $719,440 $1,186,364 $2,893,589 $3,162,352

Sorting Past Present - Future?

Proven Technology
Sorting machines exist
Metallurgy proven
Concepts for mine designs developed
Economics demonstrated
Challenges of implementation
Better sensors
Higher capacity sorters
Technology transfer - Risk averse industry

397
 Future - How can we make better use of sensors?
Sensors organic part of mining system
Apply to all aspects from exploration (geophysical,
borehole sensors) to mining and processing
Embedded sensors in material handling systems
(ore passes, scoops, shovels, bins, chutes,
conveyors)
Transmission, recording, analysis technologies
Wireless data transmission (WiFi)
Data available to GEMCOM, MineSight, process
control
Intelligent connected mines with active online telemetry

Innovative Use of Sensors

Core logging equipment

Boreholes
Blast hole drill rigs
Face shovel
Belts
Sorter

398
Sensor Based Systems in Surface Mining

Sensor-based technologies and U/G Mining

399
 Conventional vs Sensor Based Mining
Conventional mining:
- people-orientated, plan-based, subjective, time
consuming
Future mining:
Application of on-line telemetry from in-mine
sensors: Production scheduling, grade control,
plant process control settings:
- flexible
- objective
- real-time
- simultaneous

Conclusions
There is significant potential to reduce energy usage in mining
by applying several existing technologies

Relatively new, but proven. comminution technologies are more

energy efficient than conventional technologies, but industry is
slow to adopt these technologies.

The outcomes of sensor-based technologies and sorting are

significant in economic and environmental measures these
can be applied to making mining smarter.

Challenges to the application of these technologies relates

primarilly to aspects of technology transfer and mining culture
rather than technical challenges.

400
 NBK Research Centre

Acknowledgements
Natural Sciences and Engineering Research Council of Canada
Canada Foundation for Innovation
Teck Ltd
Inco
Falconbridge
Xstrata Technology
COREM
Vale
Xstrata
Knelson
Metso Minerals
BC Hydro
CSIRO Australia
Koeppern
CAMIRO
Placer Dome
Barrick
SGS

401
Statistical Experimental
Design

The problem of Experimental design is deciding what pattern of design points will
reveal aspects of the situation of interest (Box &Hunter 1978)

Outline
Introduction
Factorial Design
Fractional Factorial Design
Response Surface Designs
Central Composite Design

402
 Definitions
Experiment:testorseriesoftests
Experimentaldomain:theexperimentalareaordesignregion/domainis
definedbythevariationoftheexperimentalvariablesandtheircombinations
Factors:experimentalvariablesthatcanbechangedindependentlyofeach
otheralsocalledindependentvariables/parameters
Response:measuredresultoftheexperimentsorperformancevariableor
qualitymeasure
MainEffect:thechangeinresponseproducedbyachangeinthelevelofthe
factormeasuredbythedifferencebetweentheaverageresponseatthehigh
levelofthefactorandtheaverageresponseatthelowlevel

Introduction

Thevalidityoftheconclusionsthataredrawnfromanexperimentdependstoa
largeextentonhowtheexperimentwasconducted (Whenexperimentsare
performedrandomlytheresultwillalsoberandomLundstedt etal1998)
ExperimentalDesignaimsatmaximisinginformationgainedfromaminimum
numberofexperimentswithrespecttodefinedexperimentalvariablesandthe
responses.
ThefailureofOnevariableatatimeApproach
Anengineerisinterestedinfindingthevaluesoftemperatureandpressurethatmaximizeyieldina
chemicalprocess:

Iftheonevariableatatimeapproachisapplied
Byfixingthetemperatureat155F(currentoperatinglevel
Andvaryingthetimeatincrementallevelsof0.5from
0.5hrsto2.5.Theresultantvariationofyieldwith
timeshowstheoptimumtimetobe1.7hrs

Montgomery & Runger (2002)

403
 Introduction

Why statistical Design Experiments ?

However,thecontourplotofactualprocessyieldasafunctionoftemperatureandtimewiththe
onefactoratatimeexperimentssuperimposedonthecontoursshownbelowshowsthatthe
approachhasfailedtolocatetheoptimum

OneVariableaTime
failswherethereare
Interactioneffects

Montgomery & Runger (2002)

404
 Design Selection Guideline
Choice of an experimental design depends on the objectives
Screening Objectives: Removing less important factors, normally precedes
determination of response surface
Response Surface Objectives: Give an indication of the local shape of response
surface

Numberof Screening ResponseSurface

Factors Objective Objective

Centralcompositeor
24 FullorFractional BoxBehnken

Fractionalor Screenfirstto
5ormore Plackett Burman reducenumberoffactors

Factorial Experimental Design

Definition:Iswhenexperimentaltrials(orruns)areperformedatallcombinations
offactorlevels.
Forexample,
AMetallurgistisinterestedininvestigatingtheeffectsofBowlspeed(BS)andFluidisation(Fw)water
onGravityrecoveryofgoldusingaKnelsonConcentrator.IftwolevelsofBS(30Gand90G)andtwo
levelsofFw (5gpmand10gpm)areconsideredimportant,afactorialexperimentwouldconsist
ofmakingexperimentalrunsateachofthefourpossiblecombinationsoftheselevelsofBSandFw.
Usefulinscreeningstudies
Fullfactoriallimitedtodetermininglinearinfluenceofvariables
FractionalFactorial:Allowsforevaluationofinteractionsbetweenvariables

405
 Two Level 2k full Factorial Design

If the combination of k factors are investigated at 2 levels then the total

number of runs is 2k
Factor level are given by (-) for lower level and (+) for high level

Fractional Factorial Designs

Asthenumberofvariablestobescreenedincrease,thenumberofrequiredtests
increasesdramaticallyforFullfactorialdesignandmostoftheexperimentalruns
becomeredundant,FractionalFactorialdesigndealswiththisredundancy
Definitions:
HalfFraction:isa2k1i.e. 2 2 = 1 2 factorialdesign
1 k k

2
FractionalFactorialdesign:Consistof2kfractionofthetotalfull
factorialrunsfornvariablesat2levelsgivenby2nki.e. 2 2 = k n

1 n
2 factorialdesign
2k
Generator: is the defining relation of the design e.g. If settings for a 25-1 factorial design is constructed
such that the 5th variable settings are given by 5 =1234 such that 5x5=1234x5 Or 52=12345, the
generator can be written as I = 12345, where I the product of multiplying the elements of any column by a
column of identical elements.
Contrast: represented by lij is the linear function of the observations which estimate the ij interactions
and is the measure between two averages.

Resolution: represented by Roman numerals is the length of the shortest word in the defining relation for
example the 25-1 fraction is a resolution V design, it does not confound main effects and two factor
interactions with each other, But confound 2 -factor interactions with 3-factor interactions. Resolution R=III
does not confound main effects with one another But does confounds main effects with two factor
interactions. Whilst resolution R=IV does not confound main effects and 2-Factor interactions But does
confound 2-factor interactions with 2-factor interactions

406
 Construction of Fractional Factorial Design of Highest resolution

SeveralFractionalfactorialdesignsexist
Forthemostbasic2k1
Writeafullfactorialdesignforthefirstk1variables
Associatethekth variablewithplusorminusinteractionscolumn123...(k1).
WithFractionalfactorialdesignitisalwayspossibletoestimatetheeffects
Buttheywillbeconfounded(contaminatedbyhigherlevelinteractions)
Manyvariablescanbeinvestigatedwithoutanexcessivenumberof
experiments.
Lessinformationisgainedcomparedtofullfactorialdesigns,andtheprice
tobepaidforthefewexperimentsisthecontaminationofthemaineffects
bytheinteractioneffectsi.e.Themaineffectsareconfounded
Increaseindegreeoffractionationlowerstheresolutionofthebest
fractionandincreasesconfoundingbetweeneffectsofvariousorder

Example of Construction of Fractional Factorial Design

Model matrix X from factorial design is used to define the design matrix in fractional factorial designs
and the settings for the remaining variables are defined using the Columns in the matrix.
Example:
Sevenvariablescanbestudiedina274fractionalfactorial variablesx4tox7aredefinedby
design.Thedesignisdefinedbythemodelmatrix23 =2427 thecolumnsfortheinteractions
whichis1/16ofthefactorialdesign.Afullfactorialdesign betweenthevariablesa,bandc
wouldrequire128experiments.The8experimentsare
selectedtospanthelargestpossibleexperimentaldomain Columnsareorthogonalandthus
inthe7dimensionalspacespannedbythesevenvariables. possibletoestimatethemain
effectsindependentofeachother

407
 Case Study
TheKnelsonCVDisaheavymetalcontinuousgravityconcentratorwithprovencapabilitiestorecover
goldassociatedwithsulfides.MyraFalls(apolymetallicCuPbZincMineisloosing50%ofitsgoldto
thetailings.ItisrequiredtoassespossibleapplicationoftheCVDforgoldrecoveryfromplanttails.
Thegoldtotailsisassociatedwithpyritewhichisthemainironmineralfortheore.
TotestpotentialapplicationitisnecessarytotesttheCVDacrosstheexperimentaldomainand
determinetheparameterlevelsyieldingtheoptimummetallurgicalperformanceinbothgradeand
recovery.FeisusedasanindicatorforAu.

Firststepistoidentifyandscreenthefactors:
McLeavy(2005)identified8potentialfactorsthatinfluenceCVDperformance(Fluidisation,%solids,
FeedGrade,Heaviesparticlesize,Bowlspeed(BS),Pinchvalveopentime(PVO),Pinchvalveclosed
time(PVC),solidsfeedrate

2levelfullfactorialdesignwouldrequire256runs

Redundancyintermsofeitherorbothhigherlevelinteractionsandexcessvariablesstudiedask
increases

FractionalFactorialdesignexploitsthisredundancy

Variable level using synthetic ores for Factor Screening

McLeavy(2005)usedsixteenrun2IV84
Priortoscreeningexperimentsthelimitsoffactorsisdetermined,thetablebelow
showstypicalCVDfactorlevels

VariableLevel
Variable High Low Centrepoint
Heavies(%) 4 1 2.5
Fluidisation(gpm) 14 5 10
PVO(s) 0.05 0.03 0.04
PVC(s) 8 2 5
BS(RPM) 925 725 825
SolidsFeedrate(tph) 2 1 1.5
%Solids 45 30 37
HeaviesParticlesize(p80)microns 425 125 275

408
 Fractional Factorial design: A 28-4 Resolution IV design ,
CVD Results
Heavies Solids
% Feed
Fluidisation Particle BS PVO PVC Feed Grade
Solids Grade
size rate
run 1 2 3 4 5 6 7 8 Y(%)
1 + + + + + + + + 30.9
2 + + + + + 55
3 + + + + 18.1
4 + + + + 69.9
5 + + + + 26.4
6 + + + + 11.4
7 + + + 24.2
8 + + + + + 3.7
9 + + + 28
10 + + + + + 20.5
11 + + + + 69
12 + + + + 31.4
13 + + + + 3.5
14 + + + + 30.9
15 + + + + 12.9
16 27.1

Results for the CVD Fractional Factorial design

Calculated Contrasts for CVD Experiments showing main effects & two factor
interactions, 3 factor interactions and more are ignored
l1 = 1
L2 = 2
L3 = 3
L4 = 4
L5 = 5
L6 = 6
L7 = 7
L8 = 8
l12 = 12 + 37 + 48 + 56
l13 = 13 + 27 + 46 + 58
l14 = 14 + 28 + 36 + 57
l15 = 15 + 26 + 38 + 47
l16 = 16 + 25 + 34 + 78
l17 = 17 + 23 + 68 + 45
l18 = 18 + 24 + 35 + 67

Line9inthetableshowsthat12=37=48=56andarealiases ofeachother
andtheinteractionsareconfounded.

409
 Factor Screening Results

Rank Grade Effect(%) Rank Recovery Effect(%)
1 Solidsfeedrate 8.2 1 Heaviesparticelesize 21.9
2 BS 7.3 2 PVC 12.5
3 PVC 6.3 3 BS 12
4 %Solids 6.2 4 %Solids 5.9
5 Fluidisation 4.6 5 Fluidisation 4.8
6 FeedGrade 4.2 6 PVO 4
7 PVO 3.1 7 FeedGrade 3.2
8 Heaviesparticelesize 2 8 Solidsfeedrate 1.6

Theseresultsareusedtoscreenoutlesssignificantfactors
Ifonly3factorsaretobeusedformodelingandoptimisation:(Solidsfeerate,BS&PVC)will
beselectedbasedonGradeand(Heaviesparticlesize,PVCandBS)willbeselectedbasedon
Recovery.
IfbothRecoveryandGradearetobeusedasperformancemeasuresthentheexperimenter
wouldoptfor(PVC,BS&%Solids).

Response Surface Designs

Useful in fitting the second order models to the response with the use of a
minimum number of runs
Allows estimation of interactions and quadratic effects
Consist of:
3 level Factorial Design
Central composite:
- Circumscribed Central Composite Design
- Inscribed Central Composite Design
-Face centred Central Composite Design
Box Behnken
Doehlet Designs

410
 3 level Full factorial design
Given k factors, each at 3 levels a Full factorial design has 3k
runs.

When the number of factors is greater than 3, a full factorial design requires
a large number of runs and is not efficient
Best suited for screening out the few important main effects from the less
important

Comparison of Response Surface Designs

a b c

FullFactorialDesign Box- Behnken Design CircumscribeCentralCompositeDesign

Box Behnken design is economical in selecting points from three level factorial
arrangements, which allows the efficient estimation of coefficients for either first
or second order models
Central Composite design (CCD) is preferred because of its flexibility and
allowance for sequential experimentation i.e. Design can build upon factorial
design experiments

411
 Central Composite Designs

Circumscribed central composite design matrix:

CVD example
PVO PVC Bowl Speed Features
a
-1 -1 -1
-1 -1 1 23 Factorial design
-1 1 -1 component
-1 1 1
1 -1 -1
1 -1 1
1 1 -1
1 1 1
-1.6818 0 0 Factorial design plus
1.6818 0 0 Star design points
0 -1.6818 0 6, star design points allow for
0 1.6818 0 5 level to be assessed
0 0 -1.6818 Centre repeat runs for error
0 0 1.6818 analysis and to measure
0 0 0 significance of change in
0 0 0 Repeat centre runs response due to variation in
0 0 0 factor levels.
0 0 0

412
 CVD Results for Circumscribed Experimental Design

X1 X2 X3 PVO PVC BS Fe Grade Fe Recovery

-1 -1 -1 0.2 5 30 15.0 26.0
-1 -1 1 0.2 5 90 24.0 30.4
-1 1 -1 0.2 15 30 33.5 34.0
-1 1 1 0.2 15 90 25.7 30.0
1 -1 -1 0.6 5 30 24.2 19.4
1 -1 1 0.6 15 90 18.0 18.1
1 1 -1 0.6 5 30 15.3 24.0
1 1 1 0.6 15 90 13.0 23.0
-1.6818 0 0 0.1 10 45 26.0 19.4
1.6818 0 0 0.9 10 45 8.0 15.7
0 -1.6818 0 0.4 2 45 15.0 24.0
0 1.6818 0 0.4 25 45 33.5 25.9
0 0 -1.6818 0.4 10 20 25.7 23.0
0 0 1.6818 0.4 10 100 24.2 38.0
0 0 0 0.4 10 45 15.1 19.1
0 0 0 0.4 10 45 13.6 21.0
0 0 0 0.4 10 45 14.0 19.0

Repeatcentrerunsareusedtoassescurvatureandforerroranalysis

Empirical Modeling
Experimentalresultsareusedtomodeltherelationshipbetweenthemetallurgical
performancemeasures(Grade&Recovery)asafunctionofthedesignvariables.
TheResponseSurfacedesignsallowsforstrategicexplorationofthedesignspacesuch
thatarelationshipbetweenkeyvariablesandresponsecanbedefined
ResponseSurfacescanbegeneratedandthevariableslevelcombinationyieldingthe
minimum/maximumresponsegivestheoptimumsettings
Variousoptimisationstrategiesexistbutarebeyondthescopeofthischapter.

413