Summary
• Geology, mineralogy, sampling, mineral processing
– Used to practice as separate disciplines
• Henley, 1983
– Proposed their integration to draw more value into flowsheet diagnosis and development
• Modern practice developed
– 1st generation: method development, reactive support of operations
• Initial sampling model for plant surveys
• Initial sampling model for flotation testing
• Replicate flotation testing with QC
• Imaging and interpretation of polished thin sections
– 2nd generation: method development, predictive support of new flowsheet development
• Improved sampling models
• Improved sample presentation for ore mineralogy
• New liberation model for grinding strategy
• Case studies reviewed
Integration brings synergy
Desired Outcome
• Project Requirements of Flotation Testing
– Need clear, unambiguous “before” and “after” test conditions
– This requires:
• Representative Sample Material
• Comparable Flotation Test Charges
• Replicate Testing and QC
• Tight Metal Balances
• Reproducible Results
• Accurate Sizing of Survey Samples
• Multiple Polished Sections
• Mineralogical Skill and Experience
• Project/Operations Manager
– May not see the connection between the above and the reduction of risk
Clear proofs
These constitute parts of best
practice
The Milestone - Questions
• Interface Between Operating Plants and Automated Mineralogy
– What standard do we have to work at?
– What methods and procedures must we develop?
– What responsibilities do we face in subsequent project implementation?
• Capital cost
• Expectations of improved performance
– What type of decision do we want to make?
• Every decision has a consequence
– Some Decided that only Best Practice Would Suffice
• This paper describes best practice
Diagnose Existing Conventional Plants
1st Generation: Reactive
• Prototype Toolbox – Initial Sampling Models
– Basic Diagnostics
– Value Delivery
• Mt Isa, Queensland – Pease et al.
– Johnson et al.
• Lac des Iles, Ontario – Martin et al.
• Clarabelle, Ontario – Kerr et al.
• Raglan, Québec – Lotter et al.
• Candalaria, Chile – Baum et al., Kendrick et al.
Diagnose Existing Conventional Plants
Mt Isa: Johnson and Pease
• Zn operations experienced decrease in recovery
– Caused by decreasing sphalerite grain size
Zn Recoveries Fall Due to Finer Sphalerite Mineralogy
82%
67.5%
74%
Liberation Dropped by 15%
Recovery Dropped by 20%
Time
Mt Isa: Johnson and Pease
• Finer Grinding Stategies (IsaMill) restored recovery
– Caused by improved sphalerite liberation
Zn Recoveries Improved by Finer Grinding
82%
67.5%
74%
Finer Grinding Introduced
Zn Recovery
Improved to
80% from 52%
Lac des Iles, Ontario: Martin et al., 2003
• Operation Expanded from 2,400
to 15,000 tpd
• New Concentrator – Design Recovery was 82% Pd
– Commissioned Plant Recovery only 67.5% Pd
• Several Diagnostic Surveys – Discovered Bimodal Distribution of PGM Grain Size
• Modes at 20 and 5 microns
– Most Tailings Losses Found in the Finer Subdistribution
– Reduce Talc Depressant Dose in Rougher to Stabilise
Froth and Increase Rougher Mass Pull
– Regrinding of all Rougher Concentrates
– Additional Cleaner Circuit
• Implemented Changes – Recovery Improved to 74% Pd
– Grade Improved to 240 g/t Pd from 174 g/t
Pd Grades and Recoveries Improved
82%
67.5%
74%
60
65
70
75
80
85
Design Comm Improved
Pd
Reco
very
%
150
170
190
210
230
250
Design Comm Improved
Gra
de g
/t P
d
170 174
240
82
67.5
74
Pd Recovery
Pd Grade
Clarabelle, Ontario: Kerr et al., 2003
• Circuit Expansion 1991
• Key Part of Circuit was Mag Sep on Float Feed
– Monoclinic Pyrrhotite
• New Orebody in North Mine Commissioned1998
– Only Makes Up 5% of Total Mill Feed
– But this Pyrrhotite is Hexagonal (found by mineralogy after the fact…)
– Damaged Flotation Selectivity
– Ni Recovery Fell from 78% to 70%
• Testwork performed
– Reallocate the regrind mill
– Add TETA/sulphite to depress Hexagonal Pyrrhotite
• Changes Implemented
– Recovery was restored
Loss in Ni Recovery Restored
82%
67.5%
74%
Raglan, Québec: Lotter et al., 2002
• Concentrator Commissioned
Jan 1998 – Conventional Design
– Excellent Agreement Between Design and
Commissioned Performance
• Ni Recovery Design 87.0% Actual 86.7%
• Ni Grade Design 16.0% Actual 16.0%
• What Opportunities? – Survey June 1998- QEMSCAN – Float Tests
• Reroute Recleaner Tailings to Cleaner Circuit Feed
• Regrind Cleaner Tailings Before Scavenger Flotation
• Add CMC Depressant to Rougher
– Implemented Changes
• Recovery Gains:
– 2.1% Ni, 1.5% Cu, 1.9% Pd, 4.1% Pt
• Grade Gains:
– Concentrate from 16 to 18% Ni
Ni Recovery Advanced Beyond Design
82%
74%
85.5
86
86.5
87
87.5
88
88.5
89
Design Comm Improved
Ni R
eco
very
, %
87 86.7
88.8
15
15.5
16
16.5
17
17.5
18
18.5
Design Comm Improved
Ni G
rad
e,
%
16 16
18
Modern Toolbox
• Many Contributors
– Advance the QEMSCAN and MLA
• Gottlieb, Gu
– Better Sampling and Sizing
• Restarick, Hartley et al., Lotter
– Understanding Grinding and Grinding Media
• Greet, Peng, Grano, McIvor, Finch
– Understanding Flotation Synergy/Mixed Collectors
• Bradshaw, Lotter
– Advancing Modelling
• Wightman, Evans
More Powerful
2nd Generation: Predictive
• Advanced Toolbox – Improved Sampling Models
• Stratified Sampling of Drill Core – Lotter
• Distribution Modelling for Sampling Drill Core – Oliveira
• Statistical Benchmark Surveying - Lotter
– Ore Characterisation at the Geomet Unit Level
– Improved Liberation Models – Wightman, Evans
• Advancement of QEMSCAN to FEG Platform
• Addition of Advanced Microprobe CAMECA SX100
– Advancement of Mixed Collector Models – Bradshaw, Lotter
– Statistical Models for Flotation Testing – Napier-Munn
– Small Scale Flotation Testing - Bradshaw
• Prominent Hill, Australia – Barns et al.
• Kamoa, DRC – Lotter et al.
Contribute to New Designs
Prominent Hill
• Drill Core Mineralogy - MLA
– Chalcocite, bornite, chalcopyrite at sizes 24-40 µm
– 70% associated with haematite
– 15% associated with pyrite
– Fluorite present…penalty element
• Flowsheet Development
– Primary grind of 106 µm selected from tests
– Xanthate-based collector suite
– Lime added to control pyrite
– Regrind rougher concentrate to 25 µm
Contribute to New Designs – Barns et l., 2009
80
81
82
83
84
85
86
87
88
89
90
Design Comm
Cu
Re
cove
ry
% Month 4 87.0
85.0
Kamoa Project, DRC
• Sampling Equations
Contribute to New Designs
Hypogene
Supergene
fv
Ms95.424
fv
Ms10.218
Ms = 66.4 kg; Project required 265 kg
Ms = 34.08 kg; Project required 193 kg
Kamoa Project, DRC
• Safety Line
Contribute to New Designs
0.001
0.01
0.1
1
10
100
1000
0.1 1 10 100
Mass,
Kg
Topsize, mm
E
A B
C
D
A: 250 kg HQ quarter
core B: 250 kg crushed
sample
C: 125 x 2 kg replicate
test charges
D: 2 kg milled ore at
float feed size
E: Chemist‟s analytical
sample
UNSAFE SIDE
SAFE SIDE
Kamoa Project, DRC
• Mineralogy Identified and Quantified
a Wide Range of Copper Sulphides
- Complex Electrochemistry
– Fine to Ultrafine CuS Grain Sizes
• How to Float all of these in
1 Flowsheet?
Contribute to New Designs
0
10
20
30
40
50
60
70
80
90
100
Hypo Super
Dis
trib
ution %
Other
Azurite
Covellite
Chalcocite
Bornite
Chalcopyrite
Kamoa Project, DRC
• Mineralogy Quantified
Liberation of Copper Sulphides
in Rougher Float Feed
- Incomplete Liberation at
80% -75 microns
• How to Float all of these as
liberated and middling particles?
Contribute to New Designs
0
10
20
30
40
50
60
70
80
90
100
Lib
Midds
Locked
Hypo Super
Dis
trib
ution %
62.5
16.6
20.9
53.1
22.2
24.7
Rougher Float Feed at 80% - 75
microns
Kamoa Project, DRC
• Process Implications
– Use an MF2 Rougher-Scavenger Circuit
– Formulate a Mixed Collector to Recover the Range of Copper Sulphides at Incomplete Liberation
Through a Mixed Potential
– Regrind the Rougher and Scavenger Concentrates before Cleaner Flotation
– Use Two Separate Cleaner Circuits (Fast and Slow)
Contribute to New Designs
Kamoa Project, DRC
• Metal Balance
Contribute to New Designs
Cu Accountability is 99.7%
Or -0.3% error
High Confidence Flotation
Testing
95% Confidence Level
Assay Head 3.30% Cu Built Up Head 3.29% Cu
0
5
10
15
20
25
30
35
40
-4 -3 -2 -1 0 1 2 3 4
Fre
qu
en
cy %
Percent Error Bin
• HCFT uses replicate tests
and averages
• Distribution of Errors
• 80% of observations fall
between -2 and +2%
Distribution of residual errors
assumes an almost Normal
Distribution
Kamoa Project, DRC
Contribute to New Designs
0
5
10
15
20
25
30
35
40
45
20 30 40 50 60 70 80 90 100
Prototype
Breakthrough
Milestone
31.95 65.58 28.17 83.43 32.78 85.37
Grade% Rec%
Cum. Recovery % Cu
Cu
m. G
rad
e %
Cu
Each flowsheet revision advances the grade-recovery curve
Kamoa Project, DRC
Contribute to New Designs
• Hypogene
– 32.78% Cu Grade
– 85.3% Cu Recovery
– Mass Pull 8.6%
• Dec 2011 Supergene
– 45.11% Cu Grade
– 83.20% Cu Recovery
– Mass Pull 6.9%
Milestone Flowsheet Delivered in 10 Months
0
10
20
30
40
50
60
70
80
20 30 40 50 60 70 80 90 100
Hypogene
Supergene
Kamoa Project, DRC
Contribute to New Designs
Further Grinding Requirements Modelled from Scavenger Tailings
0
10
20
30
40
50
60
0 10 20 30 40 50 60 70
Gra
de o
f C
u S
ulp
hid
es
Co
nta
inin
g
Pa
rti
cle
s, V
ol
%
Mid-Size, Micrometers
L0
Initial regrind size of 20 um indicated for further liberation of locked copper sulphides
Conclusions
• Conventionally-designed concentrators offer an easy platform from which to deliver performance
improvements
• Greenfield projects can now draw on the modern practice with improved capabilities to deliver better
performance at startup
• The High Confidence Flotation Testing System has been proven to scale up reliably
• The JKSMI microflotation system has accurately scaled up to an operation
• Variability testing at several levels, including mineralogy, is key to reducing project risk