INTEGRATED MINING, PRE-CONCENTRATION AND WASTE DISPOSAL SYSTEMS FOR THE INCREASED SUSTAINABILITY OF HARD ROCK METAL MINING By ANDREW SHERLIKER BAMBER, B.Sc. (Mechanical Engineering) University of Cape Town, 1993 Pr. Eng. (Engineering Council of South Africa), 1999 M.A.Sc. (Mining & Mineral Processing) University of British Columbia, 2004 A THESIS SUBMITTED IN FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy in THE FACULTY OF GRADUATE STUDIES (Mining Engineering) UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2008 © Andrew Sherliker Bamber 2008 ii Abstract In the hard rock metal mining industry, both in Canada and globally, a decreasing number of economic mineral deposits are found at shallow to medium depth, and most of the deposits that remain are close to sub-economic and are required to be mined at high tonnages in order to show a return. The majority of remaining deposits are presented in challenging geological or geotechnical settings making the deposit sub-economic. The integration of ore pre-concentration and waste disposal functions into the hard rock metal mining system is proposed as a novel interpretation of Mine-Mill Integration for improving the economics and environmental impact of exploiting such deposits. The proposed approach seeks to reject waste as early as possible in the mining cycle, and safely dispose of it as backfill. This is proposed as a ‘Lean’ alternative to improving the economics of mining simply by increasing the throughput. ‘Lean’ philosophy seeks to design out overburden, smooth production, and eliminate waste from the manufacturing system. It is suggested that the proposed approach addresses all three areas, and is thus an important strategy to be considered for mining companies wishing to simultaneously improve their efficiency, economics and environmental performance, thus increasing their sustainability. Technologies specific to the success of the approach including pre-concentration systems, composite fill systems, and continuous mechanized mining methods are discussed. The impacts and benefits of integrating these technologies are defined and quantified through research, testwork, systems design and analysis. Custom geo-metallurgical evaluation tools incorporating mineralogical, metallurgical and geotechnical methods have been developed to assess ores in terms of their potential for the adoption of the proposed approach. A computerized parametric evaluation model has been developed to quantify the potential impacts and benefits using data from these evaluations. Data from over 26 case studies combined with the literature indicates that the opportunity for ore pre-concentration appears to be a general case in hard rock ores. A wide range of impacts and benefits arising from this potential have been identified and quantified through the research, indicating positive overall outcomes for the majority of cases studied. iii Table of Contents Abstract…………………………………………………………………………… ii Table of Contents……………………………...………………………………….. iii List of Tables……………………………………………………………………... viii List of Figures…………………………………………………………………….. x List of Symbols……………………………………………………………............ xiv Glossary…………………………………………………………………………... xvi Acknowledgements………………………………………………………………. xx Chapter 1 – Introduction and Thesis Outline………………………............ 1 1.1 Introduction………………………………………………………………….. 1 1.2 Significance of the Research……………………………………...….............. 7 1.3 Research Contribution……………………………………………………….. 8 1.4 Thesis Outline………………………………………………………………... 8 Chapter 2 – The Application of Integrated Mining, Pre-concentration and Waste Disposal Systems in the Hard Rock Metal Mining Industry………………………………………………………………………….. 10 2.1 Introduction………………………………………………………………….. 10 2.2 The Application of Lean Manufacturing Concepts to the Hard Rock Metal Mining System……………………………………………………………….. 13 2.3 Ore Pre-concentration as an Example of Lean Manufacturing in Mining……………………………………………………………………….. 14 2.4 ‘Lean’ Case Study – Xstrata Nickel Ontario Operations…………………….. 21 2.5 Conclusions…………………………………………………………………... 25 Chapter 3 – Enabling Technologies………………………………………….. 27 3.1 Introduction…………………………………………………………………… 27 3.2 Process Technologies…………………………………………………………. 29 3.2.1 Ore Pre-concentration by Comminution and Size Classification…………... 32 22.214.171.124. Development of the Concept…………………………………………… 32 126.96.36.199. Testwork at UBC………………………………………………………. 34 3.2.2 Pre-concentration by Sorting……………………………………………….. 37 iv 188.8.131.52 Sorting Practices in the Minerals Industry……………………………… 37 184.108.40.206 Radiometric Methods…………………………………………………… 41 220.127.116.11 X-Ray and Laser Methods………………………………………............ 42 18.104.22.168 Optical Sorting………………………………………………………….. 42 22.214.171.124 Conductivity Sorting……………………………………………………. 43 3.2.3 Pre-concentration by Dense Media Separation……………………………... 45 3.2.4 Coarse Particle Flotation Techniques for the Pre-concentration of Base Metal Sulphide Ores………………………………………………………… 48 126.96.36.199 Development of the Concept……………………………………………. 48 188.8.131.52 Theoretical Basis………………………………………………………... 50 184.108.40.206 Scoping Testwork Results………………………………………………. 53 3.3 Waste Disposal Technologies………………………………………………… 56 3.3.1 Background………………………………………………………………... 56 3.3.2 Development of A ‘Rocky’ Paste Fill For Use With Underground Pre- Concentration Systems…………………………………………………….. 59 3.3.3 Fill Preparation and Delivery Systems…………………………………….. 62 3.4 Interfacing the Technologies with the Mining Activity………………………. 65 3.4.1 Cut-and-Fill Mining……………………………………………………….. 69 3.4.2 Open Pit Methods…………………………………………………………. 71 3.4.3 Block Caving………………………………………………………………. 72 3.4.4 Open Stoping Methods……………………………………………………. 75 3.4.5 Room and Pillar……………………………………………………............ 76 3.5 Conclusions……………………………………………………………............ 77 Chapter 4 –Experimental Methods for the Geo-metallurgical Evaluation of Ores for Pre-concentration and Waste Disposal ………… 78 4.1 Introduction……………………………………………………………….. 78 4.2 Mesotextural Evaluation, Fragmentation and Liberation Analysis……….. 78 4.3 Separability Evaluation……………………………………………………. 84 4.3.1 Dense Media Separation Testwork…………………………………… 84 4.3.2 Optical Characterization……………………………………………… 85 4.3.3 Conductivity Evaluation……………………………………………… 89 220.127.116.11 Testing of the INCO 'B2' Sensor……………………………….. 90 v 18.104.22.168 Development of the MineSense B2 ‘MkII’ Sensor…………….. 93 4.4 Grinding Work Index Testing……………………………………………… 97 4.5 Additional Geotechnical, Geo-metallurgical and Rheological Testwork….. 97 4.5.1 Geotechnical Characterization of Rejects…………………………... 97 4.5.2 Mix Design and Testing…………………………………………….. 99 4.5.3 Rheological Tests……………………………………………............ 100 4.6 Conclusions…………………………………………………………............ 101 Chapter 5 – A Parametric Model for the Economic Evaluation of Pre- concentration as a Lean Manufacturing Technique in Hard Rock Metal Mining……………………………………………………….................................. 102 5.1 Introduction…………………………………………………………………... 102 5.2 Evaluation Methodology……………………………………………………... 103 5.3 Model Assumptions………………………………………………………….. 107 5.4 Cost and Revenue Impacts……………………………………………............ 108 5.4.1 Operating Cost Impacts……………………………………….. 108 5.4.2 Estimation of Capital Costs for Hard Rock Mines…………… 111 22.214.171.124 Estimation of Open Pit Mining Costs…………………………............ 111 126.96.36.199 Estimation of Underground Mining Costs……………………………. 112 188.8.131.52 Estimation of Plant & Surface Infrastructure Capital Costs………….. 114 184.108.40.206 Adjustments for Variations in Mine Capacity and Effect of Inflation……………………………………………………….………. 116 5.4.3 Revenue Impacts………………………………………………. 117 220.127.116.11 Impacts on Recovery…………………………………………………... 117 18.104.22.168 Variations in Metal Price………………………………………............ 120 5.5 Impact of Cost Variations on Cutoff Grade, Resources and Reserves………. 122 5.5.1 Estimating the Ore Reserve…………………………………… 122 5.5.2 Cost – Cutoff Grade Interactions……………………………... 126 5.5.3 Cutoff Grade – Tonnage Interactions for an Idealised Resource………………………………………………………. 128 5.6 Impact Valuation and Evaluation Methodology……………………………... 131 5.7 Conclusion……………………………………………………………............ 136 vi Chapter 6 – Integrated Mining, Processing and Waste Disposal Case Studies…………………………………………………………………………… 137 6.1 Introduction…………………………………………………………………… 137 6.2 Gold Ores……………………………………………………………………... 138 6.3 Lead-Zinc Ores of the Mississippi Valley Type………………………............ 139 6.4 Pre-concentration of Copper Porphyry Ores………………………………….. 142 6.5 Footwall and Contact Type Ores of INCO’s Sudbury Operations……............ 146 6.6 Polymetallic Base Metal Sulphide ores of Xstrata Nickel’s Ontario Operations…………………………………………………………………….. 147 6.6.1 Preliminary Core Evaluations for 2 Mines……………………………….. 148 6.6.2 Phase II Study for 9 Ores of Xstrata Nickel’s Ontario Operations……….. 153 22.214.171.124 Metallurgical Results…………………………………………………. 155 126.96.36.199 Investigation of Waste Disposal Aspects…………………………….. 159 188.8.131.52 Rheological Evaluations of Fill Mixes……………………………….. 165 184.108.40.206 Grinding Index Testwork…………………………………………….. 165 6.6.3 Evaluation of Impacts on Energy Usage at Xstrata Nickel’s Ontario Operations…………………………………………………………........... 167 220.127.116.11 Impact of the Pre-concentration Step…………………………………. 170 18.104.22.168 Impacts on Hoisting Energy…………………………………………... 170 22.214.171.124 Impacts on Surface Haulage………………………………………….. 171 126.96.36.199 Impacts on Grinding and Overall Beneficiation of the Ore…………... 172 188.8.131.52 Impacts on Overall Energy Usage……………………………………. 174 6.6.4 Conclusions……………………………………………………………….. 175 6.7 Low Grade Ultramafic Ni Ores of the Thompson Nickel Belt……………….. 176 6.7.1 Introduction……………………………………………………………….. 176 6.7.2 Methodology, Sampling and Testwork Results…………………………... 178 6.7.3 Design of a Surface Pre-concentration Facility for CVRD INCO’s Pipe II Deposit……………………………………………………………………. 182 6.7.4 Parametric Evaluation of the Exploitation of Pipe II with and without Pre-concentration…………………………………………………………. 186 6.8. Conclusions…………………………………………………………………... 188 vii Chapter 7 – Discussion and Conclusions……………………………………. 190 7.1 Grounds for the Generalized Application of Integrated Mining, Processing and Waste Disposal Systems to Hard Rock Mining………………………..... 190 7.2 Integration of the Enabling Technologies……………………………………. 194 7.3 Benefits of the Approach…………………………………………………….. 195 7.4 Significance of the Research…………………………………………………. 197 7.5 Conclusions…………………………………………………………………... 198 References………………………………………………………………………. 200 Appendices………………………………………………………………............ 212 Appendix A – Publications List…………………………………………………. 212 Appendix B – Parametric Valuation and Evaluation Model Spreadsheets........... 213 Appendix C – Variations to the Idealized Model……………………………….. 214 Appendix D – Case Studies: Placer Dome………………………………………. 222 Appendix E – Case Studies: CVRD-INCO……………………………………… 246 Appendix F – Case Studies: Integrated Underground Mining and Processing at Cameco Corporation……………………………………………... 264 Appendix G – Table of Densimetric, Magnetic and Conductive Properties for Selected Ores………………………………………….................. 273 Appendix H – Optical Results for Xstrata Nickel Ores…………………………. 274 Appendix I – Conductivity Results for Xstrata Nickel Ores…………............... 310 viii List of Tables Table 2.1 – Planned Dilution Levels by Mining Method………………................ 10 Table 2.2 – Functional Comparison of Manufacturing vs. Mining Systems……... 17 Table 3.1 – Metallurgical Performance of Selected Pre-concentration Technologies on Various Ore Types………………………………………….................. 30 Table 3.2 – Methods of Discrimination in Sorting……………………………….. 38 Table 3.3 – Typical Sorter Capacities……………………………………............. 40 Table 3.4 – Conductive Properties of Ore Minerals……………………………… 44 Table 3.5 – Comparative Energy Density for Flotation Cells……………............. 51 Table 3.6 – Mt Isa Mines Composite Fill Mix Properties………………………... 62 Table 3.7 – Data for Existing Large and Deep Excavations……………………... 68 Table 3.8 – Advantages and Disadvantages of Block Caving Techniques............. 73 Table 4.1 – Mesotextural and Microtextural Classifications for Visual Evaluation of Ore Types……………………………………………… 79 Table 4.2 – Typical Heavy Media Separation Results…………………………… 84 Table 4.3 – Correlation of Photometric Measurements with Grade for Xstrata Ni Ores…………………………………………………………………… 89 Table 5.1 – Cost Estimating Factors for 500 tpd Underground Mine……............. 113 Table 5.2 – Capacity-based Estimating Factors for Process Plant and Infrastructure…………………………………………………………. 114 Table 5.3 – Process Plant Cost Factors……………………………………........... 114 Table 5.4 – Previous Estimates of Mine & Concentrator Capital……………….. 115 Table 5.5 – Cost Estimation Factors for Combined 3000 tpd Pre-concentration, Grinding and Flotation Plant………………………………………… 115 Table 5.6 – Ore Tonnage Estimation Factors…………………………………….. 123 Table 5.7 – Parameters for a Hypothetical 30 000m3 Gold Ore Reserve………... 125 Table 5.8 – Net Operating Cashflow Breakdown………………………………… 133 ix Table 6.1 – Results of Doe Run Ore Evaluation…………………………............. 142 Table 6.2 – Bougainville Ore Pre-concentration by Size Classification…............. 145 Table 6.3 – Summary of DMS Testwork Results on Xstrata Nickel Ores……….. 157 Table 6.4 – Summary of Conductivity Sorting Results on Xstrata Nickel Ores…. 158 Table 6.5 – Geotechnical Properties of Pre-concentration Rejects………............. 161 Table 6.6 – Fill Mix Ratios for Xstrata Fill Testwork……………………………. 162 Table 6.7 – UCS Results for Xstrata Fill Mixes………………………………….. 164 Table 6.8 – Average of Slump Test Results for Xstrata Fill Mixes……………… 165 Table 6.9 – P80, F80 and Work Indices Measured for Xstrata Ores and their Concentrates………………………………………………….............. 166 Table 6.10 – Estimated Annual Energy Usage & Costs for Pre-concentration Plant…………………………………………………………………... 170 Table 6.11 – Annual Hoisting Energy Cost Savings………………………………. 171 Table 6.12 – Overall Projected Impact of Pre-concentration on Energy Costs for Xstrata Nickel’s Ontario Operations………………………………….. 174 Table 6.13 – Preliminary Sorting Results for Pipe II……………………………… 180 Table 6.14 –Sorted Products for use in Comparative Flotation Testing…............... 180 Table 6.15 – Pipe II – Sorting Results by Ore Type………………………………. 182 Table 6.16 – Projected Metallurgical Performance of the Pipe II Sorting Plant…………………………………………………………............... 183 Table 6.17 – Present INCO Thompson Mill Operating Parameters………………. 186 Table 6.18 – Pipe II – Base Case Model Capital Cost Estimate………….............. 187 x List of Figures Figure 2.1 – Typical Composition of a ROM Ore Sample……………………….... 11 Figure 2.2 – Variance in Iron Grade and Tonnage Shipped ex Pilbara, WA………. 15 Figure 2.3 – Generic Manufacturing Value Chain……………………………….... 17 Figure 2.4 – Modified Mining Value Chain with Re-cycle………………………… 18 Figure 2.5 – Mining Value Chain with Surface Pre-concentration………………… 19 Figure 2.8 – Mining Value Chain with Underground Pre-concentration………….. 20 Figure 2.7 – Geographical Disposition of Xstrata’s Ontario Operations………….. 22 Figure 3.1 – Integrated Mining, Processing and Waste Disposal Approach……….. 28 Figure 3.2 – Integrated Underground Mining Processing and Waste Disposal Approach………………………………………………………………. 29 Figure 3.3 – Size Assay McCreedy 153 Orebody………………………………….. 32 Figure 3.4 – Pt Enrichment of Fines in Bushveld Ores…………………………….. 33 Figure 3.5 – Concentration Results from Autogenous Grinding and Classification of the 153 Ore…………………………………....................................... 35 Figure 3.6 – Grade Distribution by Size in 153 Ore after 60 minutes of Grinding………………………………………………………................ 36 Figure 3.7 – Concentration by Comminution and Size Classification……………... 37 Figure 3.8 – Features of Sensor-Based Sorters……………………………............... 39 Figure 3.9 – 100tph DMS Cyclone Module………………………………………... 46 Figure 3.10 – Hydrodynamic Zones in Flotation Tank Cell………………................. 52 Figure 3.11 – Comparative Size Distribution of Flotation Concentrates……………. 56 Figure 3.12 – Comparison of Backfill Strength by Aggregate Type………………… 61 Figure 3.13 – Rocky Paste Fill System at Mount Isa Mines………………................ 63 Figure 3.14 – Composite Fill Preparation System…………………………………… 64 Figure 3.15 – Thomas Katts BS907A Concrete Pump……………………................. 65 Figure 3.16 – 3D Layout of Excavations for DMS or Sorting Based Pre- concentration System…………………………………………………. 67 Figure 3.17 – Integration of Pre-concentration and Waste Disposal with Cut and Fill Mining……………………………………………………………. 70 xi Figure 3.18 – Process Integration for Waste Management in Open Pits…………….. 72 Figure 3.19 – Representation of Block Cave Stope………………………………….. 73 Figure 3.20 – Integrating Pre-concentration with Block Cave Operations………….. 74 Figure 3.21 – Typical Open Stoping Method………………………………………... 75 Figure 3.22 – Adaptation of Open Stoping Methods to Modified AVOCA with Composite Fill………………………………………………………… 76 Figure 4.1 – Observation of Massive Vein Massive Sulphide Mesotexture In Situ……………………………………………………………………. 80 Figure 4.2 – Mesotextural Evaluation of Musselwhite Core……………………….. 80 Figure 4.3 – Screened Sample Fractions for Size Assay…………………………… 81 Figure 4.4 – Footwall Ore Size Assay Showing α log-normal Grade Distribution…………………………………………………………… 82 Figure 4.5 – Kuz-Ram Model Results for Pipe II ROM Ore………………………. 83 Figure 4.6 – UBC DMS Pilot Testing Apparatus………………………….............. 85 Figure 4.7 – Footwall Ore Sink & Floats Fractions by Size……………….............. 85 Figure 4.8 – Pattern Matching Applications in Electronic Components…………… 87 Figure 4.9 – UBC National Instruments Machine Vision Station…………………. 88 Figure 4.10 – ‘Pancake’ Type Inductive Coil……………………………………….. 91 Figure 4.11 – INCO ‘B2’ Sensor Setup……………………………………………… 91 Figure 4.12 – Correlation of B2 Readings with Ni Grade…………………………… 92 Figure 4.13 – Plot of Field Strength with Coil Diameter……………………………. 94 Figure 4.14 – MineSense B2 ‘MkII’ Conductivity Sensor Schematic……................. 95 Figure 4.15 – MineSense B2 ‘MkII’ Conductivity Sensor Setup……………………. 95 Figure 4.16 – Correlation of B2 ‘MkII’ Conductivity Readings with Grade for Xstrata Ni Ores…………………………………………....................... 96 Figure 4.17 – MTS 815 UCS Testing Machine……………………………………… 99 Figure 4.19 – Slump Test Photographs for Fraser Copper Fill Mixes………………. 101 Figure 5.1 – Mineral Deposit Evaluation Flowchart………………………................ 105 Figure 5.2 – Synergistic Impact of Pre-concentration on Metal Recovery………….. 118 Figure 5.3 – Variation in Mill Recovery with Varying Feed Grade…………………. 119 xii Figure 5.4 – Characteristic Mineral Commodity Cycle……………………………... 121 Figure 5.5 – Break-even Production Rate…………………………………………… 127 Figure 5.6 – Change in Breakeven Production Rate………………………………… 127 Figure 5.7 – Impact of Change in Fixed and Variable Cost………………................ 128 Figure 5.8 – Idealized Grade-Tonnage Curve……………………………………….. 128 Figure 5.9 – Real Grade-Tonnage Curve for the Navidad Deposit, Arizona………... 129 Figure 5.10 – Grade Tonnage Curves for Selected Global Gold and Base Metal Deposits………………………………………………........................... 130 Figure 5.11 – Cost Impacts on Grade Tonnage Curve………………………………. 131 Figure 5.12 – Parametric Valuation and Evaluation Flowchart……………………... 132 Figure 5.13 – Examination of the Robustness of a Project using the NPV Profile Method………………………………………………………………... 135 Figure 6.1 – Banded Ore of the MVT Type………………………………………… 140 Figure 6.2 – Breccia Ore of the MVT Type………………………………………… 141 Figure 6.3 – Close-up of Breccia Ore Showing Massive and Disseminated Mineralization………………………………………………………….. 141 Figure 6.4 – Massive to Disseminated Chalcopyrite Mineralization along Fractures and Foliations in Biotite Porphyry……………………………………… 143 Figure 6.5 – Panguna Andesite and Biotite Textures from the Bougainville Mine.... 144 Figure 6.6 – Size/assay Relationships in Bougainville Copper Porphyry Ores…….. 145 Figure 6.7 – Fraser Cu Massive Vein Sulphides In Situ……………………………. 148 Figure 6.8 – T-L Cut and Fill Panel Showing Banded Pentlandites………............... 149 Figure 6.9 – Size Classification and Optical Sorting for Narrow Vein Footwall Copper Ores……………………………………………………………. 151 Figure 6.10 – Size Classification and Dense Media Separation for Massive Sulphide Ni Ores………………………………………………………………... 152 Figure 6.11 – Xstrata Nickel’s Ontario Operations………………………………….. 154 Figure 6.12 – Sudbury Breccia/Matrix Type Ore…………………………................. 154 Figure 6.13 – Disseminated Ore Texture…………………………………………….. 155 Figure 6.14 – Typical Dense Media Separation Rejects…………………………….. 160 Figure 6.15 – Size Distribution of Pre-concentration Rejects……………………….. 161 xiii Figure 6.16 – UCS vs. Axial Strain Curve for Rockfills…………………………….. 163 Figure 6.17 – UCS vs. Axial Strain for Composite Fills…………………………….. 164 Figure 6.18 – Work Index : Ore Grade Correlation…………………………………. 167 Figure 6.19 – Xstrata System Evaluation Parameters……………………………….. 169 Figure 6.20 – Hoisting Impact Evaluation Model…………………………………… 171 Figure 6.21 – Haul Energy Cost Savings from Pre-concentration of the Ore 172 Figure 6.22 – Grinding Energy Requirements with- and without Pre- concentration………………………………………………………….. 173 Figure 6.23 – Impacts on Power Requirements for Full Beneficiation……………… 174 Figure 6.24 – Extraparental Massive Sulphides at Pipe……………………………... 179 Figure 6.25 – Intraparental Brecciated Sulphides at Pipe…………………................ 179 Figure 6.26 – Low Grade Ore Stockpile at Pipe……………………………………... 181 Figure 6.27 – 13000 tpd Conductivity Sorting Flowsheet…………………………… 184 Figure 7.1 – Plot of Recovery vs. Feed Particle Size………………………............... 191 Figure 7.2 – Plot of Recovery vs. Wt% Rejects……………………………............... 192 Figure 7.3 – Plot of Recovery vs. Ore Feed Grade………………………………….. 192 Figure 7.4 – Projected Overall Capital, Operating and Revenue Impacts Based on Research Results……………………………………………………….. 193 Figure 7.5 – Worldwide Locations for Research Data from the Literature and the Testwork……………………………………………………………….. . 198 xiv List of Symbols (M)tpa - (Million) Tons Per Annum (N)AG - (Non) Acid Generating (S)AG - (Semi) Autogenous Grinding AC - Alternating Current ARD - Acid Rock Drainage ASTM - American Society Of Testing And Materials CDN$ (m) - Canadian Dollars (millions) CHF - Cemented Hydraulic Fill CRF - Cemented Rock Fill d80 - Screen Size At Which 80% By Weight Of Sample Passes dm3 - Cubic Decimetre DMS - Dense Media Separation EIT - Engineer In Training f80 - d80 Of Feed Material To Process g/t - Grammes Per Tonne HM - Heavy Media HPGR - High Pressure Grinding Roll ICP - Inductively-Coupled Plasma Mass Spectrometry kg - Kilogrammes kHz - Kilohertz kW - Kilowatts kW/m3 - Kilowatts Per Cubic Metre kWh/a - Kilowatt Hours per Annum m2(3) - Metres Squared (Cubed) m3/hr - Metres Cubed Per Hour MJ/m2 - Mega Joules Per Metre Squared MMI - Mine-Mill Integration MPa - Megapascals mV - Millivolts Ø - Diameter p80 - d80 Of Product From A Process xv PCI - Peripheral Component Interconnect PC Card PGE / PGM - Platinum Group Element (Metal) QEM/SEM - Quantitative Scanning Electron Microscopy RD - Relative Density RGB - Red-Green-Blue ROM - Run-Of-Mine RQD - Rock Quality Designation SG - Specific Gravity tpd - Tonnes Per Day TPM - Total Precious Metals UCS - Uniaxial Compressive Strength US$ (m) - United States Dollars (millions) VCR - Ventersdorp Contact Reef VMS - Volcanogenic Massive Sulphide Wt% - Proportion By Weight σ - Stress xvi Glossary Automation - autonomous monitoring and control of mechanical processes by means of electronics and microprocessors Backfill - waste material placed in underground excavations for mechanical support Barren - containing no, or low metal values Batching - the process of preparing concrete mixtures in discrete batches Binder - cement or other cementitious material used to bind other particles together Blasthole - large bore hole drilled in rock to house explosives for breaking the rock Breccia - rock composed of angular fragments in a matrix cementing material of typically different composition Bridge - an electrical circuit of typically 4 resistors used to measure a voltage Bushveld - igneous intrusive complex in South Africa Closure (1) - the process of closing and rehabilitating a mine site Closure (2) - time-dependent collapse of an excavation in hard rock Colorimetric - the measurement of spectral values in reflected light Comminution - the process of reducing the particle size of a mineral by crushing or grinding Composite Fill - backfill created by mixing traditional rock fill with paste Concentrate - a mineral product of sufficient metal grade to be considered for treatment in a furnace Core Logging - visual examination of drill core to determine its mineralogical and geotechnical features Dense Media - suspension of fine, dense particles in water used to raise the apparent density of the fluid Disseminated - fine, discrete particles of valuable mineralization evenly distributed in the host rock Drawpoint - location from which ore is extracted from a stope by LHD Extraction - proportion the original mineral resource which is removed by mining Fill Sequence - the order in which stopes are mined, and then again backfilled Float - light, barren rock which floats in the dense media separation process Flotation - separation of valuable from non-valuable minerals by means of differences in surface chemistry Footwall - barren rock located below the longitudinal axis of the orezone xvii Fuzzy Logic - Matrix transformation function used to give an approximation instead of a precise result Gabbro - basic, intrusive igneous rocks Gangue - non-valuable components in a mineral assemblage, typically silica and metal oxides Geo- metallurgy - interdisciplinary field linking the mineralogical and geo-technical properties of an ore with its behaviour during mineral beneficiation Grade Control - the practice of monitoring and controlling the metal content of the ore delivered from the mining operation to the mill Grade Distribution - the distribution of metal values by particle size in an ore Greenhouse Gas - gases such as carbon dioxide and water vapour which are known to increase the retention of radiation within the Earth’s atmosphere Hangingwall - barren rock located above the longitudinal axis of the orezone Hard Rock - rock that specifically requires drilling and blasting for excavation Hydraulic Hoisting - the transport of mineral slurries vertically by means of pumping Hydrodynamic - the measure of forces and motion in liquids Igneous - rock of magmatic origin that cooled and solidified within the Earth’s crust INCO - International Nickel Company Of Canada Leaching - dissolution of valuable minerals from the rock using acid or complexing agents such as cyanide LHD - front end loader for loading, hauling and dumping broken rock underground Mafic - igneous rock with relatively low silica content formed under equilibrium redox conditions Marginal - mining operations with low profit margins likely to lose money under adverse economic conditions Massive - valuable mineralization occurring as large, homogenous inclusions in the host rock Mesotexture - the supra-microscopic disposition of mineralization in rock Meta- sedimentary - metamorphosed sedimentary rocks Microtexture - microscopic disposition of mineralization in rock xviii Mineralogical - qualitative and quantitative evaluation of minerals in rock Mining Cycle - the sequence of activities in the mining of rock: drilling, blasting, mucking, ground control, haulage, hoisting Ore / Orebody - collective of mineralized rock within a mineral deposit which is economically valuable Ore Zone - well bounded localized area of valuable mineralization (reef or seam) Overburden - typically fractured, weathered waste material overlying a near-surface mineral deposit Oxide - valuable metals compounded with oxygen in rock e.g. Cr2O3 Paleo-Placer - fossilized alluvial mineral deposit Parametric - the use of arbitrary values as independent variables in a model or equation in order to determine the value of the dependent variables Particle Size Distribution - the proportion by weight of particles in a sample falling within various size classes Paste - a concentrated suspension of fine tailings, mixed with cement and used as backfill in mining Photometric - measurement of the properties of transmitted or reflected light Pillar - vertical columns of ore left behind in the orebody in order to provide local or regional mechanical stability Porphyry - a fine-grained igneous host rock containing conspicuous coarse phenocrysts of alkali felspar or other distinctive minerals Positive Displacement - method of pumping fluids by mechanically by means of an expanding and contracting chamber Productivity - the output of an industrial process measured per unit input Radiometric - measurement of the level and nature of radioactivity in materials Recovery - percentage by mass of the valuable mineral recovered to the concentrate in a metallurgical process Selectivity - measure of the ability of a mining method to extract valuable mineralization only Siliceous - containing a high proportion of silica or alumino-silicates Sill Pillar - extensive horizontal pillar left in massive orebodies for regional support Sink - dense, valuable rock which sinks in the dense media separation process Size-Assay - process to determine the distribution of metal values by particle size xix Sloughing - inadvertent collapse of typically hangingwall rock into ore after blasting Slump - measure of the consistency of typically a concrete mix Slurry - a heterogenous suspension of solids in water Stripping Ratio - the ratio by mass of non-valuable to valuable material removed from an open pit mine Sulphide - valuable metals compounded with sulphur such as Cu2S and NiFeS Surface Footprint - total surface area covered by the structures and excavations related to the mine Sustainability - combined economic, environmental and social accounting as applied to the mining industry Tele-operation - operation of multiple units of mobile mechanical equipment by one operator using telemetry Tertiary Regrind - third stage grinding of an ore in order to further liberate the valuable mineral Throughput - the mass flowrate of dry ore passing through a process Ultramafic - igneous rock with low silica content formed under slightly oxidising conditions UG2 - chromite seam of the Upper Group of the Bushveld Igneous Complex Value Chain - the definition and ordering of sequential value-adding steps in an industrial process Volcanogenic - mineralization of extrusive volcanic origin Waste - collective definition of all gangue material generated during the mining and processing of an ore into a concentrate xx Acknowledgements I would like to acknowledge the contribution and support of a number of parties during the course of the research. Firstly to my parents for their unwavering love and support throughout this period; to my co-supervisors, Professor Malcolm Scoble and Dr. Bernhard Klein for their support over the course of the research; my colleagues Mark Stephenson and Trent Weatherwax for providing complementary research efforts in this field, as well as the excellent support of Cristian Caceras and Paul Hughes of the UBC Geomechanics Group. I would also like to thank people from the organizations who supported the research: Simon Nickson, Dr. Harvey Buksa, Dr. Larry Cochrane, Boris Shepertecky, Kate Rubingh, Scott Mooney and Samantha Espley at CVRD-INCO Ltd (now Vale), John Vary (Retd.), Justin Widdifield and Dawson Proudfoot at Falconbridge Ltd. (now Xstrata Nickel); Joe Hunter at Placer Dome Ltd (now Barrick Corp); Bruce Fraser of AMIRA (Australia) and Chip Jones of the Doe Run Company in Missouri for providing supporting information as well as numerous geologists, mining engineers, metallurgists, summer students and others who assisted during the sharp end of the research. I would like also to thank the Collaborative Research and Development Grant Programme of the Canadian Natural Sciences and Engineering Research Council (NSERC) for matching support during the course of the research. 1 1. Introduction and Thesis Outline 1.1 Introduction In the Hard Rock Mining Industry, both in Canada and globally, a decreasing number of economic mineral deposits are found at shallow to medium depth, and most of those shallow deposits that remain are close to sub-economic grade and are required to be mined at high tonnages in order to show a return (Scoble, 1994). A large proportion of the remaining deposits are higher grade deposits which are presented in situations of either extreme depth, complex structure, poor ground conditions, or remote geography, one or a combination of which factors results in extreme pressure on costs, making the deposit sub-economic. Variations in the quantity and quality of ore delivered also negatively affects the performance of the overall mining system as the process control system takes time to adjust to these variances. Grade control and the maximizing of the grade of ore delivered to surface is thus of paramount importance in these situations. The commodity market which the industry supplies continues to be cyclical (Baxter & Parks, 1957) and economic pressures are further exacerbated in periods of declining or low commodity prices. A common solution to improving the economics of mining low grade orebodies is to increase the mining rate, with consequent decreases in unit costs and an increase in capital costs. However, increasing the mining rate reduces the mine life and significantly increases the physical and environmental footprint of mines, with negative impacts on public perception and closure requirements. Increasing the mining rate can also exacerbate ground control problems, with unforeseen negative impacts on costs. There is significant public pressure on the industry to reduce surface footprint, and it has been suggested that the productivity, economics as well as the environmental acceptability of mining marginal orebodies would be greatly improved by introducing innovative processes for the reduction of waste into the mining activity (Parsons & Hume, 1997; Warhurst & Bridge, 1996; Feasby & Tremblay, 1995). An effective approach for deep or otherwise marginal deposits would thus be to increase the mining rate, while simultaneously minimizing dilution, ground control problems and reducing the surface environmental footprint of the mine. This concept has found wide support from industry proponents including Rio Tinto, BHP and Placer Dome (Batterham, 2003, 2004; Hames, 2005; Cross, 2006), from whom the ultimate interpretation of the approach is presented in terms of 2 positing the ‘invisible mine’ where the maximum of the mining and processing activity is undertaken below surface. However, little work has been done in terms of a practical realization of this vision. A groundbreaking approach is required to achieve this and thus improve the feasibility of exploiting such deposits. The focus of this thesis is the specification, design and evaluation of an integrated set of technologies, the acceptance of which significant steps towards the realization of this concept can be made. Substantial efforts in the area of concept dissemination, technology transfer, and skills development together with appropriate change management are now required to begin implementation. The mining industry is innovative, and the potential for successful transfer of this technological approach into the industry is considered high. Substantial innovation has recently been seen in other fields such as process and environmental technologies, innovations which include Outokumpu and INCO’s adoption of the flash smelting process, developments in pre- concentration and waste disposal at Mount Isa Mines in Australia, the ISA Mill, Jameson Cell and the ISASmelt process also at Mt Isa, and ongoing developments in tailings and ARD management in Canada and worldwide (Bamber & Scoble, 2007). Recent developments in hydrometallurgy have enabled the lower cost processing of lower-grade copper and gold ores. Novel technologies such as Pressure Acid Leach and heap leaching have increased the range of ores that can be processed economically. Low-pressure, low-temperature metal leaching technology is enabling the exploitation of lateritic type base metal ores as well as methods such as in-situ leaching. Notable improvements have been made in the area of sulphur- and particulate emission levels at base metal smelters such as Copper Cliff, Falconbridge and Kidd Creek, as well as at Anglo’s Waterval smelter in Rustenburg, South Africa. There have also been ongoing developments globally in area of process control and operational data management, led by companies such as Western Mining, INCO and Outokumpu. However, underground hard rock mining faces particular challenges in the area of innovation, and it is recognized that substantial care in technology transfer and change management is required for success. There are a number of competing strategies whose effectiveness must be compared to the proposed approach in order for the benefits to be fully realized. Highly selective or resue mining methods can increase the viability of mining marginal deposits, particularly in narrow-vein situations. Resue mining is a two-phase mining process for narrow vein ores whereby waste in 3 the mining face is drilled, blasted, and stowed separately underground first, followed by drilling, blasting and removal of the mineralized zone. Selective blast mining methods (SBM) have also been suggested where the narrow vein ore and waste are blasted simultaneously, but are segregated due to a blast design which ‘throws’ the ore and waste into different areas of the stope (Bock, 1998). However, adopting the abovementioned methods impact negatively on the productivity of the method when compared to bulk mechanized mining at conventional widths, and other solutions must be found. Alternatives to resue mining in narrow-vein situations have thus been developed. Randfontein Estates in South Africa experimented with thick-seam mechanized gold mining (‘TM3’) in the 1980’s. However the resultant decrease in head grade was considered economically unacceptable. A mechanized alternative to the resue mining of steeply-dipping narrow vein gold deposits of the Witwatersrand was proposed by Pickering et al (1999). A similar mining method has since been adopted at Placer Dome’s South Deep Mine in some sections: it was found that the cost of the bulk method was 30% lower than the conventional method, however the planned grade of ore delivered from the thick seam mechanized sections was some 30% lower than from the conventional longwall mining sections, which can be attributed solely to additional dilution arising from the mining method (James & Rafffield, 1996). The development of low-profile diesel LHD’s has now enabled higher- productivity methods to be employed while maintaining grade in mining heights as low as 1.5m at dips of up to 12º, and many manufacturers now offer reliable, high productivity mining suites for mechanized mining at these mining widths. More recent efforts to increase the efficiency and effectiveness of underground hard rock mining include tele-operation of remote LHDs, continuous face advance techniques and Vertical Crater Retreat (VCR) Mining have been made, with mixed results. In the tele-operation of mining fleets in order to improve productivity underground there has been a significant effort, particularly by INCO locally in Canada. The Mine Automation Project (MAP) was a CDN$20 million, 5 year long research program involving collaboration between INCO, Tamrock, Dyno Nobel and CANMET, for the purposes of fully automating the mucking process (Graham & Morrison, 2003). The project was taken to the stage of an operating underground stope but further development of the concept was abandoned in 2002. Continuous mining systems are also attractive as a substantial component of the potential underground mining productivity is lost between conventional drill, blast and muck cycles. Recent research into these systems is aimed at overcoming this problem. In 1995, INCO 4 and Noranda entered into a 10 year, $35m partnership as HDRK Ltd with Tamrock and Wirth GmbH of Germany in order to develop a continuous hard rock development machine (Dessuerault et al, 2002; Graham 2001). A modified Eimco T60 roadheader capable of mining medium hard rock at acceptable rates was installed in a Manitoba mine, but further installations have not been forthcoming. A second effort comprising a C1000 Oscilloader in series with a mobile horizontal crusher and feeder unit, feeding an extendable belt which would follow the mining face by extending the head end of the belt was taken to the prototype stage, but the project was not pursued further. More effective efforts have been made at adapting mining methods for continuous operation and greater productivity. The Vertical Crater Retreat (VCR) method was developed in the 1970s as a means of increasing stope sizes and smoothing the production of ore from these stopes and facilitating the automation of mucking and hauling operations (Gertsh & Bullock, 1998). This increase in stope sizes was facilitated through the development of large bore underground drills and spherical charging of the holes, thus eliminating raise boring and slot cutting from the development cycle. VCR was adopted widely by INCO in the 1980’s as a replacement for massive cut and fill stopes in at their Ontario and Manitoba operations. However, VCR, as with TM3 and block caving, has not achieved the level of continuity and productivity originally envisioned, and furthermore results in a massive decrease in the grade of ore delivered from the stopes when compared to cut and fill methods, and thus results of the implementation are widely considered to have been negative and several mining sections at INCO have since reverted to cut and fill techniques (Cochrane, 2006). Several other innovative techniques for the enhanced mining of base metal hard rock ores have previously been researched at the US Bureau of Mines. These include in-situ leaching of copper and gold, as well as the leaching of previously unresponsive base metal ores by means of novel lixiviants, either in-situ or by heap leaching on surface (Davin, 1979; Rogich, 1982). However, the technology for leaching of unconventional base metal ores is presently inefficient and delivers unacceptable economics. Other solutions for the enhanced extraction of base metal ores must be found. The pre-concentration of Run–of- Mine ore (ROM) and subsequent disposal of the waste rejects underground, prior to transport and conventional treatment of the ore by fine particle processes such as flotation or leaching, is proposed as an extension of the concept of 5 Mine Mill integration philosophy as a more general approach to improving the efficiency, economics and environmental performance of hard rock metal mining. It has been previously stated that “pre-concentration is the rule rather than the exception” (Salter & Wyatt, 1991) yet the concept has not found wide application in the industry. Opposition to the concept in industry ranges from basic perceptions of unacceptable metal losses, high cost and low capacity of the technologies to a lack of understanding of the systemic impact of applying the concept. Further challenges to the application of pre-concentration, particularly when integrated into the underground environment are cultural: mining engineers do not understand mineral processing sufficiently well, and metallurgists are unwilling to extend their activities into the underground environment. Since 2003, strategic research at UBC has been directed primarily at developing applications for the technology at both surface and underground operations for several mining companies both in Canada as well as globally. Through the research, the benefits of ore pre- concentration, particularly when it is allied to technologies for the simultaneous preparation and disposal of the waste rejects and applied underground, have been shown to be sufficient to merit an objective consideration of this approach. The concept of ore pre-concentration itself is not new, and has been practiced on surface through modern mineral processing technologies such as Dense Media Separation (DMS) and sorting since the 1930’s (Munro et al, 1982). The concept of underground pre-concentration is also not new (Agricola, 1556; Lloyd, 1979); however, there are currently no examples of underground ore pre-concentration plants in operation. Several examples of sorting plants which operate underground are cited; however, sorting in this application is typically used for classification and diversion of the whole ore stream, and thus cannot be strictly considered pre-concentration (Kruuka & Briocher, 2002). The integration of pre-concentration and waste disposal activities into the mining process is considered a specific interpretation of the philosophy of Mine-Mill or Mine-to-Mill Integration (MMI). The Mine-Mill Integration philosophy arises from the desire to integrate the downstream processes such as ore handling and mineral processing more closely with the mining activity in order to improve efficiencies in the whole operation (Stephenson et al, 1971). Several exponents of this philosophy can be found in the literature, and typical aspects include blast fragmentation analysis, blast fragmentation – comminution interactions, and the application of down-the-hole electromagnetic or optical sensing for improved geo-metallurgical characterization of the in-situ ore in preparation for extraction and processing. The integration of automated ore pre- 6 concentration and waste disposal systems into the mining environment, whether on surface or underground, is proposed as an interpretation of this philosophy as a means of reducing the selectivity required of the mining method, increasing productivity and lowering costs while simultaneously reducing the quantity and increasing the grade of the ore delivered to downstream processes (Bamber et al, 2005). Due to the prospective rejection and disposal of a significant quantity of waste from the ROM ore stream, the capacity and thus size of surface facilities such as milling, tailings disposal and associated surface infrastructure would also be significantly reduced compared to the bulk mining scenario. The concept has been widely supported in principle by various industry and academic proponents (Hames, 2005; Batterham, 2003; Cross, 2006; Scoble et al, 2006; Martens, 2007) yet has presently not found application in the industry. It is suggested that present resistance to the application of the concept on technical or economic grounds is unfounded, and the thesis seeks to define appropriate parameters for the performance of the technologies on typical hard rock ores as well as identify and quantify the benefits for a range of operational situations. Pre-concentration technologies seek to identify and reject barren siliceous waste from the Run-of-Mine ore stream by means of several available coarse-particle separation processes such as optical sorting, conductivity sorting, or dense media separation (DMS) and reject it. The process results in the production of a large quantity of coarse siliceous waste and thus consideration of appropriate solid waste disposal strategies is also required for a complete evaluation. It is suggested through the research that the opportunity for the pre- concentration of ore is a general case in hard rock metal mines, and that there is a specific window of operational features such as ore mineralogy, operating cost structure, level of capitalization and organizational topography within which the technical, economic and environmental impacts and benefits of pre-concentration are positive. Where applicable, the adoption of the integrated mining, pre-concentration and waste disposal approach can reduce the quantity of material transported, processed and disposed of on surface thus significantly reducing the surface footprint of mines. Ore reserves are increased through the reduction in costs and it is thus believed that through these impacts, the sustainability of hard rock mining is significantly increased. 7 A wide range of ores have been tested, and results from the research and testwork have been augmented by additional supporting data from the literature. The research was conducted using a team approach; central to this thesis is the design of the research programme, research planning and supervision, target generation and the conduction of specific site investigations on the research targets, sampling and testwork, analysis and evaluation of results. Additional supporting testwork was provided by two Masters students working in the Mine-Mill Integration Group: Mark Stephenson on the testing of conductivity sorting on the Pipe ores, and Trent Weatherwax on the metallurgical and geotechnical testing of the Xstrata ores. It is thus suggested from the results that there is a more universal case for ore pre-concentration, either on surface and underground, the adoption of which is considered an appropriate application of Lean Manufacturing in Hard Rock Mining, a concept which has had broad success in the manufacturing sector in improving production efficiency. An appropriate envelope of overall cost to ore value ratios has been defined for mining operations within which a significant degree of waste rejection at acceptable metal recoveries can deliver substantial technical, economic and environmental benefits for the enhanced exploitation of these deposits. 1.2 Significance of the Research The anticipated impact of the proposed approach on hard rock mining in terms of delivering these technical advantages, economic benefits and improvements in environmental performance and public perception is considerable, and there are several mineral districts in particular which are expected to benefit from the application of the research outcomes. These include: • low grade ultramafic Ni ores of the Thompson Nickel Belt type • deep massive sulphide ores of the Sudbury Igneous Complex for example Onaping Depth and Nickel Rim • narrow vein UG2- type ores of the Bushveld Igneous Complex situated below 1500m which are presently uneconomical to exploit through conventional methods • Elsberg and VCR-type ores of the Witwatersrand basin situated below 3000m which are presently uneconomical to exploit • Cordilleran-type copper porphyry ores at block cave operations such as the Candelaria and North Parkes mine 8 Based on the anticipated impact of the research, two commercial spin-offs have been developed in the course of the research. One is focused on the evaluation of opportunities and design of the technologies described for mining companies both in Canada as well as globally. A second company is being been formed to focus on commercializing the sensing technologies that have been developed in the course of the research. 1.3 Research Contribution Significant challenges to the continued sustainability of hard rock mining are extant. A novel integrated mining, processing and waste disposal approach is proposed, where this potential can be identified, for improving the sustainability of extracting a wide class of mineral deposits compared to the conventional approach. A general case for the pre-concentration of ore prior to conventional treatment by grinding and flotation has been identified in the course of the research which is considered a significant contribution. The research tools that have been developed through the course of the work are also a significant contribution in terms of enabling the rapid and cost-effective assessment of a particular opportunity for pre-concentration, together with a rapid quantification of the expected benefits. Innovative improvements in laboratory techniques such as dense media testing, and the characterization of the optical and conductivity properties of ore with reference to sorting have also been made. In particular, the induction-balance sensor that has been designed and tested is a particularly unique contribution, enabling the characterization and grade evaluation of a class of low grade ores previously unresponsive to conventional conductivity methods. The future potential of this sensing technology in terms of exploration, down-the-hole sensing and ore sorting has yet to be explored. 1.4 Thesis Outline Ore pre-concentration, prior to conventional fine-particle processing such as leaching or flotation, and the subsequent disposal of the solid waste in the mining void, is proposed as a specific interpretation of mine-mill integration in order to improve efficiency, economics, and environmental impact for the increased sustainability of hard rock metal mining. The thesis is a comprehensive treatment of the design, application and evaluation of the ore characteristics as well as the mining technologies applicable to the realization of this concept. 9 In Chapter 2 the philosophy of mine-mill integration, and in particular the integration of ore pre- concentration and waste disposal technologies into the hard rock mining system, is presented as an application of Lean Manufacturing for the increase in efficiency of the. Chapter 3 presents a review of the enabling technologies such as sorting and dense media separation, as well as related waste disposal technologies such as paste fill, cemented rock fills, and composite ‘rocky’ paste fills. A discussion of the integration of these technologies into the hard rock mining environment is also presented. In Chapter 4 the geo-metallurgical experimental methodology for assessing the potential for ore pre-concentration of a candidate ore, characterizing the products of pre-concentration, characterizing the waste rejects for disposal, as well as designing the appropriate pre-concentration and fill disposal systems is presented. Several examples from the testwork are cited to illustrate the results of a typical investigation. A computerized parametric methodology for the valuation and evaluation the concept versus the conventional case has been developed over the course of the research. The model uses the results of the geo-metallurgical evaluation as well as additional field data on the deposit for the assessment of the impacts and benefits of implementing the integrated mining, processing and waste disposal approach on a typical case study. Chapter 5 presents the features of this model which has been used to evaluate the potential of the approach on a number of deposits. Case studies conducted in the course of the research have typically comprised literature review, site investigation, sampling, testwork and evaluation using the methods described in Chapters 4 and 5. Ores from over 26 case study deposits have been sampled and tested, and complemented by additional data for ores from the literature. Case study results from the research to date are summarized in Chapter 6. Additional details of the case studies are presented in the Appendices. Chapter 7 presents a discussion of the results obtained to date, data supporting the concept of a general case, conclusions as to the significance of the research as well as recommendations for future development of the concept. 10 2 On the Application of Integrated Mining, Processing and Waste Disposal Systems in the Hard Rock Mining Industry 2.1 Introduction The pre-concentration of Run–of- Mine ore (ROM) and subsequent disposal of the waste rejects underground, prior to transport and conventional treatment of the ore by fine particle processes such as flotation or leaching, is proposed as an extension of the concept of Mine Mill Integration philosophy, as an improved approach over conventional techniques such as more selective mining, resue techniques or simply increased mining rate, for improving the efficiency, economics and environmental performance of hard rock metal mining. The approach is proposed as an effective application of the Lean Manufacturing philosophy (Shingo, 1992) to Hard Rock Metal Mining. Success in ore pre-concentration involves the identification and rejection of a significant quantity of barren, siliceous gangue material from the Run-of-Mine (ROM) ore prior to conventional fine-particle processing. It is important to properly classify what is meant by ROM ore in this context. ROM ore typically includes both valuable and non- valuable components in the form of dilution. Dilution of the ROM ore is usually defined as follows: %100×⎥⎦ ⎤⎢⎣ ⎡ +=∂ toretwaste twaste - (1) Where twaste is the quantity of non-valuable rock component Levels of planned dilution can vary widely depending on the mining method (Table 2.1). Table 2.1- Planned Dilution by Mining Method (Canadian Mining Journal, 2007) Method Ave (%) Min (%) Max (%) Shrinkage stoping 13.3 5.0 20.0 Open Stoping 14.0 5.0 32.0 Blasthole 15.6 6.0 25.0 Cut & fill 15.7 5.0 30.0 Longhole 23.3 15.0 30.0 VRM 27.5 18.0 37.0 Caving 52.0 52.0 52.0 A second observation to make in this context is that the mineralization in most hard rock ores, including ores such as porphyry ores conventionally considered low grade and homogeneous, are in fact highly heterogeneous in terms of the macro-mineralogy of the orezone. Other factors 11 impact on the relative value of ore delivered to surface when compared to the value of the ore resource. In the process of creating the ore reserve by delineating the resource into ore blocks using the Smallest Mining Unit method (SMU), material which is not mineralized is included in the definition of ore, thus reducing the grade of the reserve; furthermore material which is mineralized is inadvertently excluded from the reserve through simple lack of resolution in the method, thus reducing overall extraction of the resource. ROM ore, in addition to planned dilution, thus typically includes additional barren material which may be rejected by the pre- concentration process, including unplanned dilution, barren gangue minerals included in the definition of ‘ore’, and gangue minerals (such as metal oxides and sulphides) directly associated with the valuable metal component (Figure 2.1). 0 20 40 60 80 100 W t% Uplanned dilution Planned dilution Included gangue Associated silica Associated sulphides Associated metals Metal of interest Figure 2.1 Typical Composition of ROM Ore A large proportion of the ROM ore thus comprises both planned, unplanned, and internal dilution by gangue materials, as well as valuable minerals from material defined as ore as well as valuable minerals from material defined as ‘waste’. Unplanned dilution also varies widely by mining method, in similar proportions. Loss of valuable fines has also been consistently noted to result in negative deviations to the head grade of ROM ore, particularly in gold, as well as high- grade base metal operations. The grade of ROM ore can thus vary significantly from plan (Howard et al, 2005), and this presents huge challenges in the efficient operation and control of the production system. Both grade and tonnage variations can be principally attributed to variations in the level of ore dilution as defined above, as well as the loss of valuable fines. The aim of an appropriate process strategy would be to maximize the liberation of these gangue 12 components at a coarse particle size, identify and reject them from the ROM ore as early as possible in the mining cycle, while retaining valuable fines components, thus maximizing the retention of the valuable components. Ore quantity is minimized while maximizing quality. The capacity required of all systems downstream of this process is significantly reduced and is thus considered to be a Lean approach to hard rock mining. Lean Manufacturing represents a philosophy for the structuring, operation, control and management of production systems for the elimination of waste, unnecessary work and hence an increase in efficiency of the industrial system, the application of which has resulted in enormous improvements in productivity and efficiency at many corporations. It is a concept that developed in the context of the motor manufacturing industry in the 1950’s, and has found its most significant application at Toyota Motor Corporation in the form of the Toyota Production System. Toyota is considered the most globally successful motor manufacturer, and the Lean Manufacturing philosophy is widely credited as the single most important contributing factor to this success. More recently, Lean Manufacturing as it is applied at Toyota has been studied and put into practice at companies such as GM, Motorola and Boeing Corp (Yingling et al, 2000; ). The fundamental concepts of Lean Manufacturing as originally applied at Toyota can be broken down into three principal areas (Shingo, 1992): design out overburden (muri) smooth out production (mura) and eliminate waste (muda) from the manufacturing system Overburden can comprise unnecessary manufacturing capacity, systems or people. The smoothing out of production entails efforts to remove variance from the quantity as well as quality of the product in question. The elimination of waste embodies the physical removal of unnecessary product (as in machining), the removal of waste products (whether solid, liquid, or gaseous), as well as the elimination of re-work due to poor product quality. Seven types of waste are typically identified in this context (Ohno, 1988): wasteful processes; overproduction; excess inventory; rejects/rework; transport; waiting time; and unnecessary motion or activities. Lean Manufacturing principles are typically applied to the behaviour and actions of workers and supervisors, which encourage the identification and removal of examples of such waste from the manufacturing system through the application of ‘Kaizen’ (Shingo, 1992). Only in rare cases does the application of the Lean Manufacturing philosophy lead to the re-structuring of a 13 business or process, although this can result in substantial benefits in terms of the objective of Lean Manufacturing. The application of ore pre-concentration and subsequent waste disposal technologies to the hard rock mining system is expected to result in a substantial reduction in system overburden, and a significant smoothing of production through the rejection of significant amounts of waste as early as possible in the mining cycle. It is this application of pre-concentration and waste disposal to the basic design and organization of the Hard Rock Mining production system which is the focus of this thesis. In order to evaluate these impacts to the capacity and performance of hard rock mining system, the system must first be adapted and configured appropriately for analysis in terms of the Lean Manufacturing concept. 2.2 The Application of Lean Manufacturing Concepts to the Hard Rock Metal Mining System Lean manufacturing philosophy is fundamentally and originally a concept for addressing these issues of waste and inefficiency in discrete production systems such as manufacturing. However, with adaptation, the concepts of Lean Manufacturing can be applied to continuous systems such as the mining system (Yingling et al, 2000; Vagenas et al, 1995). Companies such as Rio Tinto and Barrick Gold Corp have applied some concepts of Lean Thinking to their operations (Krawchyk & Ghuldu, 2006; Dunstan et al, 2006), particularly at milling and smelting operations, with some success. However, these previous applications of Lean Thinking have focused mostly on organizational systems and personnel behaviour in the mining environment. Six Sigma, for example, is a system for the adoption of waste removing behaviours which has found wide application in the industry. However, Six Sigma itself is fundamentally a measure of quality alone, and does not embody the full breadth and depth of thinking implied in Lean Manufacturing as it does not address potential changes to the fundamental characteristics of the production system, and the results of applying Six Sigma in order to improve the quality of the core product (ore) in the mining industry have been at best marginal. Vagenas, Scoble and Baiden (1995) explore a more structural interpretation of Lean mining through the application of automation and tele-operation to reduce the total complement of mining equipment required for a given production rate. However, even this interpretation of Lean mining fails to address the significant problem of varying quantity and quality of the product of the mining system. Further application of the concept is required to address this. Product quality variations in mining can be significant: the mass flowrate, particle size and chemical composition of ROM ore delivered from the mining operation to the metallurgical 14 operation is known to vary widely on a daily basis (Hinde et al 1986; Lloyd & van der Walt, 1986; Tamlyn, 1994; Willis, 2000; Howard et al, 2005). Reducing these diurnal fluctuations is a significant challenge, and achieving significant and permanent gains in ore quality requires more than just changes to the attitude, behaviour and performance of the personnel operating production system. Structural changes are required in the way that ore is delineated, mined and processed in order to achieve the next step forward in the quality and consistency of production from mines. Lean Manufacturing has been previously examined in this particular context (Steelman et al, 2007), where the potential applications of Lean Thinking to core activities within the mining system (such as mine planning, material handling and mineral processing) as well as on common services and support functions to the core process (such as procurement) were highlighted. Key questions asked in the Lean review process were: 1. What are the system bottlenecks? 2. What is the efficiency, reliability, utilization and productivity of the system? 3. How effectively are common services and support functions delivered? 4. How well does the management system support the existing production system? 5. How well does the operation of the system match the planning of the operations? 6. How is ore dilution, grade and ore chemistry managed? 7. What is the variability of the feed- and product quality in the process? The evaluation and recommendations of the study focused mostly on issues 1 – 5. Addressing issues 5 and 6 was seen as a significant challenge, as the present hard rock mining paradigm is primarily focused on economies of scale where quality of the product i.e. the ore is seen as secondary to achieving these economies of scale. Modifications of the process management system, or adjustments to the behaviours and attitude of operating personnel only are considered ineffective in addressing this issue. It is posited that the achievement of meaningful improvements in total product quality requires a re-thinking of mining production philosophy and a restructuring of the production system itself. 2.3 Ore Pre-concentration and Waste Disposal as an Example of the Application of Lean Manufacturing in Mining In terms of addressing this key issue of product deviation in terms of quantity and quality, it must firstly be noted that ore, and more importantly the metallic component of the ore, is the primary product in all steps of the mining process, and variance in the quality of the product, occurs on a continuous basis in a typical operation in 4 primary ways: 15 excess of ore (overproduction) paucity of ore (underproduction) under quality ore (below grade) over-quality ore (above grade) The advent of such situations cause unplanned deviations within the production system, and can cause substantial losses due to loss of product during the lead time in which the system adjusts to accept the new product quantity or quality. Losses can occur in delivering low grade ore as the increased content of gangue material and deleterious elements can negatively affect recovery to concentrate in the mill. Losses also occur in the delivery of ore of higher grade than planned due to slow responses in the plant control system in adjusting process parameters such as sump densities or reagent dosing appropriately. Such deviations can arise due to a total system failure, as in a hoist or shaft failure, or simply the arrival of a batch of above-grade ore in the flotation circuit, where sulphides may be inadvertently lost to tailings due to the overloading of the froth. Any of these situations may occur individually or simultaneously (e.g. overproduction of low grade ore), further exacerbating the projected loss. Data from field operations confirms this observation. Previous studies of ore quality variability at BHP Pilbara indicate that variations of up to 2% in head grade and variations of between 0 and 200% of the planned tonnage to the mill led to a long term decrease of up to 1% in recovery of metal to final product (Kamperman et al, 2001). Figure 2.2 - Variance in iron grade and tonnage shipped ex Pilbara, BHP Billiton, WA (after Kamperman et al 2001) The results of studies on the impact of variable of ore quality and quantity on operational performance at the Ulanskii and Krivbas mines in Russia made by the Leningrad Mechanobr Institute through sampling, modeling and simulation support this data (Kazanskii, 1973; Fugzan et al, 1971). An increase of 18% in copper grades, coupled with a reduction in grade variability of 35% was shown to improve overall metal recovery at Ulanskii by 0.84%. Several methods for 16 stabilizing ore quality at the mine, including increasing the selectivity of mining, and removal of ore dilution were recommended. In the study by Epelman and Filiminov (1972) at Krivbas iron ore mine, it was shown that ore grade and tonnage varied significantly against plan and that a reduction in variation in the Fe content of the ore of between 2-4%, coupled with a reduction of 10% in the deviation of tonnage from plan at the mine resulted an increase of 21500 tonnes (5%) in metal recovered to concentrate. It was identified during the study that quality deviations from stope to stope were unavoidable; however ore quality stabilization by either more selective mining, increased grade control, tactical blending of ore streams underground or in fact separation of the individual ore streams of various quality were suggested as an economic alternative to the base case. Such deviations lead to unnecessary capacity in transport and treatment processes and furthermore that are problematic to the efficient operation of the mill. Several remedies to this deviation were recommended: increasing mining selectivity; improved grade control in the stopes; re-planning of mining sections by ore type; stockpiling and blending of the ore underground; transport of all distinct ore types in discrete streams; surface stockpiling and blending through re-handling (Kazanskii, 1973; Fugzan et al, 1971; Epelman & Filiminov, 1972). All of the above remedies are, however, considered costly in return for comparatively marginal benefits and thus require careful consideration at the planning stage. The strategic application of ore pre-concentration, either on surface ore underground, and the subsequent disposal of the solid waste prior to delivery of the ore to the mill is suggested as a more efficient and cost effective means of addressing these unplanned deviations in ore quality and quantity that have been identified as deleterious to the mining and processing operation. The integration of the pre-concentration step into the surface or underground mining system implies structural changes to the basic mining system. Several aspects of the mining system thus must be considered in preparation for the evaluation of the integration of these processes as a Lean Manufacturing approach. Fundamental differences between the mining industry as a continuous production system, and a typical discrete manufacturing entity exist (Table 2.2), and thus a reduction of the typical hard rock mining system into discrete processes must be made. 17 Table 2.2 – Functional comparison of manufacturing and mining systems Mining Manufacturing Process is continuous and cannot be arbitrarily stopped Systems and products are discrete and production is discontinuous Geologically and geographically constrained Manufacturing sites can be located where market and infrastructure allow Inherently variable and challenging environment Production occurs within a stable and regulated environment Operations are typically in rural or remote environments Located near large centres transportation infrastructure Quality of the raw material itself is highly variable Raw materials and supplies are procured to specified standard and quality Products are shipped as raw materials to the next process Products are typically final products which go to market For this purpose, the concept of the Value Chain, as originally posited by Porter (1980) is applied (Figure 2.3). Figure 2.3 - Generic Manufacturing Value Chain (after Porter, 1980) Like Lean Manufacturing, the Value Chain concept was originally posited for discrete systems such as supply-chain or manufacturing enterprises, however the concept has found wide application in the mining industry, as it provides a convenient way to describe operations to investors as the physical boundaries between primary activities such as mining, hoisting, surface transport and beneficiation provide convenient junctures at which to break this otherwise continuous process chain into discrete steps. For practical purposes in this thesis, a logical boundary for consideration is set at the smelter, at which point the mineral concentrate becomes an impure metal. The lower battery limit of the system is considered to be the mineral resource 18 itself. An important interaction that has been identified in the analysis is the interaction between the mineral resource and the minable reserve, thus implying a significant impact on the sustainability of mining operations. Secondly, we introduce conventional waste disposal activities in the form of tailings disposal and the handling of development waste as an additional step in the mining value chain. Finally, the integration of coarse particle separation and solid waste disposal technologies into the mining process introduces three areas of recycle to the traditional system. Firstly, barren development waste or overburden that immediately arises during the mining sequence may be returned to the void. The balance of unreturned development waste or overburden is traditionally taken to surface and stored in dumps. The second entails the processing of this waste as fill for the underground stope. In the surface context there may be several different classes of solid waste, for example reactive- and non- reactive wastes, which must also be processed for appropriate disposal. A third area of recycle is introduced to in those operations utilizing hydraulic- or paste fills, where fine tailings traditionally disposed of on surface are prepared for placement as either cemented or uncemented fill material in the underground stopes. These steps must be incorporated into the value chain in order to represent the integrated mining, processing and waste disposal system in discrete steps, with recycle streams as appropriate (Figure 2.4). Figure 2.4 –Mining Value Chain with Recycle Each step in the value chain has certain properties which must be constantly evaluated in the terms of the ‘Leanness’ of the mining system. Four key properties for consideration are the capacity and cost of each step in the chain, the complement of personnel required to operate and manage each step, and ultimately a measure of the quality of the product delivered by the 19 process to the next step in the chain. Secondary properties may include measures of productivity, reliability, and utilization for each step of the process. In this thesis the focus is on primary properties, however some consideration of these secondary properties is made in specific cases. The focus in this dissertation is on primary impacts to the capacity, capital cost, operating cost, operational efficiency (in terms of metal recovery) and waste disposal requirements of the hard rock mining system. Capacity requirements in mining are typically fixed between the mining and the milling rate. The capacity / processing rate of the mining step in the value chain is typically determined as a function of the size and grade of the ore reserve, for example through the use of Taylor’s formula for underground mines or by Lerchs-Grossman type optimisation in the case of open pits (Gentry & O’Neil, 1984). Until a concentration step occurs, the capacity, and thus the cost and personnel requirements for each subsequent step is determined (and constrained) by the output of the previous step in the chain. This is represented in the flowchart (Figure 2.4) by the size of the block. By removing dilution, as well as interstitial gangue, pre-concentration reduces the quantity, improves the quality, and reduces the variance in the quality of the ore arriving at the following step in the value chain, thus freeing the downstream processing from this typical constraint. We can thus evaluate the impact of introducing the pre-concentration step on surface prior to conventional fine-particle processing on the traditional mining value chain, for example as at Mount Isa Mines in Australia (Figure 2.5): Figure 2.5 –Mining Value Chain with Surface Pre-concentration 20 Pre-concentration by dense media separation is introduced on surface prior to the transport of the ore by road to the mill. Additionally, the coarse waste rejects are returned via raise-bore to the underground stopes where they are mixed in a ratio of up to 25% by mass with conventional paste fill for use as composite fill (Kuganathan & Shepherd, 2001). The capacity and cost of processing in downstream processes is reduced proportional to the degree of waste rejection. An increase in mill capacity of 50% was achieved through reductions in the quantity of ore to be processed as well as an 18% reduction in the Bond Work Index of the ore (McCullough et al, 1999). This strategy is expected to benefit both open pit and shallower surface operations. The impact of pre-concentration can be further enhanced if it is introduced earlier in the mining cycle, or value chain (Klein et al, 2002). There are two logical places in the value chain where pre-concentration can be introduced in the underground hard rock mining cycle – after blasting prior to underground haulage, or after haulage prior to hoisting (Figure 2.6); Figure 2.6 – Mining Value Chain with Underground Pre-concentration Pre-concentration of the ore occurs prior to haulage or hoisting from underground. A solid waste recycle stream is now produced directly underground, and combined with tailings, water and cement as required form surface can be disposed of as fill. Capacity, quality and cost benefits are passed down all the steps in the value chain subsequent to the introduction of the process. As there are more steps in the process, benefits accruing in those processes are increased. Furthermore, the impacts and benefits are significantly enhanced for deep mines, or mines remote from the metallurgical operation. Overall impacts can be quantified parametrically using the model. A new step is introduced in the process, requiring additional capital, labour, power 21 water and materials; however, the capacity, and thus the capital cost, operating cost and personnel requirements of all subsequent operations are reduced. Gross unit revenue is decreased due to additional metal losses incurred, however based on experience from previous studies, these are minimal and the overall impact is typically net positive (Scoble et al, 2006). Furthermore, the additional recovery penalty is often offset, if not overcompensated for, by increased recovery in the surface mill due to the higher feed grade and reduction in deleterious elements such as silica and talc in the ore (Stephenson, 2006; Blower & Kiernan, 2003). In such cases, the overall reduction in costs and increase in profitability decreases the economic cutoff grade for the conversion of the ore resource into a reserve; extraction of the ore is thus increased, further offsetting the negative revenue impacts (Bamber, 2006). Additionally, more cost-intensive processes can now be used on the resultant higher-grade ore, incurring further benefits. Finally, as in the case of operations such as Mt Isa, which employ the pre- concentration waste in combination with surface tailings as fill underground, there is a reduction in the total quantity of physical waste left on surface after operations have ceased, thus reducing closure requirements, which must be quantified. 2.4 ‘Lean’ Case Study - Xstrata Nickel Ontario Operations In the context of Mine-Mill Integration, the above concepts can be extended to groups of operations in order to fully realize the power of applying the Lean approach to the planning, development and operation of mines. A typical feature of mining is that the number and diversity of the steps in the value chain decrease at every step in the chain. Thus, many mines can typically feed a mill, many mills are required to feed a smelter and typically many smelters are required to justify a refinery. Many mining operations are standalone, others are logically integrated with a dedicated mill. It is common for a group of possibly open-pit or underground mines to feed a mill; however, not many individual mining and milling operations are large enough to justify a standalone smelter, thus a battery limit in terms of location (and possibly ownership) typically occurs at the point of feed to the smelter. Due to an increase in the degrees of freedom in the integrated production system, product quality variations and inherent losses are amplified in such a case. In order to address this variability, an integrated mining, pre- concentration and waste disposal strategy is proposed for Xstrata’s Ontario Operations (Figure 2.7). 22 Figure 2.7 – Geographical Disposition of Xstrata’s Ontario Operations (shown in red) (after Romaniuk, 2006) Present mining and processing philosophy would indicate centralized milling for this group of operations in order to maximize economies of scale. At Xstrata Nickel, this is the case: mining operations are remote, however all ore is milled at Strathcona located in the North West of the Sudbury Basin. Nickel concentrate transported 75km SE to the smelter in Falconbridge Town; copper concentrate is trucked 300km NW to the Kidd Creek smelter. Ores is nominally stockpiled and blended at the mill, however significant variations in ore grade and quantity as delivered from the mine have been noted during studies. There are several reasons for this: Several different ore types are exploited simultaneously at Xstrata’s Sudbury operations. These typically fall into 2 main categories, but up to 9 sub-ore types are identified (Proudfoot, 2006). Ores range from massive vein stringer type deposits, high in copper and precious metals and mined using highly selective mining equipment, to low grade nickel ores mined using bulk underground methods. Many mines produce both types of ore, and hoist them separately, however all ore types are mixed at the mill and milled simultaneously. The geographic diversity of Xstrata’s operations is increasing as the company matures. Traditionally ores from the Strathcona locality only have been mined and processed, the exception being ores from the Thayer Lindsley Mine. A major expansion is currently 23 underway at the Nickel Rim mine, which is located 75km from the existing mill, through the town of Sudbury to the SE. Several operations require fill, and this is currently sourced from development waste at the mines, supplemented by classified tailings for those mines near Stratchona, and slag ballast trucked from Falconbridge smelter in the East to operations such as T-L which require ballast. An integrated pre-concentration strategy for such diverse operations requires complex consideration. However, on closer examination of the range of ore types and geographic disposition of these operations, there are a number of obvious scenarios for the application of pre-concentration at Xstrata which are suggested for consideration in order to explore this concept: Customized milling at each mine by ore type Centralized pre-concentration of ore at Strathcona prior to milling Pre-concentration of ROM ore at each mine on surface Pre-concentration of ROM ore at each mine underground Creation of custom ore streams from each individual ore type through appropriate pre- concentration and/or classification of the ore underground Of the nine principal ore types that were tested in the study, all displayed significant potential for pre-concentration (Bamber et al, 2005, 2006; Weatherwax, 2007). Footwall-type ores that were tested indicate a high degree of rejection of a barren waste, producing a high grade, possibly direct shipping pre-concentrate. Waste rejects were tested and found to be suitable for use as an aggregate for high-strength fill (Bamber et al, 2006). Contact-type ores presented a different scenario, where the degree of waste rejection was lower, however it was identified that a number of custom ore streams could be generated – low grade or barren waste for fill, low grade and high-grade ore streams. The results were used as inputs into the parametric model to evaluate the alternatives for applying the Lean approach to the mining and processing of these ores. At this stage of concept development, the consideration of these options through scoping level or pre-feasibility studies is obviously not practical, and the use of parametric methods as applied in the case study is suggested as a cost-effective means of defining, evaluating and comparing options for evaluating this as a Lean approach to mining. Through systematic evaluation of the options, a final strategy comprising the following was recommended: 24 • Pre-concentration of Onaping Depth ore underground • Pre-concentration of Ni Rim ore on surface, with future potential for underground pre-concentration to be explored • Pre-concentration of TL ore on surface • Disposal of the solid waste underground at Fraser, Craig, Onaping, Thayer Lindsley and Nickel Rim Applications at the Craig and Fraser Mines were not foreseen as life of mine is limited at these operations. Several major operational benefits are however expected to be enjoyed at future mines. Hoisting capacity at the two deepest mines, Onaping Depth (2500m) and Nickel Rim (1500m) is reduced by 30-50%. The quantity of ore to be trucked from Ni Rim to Strathcona would be reduced from 1.4Mtpa to 800 000 tpa, reducing the number of trucks driving through Sudbury town from 6 per hour to 4 per hour. The quantity of ore arriving at Strathcona Mill, and thus the size of the mill, could be reduced by 35% from 2.2 Mtpa to 1.4Mtpa, together with a resultant increase in feed grade of 32%. Based on the testwork, mill recoveries, especially for nickel and precious metals, are expected to increase; grade variability is projected to be reduced due to lower overall levels of dilution, and a resultant opportunity to shut down an entire grinding line at the mill has been identified. Further benefits would be enjoyed at the tailings dam. Present tailings strategy involves the deposition of a high-sulphur and low-sulphur product. The dam is presently topographically constrained, however, with pre-concentration the total quantity of tailings arriving at the pond is reduced, and an opportunity to generate a single high-sulphur tail on surface has been recognized. The generation of additional fill material from the waste streams at Onaping Depth, Thayer Lindsley and Nickel Rim would reduce the quantity of slag required to be trucked from Falconbridge smelter back to the mines from 800 000 tpa to 200 000 tpa. Total energy requirements for mining and beneficiation are projected to be reduced by 8.8% through testwork and modeling. Ontario is furthermore moving to an energy credit system, with Greenhouse Gas (GHG) credits valued at $15/tonne equivalent CO2. Energy costs are projected to be reduced by 180MkWh/annum, thus GHG credit will be significant constituting a significant additional cost saving for the group (Bamber et al, 2007). These ores represent the majority of the present and future production at Xstrata Nickel’s Ontario operations, and thus the potential impacts of adopting a Lean Approach as evaluated are essential for consideration for this group of operations. 25 2.5 Conclusions The need for improvements in the effectiveness and efficiency of hard rock mining is dire, particularly in conditions of increasing costs and declining long term metal prices. Mine-mill integration, and specifically the integration of pre-concentration and waste disposal systems into the hard rock mining process, is seen as a strategy for improving the efficiency, cost effectiveness and environmental performance of mining operations. The case for evaluating this approach as an application of Lean Manufacturing philosophy to hard rock mining has been presented. The application of the concept is challenging, particularly as a structural change in the way mining and processing systems are designed and built is required. The model of the traditional mining system must be modified in order to evaluate the impacts and benefits. While the typical manufacturing system is discrete and discontinuous, hard rock mining is a constant and continuous process and thus requires special consideration in this context. An approach for adapting Lean Manufacturing principles to the typical hard rock mining system, as well as the modelling of the hard rock mining system as a discrete series of steps in a Value Chain is presented. The successful management of ore quantity and quality is seen as a major obstacle in achieving Lean goals in the industry; ore pre-concentration, whether on surface or underground is discussed as a means of achieving these objectives. The process of extracting and processing ore is shown to be less dependent on the given grade and physical presentation of the ore, grade control decisions are very much reduced in the system and ore grade and quantity variations are minimized through the introduction of process control on these variables. Mining selectivity is automatically reduced, and productivity increased (Bamber et al, 2005). Recycle streams are introduced where appropriate, capacity and thus cost and personnel are reduced in these areas. A significant improvement in product quality and quality control, efficiency and effectiveness coupled with a decrease in costs is projected for the cases studied. The operating scenario of a major nickel and copper producer Xstrata Nickel is discussed, and it can be seen from the results that a number of alternatives exist for the application of the concept, each of which brings a significant benefit to the operations. Thus it can be seen that the application of pre-concentration to the hard rock mining system facilitates improvements in all three of the key areas of Lean Thinking: reductions in overburden, in terms of the utilization of unnecessarily large equipment, and the additional operational and supervisory personnel required in transport and processing operations 26 the ore production process is itself improved, and thus measures of the quality and quantity of ore produced are smoothed reduction in the quantity of waste material which traditionally has been unnecessarily processed in the mill, and the introduction of a new type of recycle stream which has the potential to reduce the quantity or alternately improve the quality of solid waste disposed of on surface increase in life of mine through increase in ore reserves at the same mining rate significant reductions in Greenhouse Gas (GHG) emissions due to a reduction in the energy intensity of the overall mining and beneficiation process For the ideal case, a successful application of the concept would result in the maximum extraction of ore from the resource, the optimum disposal of solid wastes underground, reductions in the capacity required of unit operations downstream of the pre-concentrator, and a reduction in the overall surface and environmental footprint of mines. By enhancing the life of a particular deposit, the duration of the economic activity is also extended with further benefits to dependent communities. The triple bottom line of corporate environmental accounting for sustainability has been stated to be ‘economy, environment and community’(Elkington,1994); it can be clearly seen that the integrated mining, processing and waste disposal approach impacts positively on all three of these areas, and thus the overall sustainability of hard rock metal mining. 27 3. Enabling Technologies 3.1. Introduction Improved mine-mill integration, in the form of the strategic application of ore pre-concentration, either on surface or underground, and the subsequent disposal of the solid waste underground prior to delivery of the ore to the mill is suggested a as a means of improving the efficiency, economics and environmental performance, and thus sustainability of the sector. As has been previously discussed, the pre-concentration of ore is not a new approach. However, it is believed that the potential for the pre-concentration of hard rock metal ores is a general case and must be investigated in each instance in order to determine the feasibility of introducing this step into the mining system, whether on surface or underground. Several enabling technologies have been identified in the course of the research. Process technologies to be considered include optical sorting, conductivity sorting and dense media separation. Pre-concentration by comminution and size classification as well as coarse-particle flotation are also suggested as technologies with great potential. The application of these process technologies to the ore generates a coarse waste rock stream, and thus necessitates the further consideration of solid waste disposal technologies such as paste fill, cemented rock fill and in particular composite ‘rocky paste’ fills for the disposal of the reject material. Integration of these pre-concentration and waste disposal steps into the value chain also necessitates additional consideration of impacts on the mining methods available. Two models for the application of integrated mining, processing and waste disposal technologies have been presented. In the first model, ore is pre-concentrated on surface prior to the transport activity. Waste rejected in this process may be classified by type and deposited on surface, as in open pit mines, or prepared as appropriate for disposal as unconsolidated rock fill in cases where no support is required, or alternately cemented rock fill or high-strength composite type fills in the case of mines which require the fill for additional geotechnical support (Figure 3.1). 28 Figure 3.1 –Integrated Mining Processing and Waste Disposal Approach For underground deposits characterized either by extreme depth, long underground haul distances, highly diluted ore, adverse ground control issues or perhaps with constraints on surface waste disposal, an integrated underground mining and processing approach is proposed, where ore is pre-concentrated underground, the pre-concentrate is hoisted to surface, and rejects are prepared in an underground fill plant together with tailings, cement and water from surface for disposal as backfill (Figure 3.2). General consideration of pre-concentration of hard rock metal ores either on surface or underground, and the subsequent disposal of the waste rejects underground is proposed where that approach can be demonstrated to be economically and environmentally superior to the conventional approach. It is the integration of pre-concentration technologies simultaneously with waste disposal technologies in the mining of hard rock base metal ores which is considered both novel and highly original and is the focus of this thesis. 29 Production Stope Water Cement LHD Fill stope Haul truck Ore pass Feed bin Feed preparation Coarse pre - concentration Fine pre- concentration Hoist to Surface Concs bin Backfill Preparation Waste bin 40-60% ROM Surface bin Pre-concentrate Transport 40-60% ROM Scalp Rockfill LHD ~5% ROM 100% ROM Surface Underground High-grade fines Classified Tailings Mill Tails To smelter Figure 3.2 –Integrated Underground Mining, Processing and Waste Disposal System Design This Chapter presents a review of the relevant technologies researched in the thesis and their application to the hard rock mining environment. Application of the system design to several of the available mining methods is also considered. Criteria and methods for measuring the potential for application of these particular technologies at selected operations have been established and tested through sampling and testwork in the laboratory during the course of the research, the methods for which are presented in Chapter 4. 3.2. Process Technologies Coarse particle separation methods such as sorting or dense media separation are traditionally used on ores such as massive oxides, as well as coal, to produce a final product. However, the same processes are considered applicable in a pre-concentration role, and indications from the literature are that a high degree of waste rejection can be achieved from a wide range of ore types at a coarse particle size (Lloyd 1979; Ferrara & Guarascio, 1980; Schena et al, 1990; Mohanty et al, 2000). More recent additional examples supporting the use of pre-concentration on previously unconsidered ore types have also been noted in the literature (LionOre, 2006; 30 Collins & Bonney, 1995; McCullough et al 1999; Wilkinson, 1985; Wright 1985; Peters 1999; Sivamohan & Forrsberg 1991, Jones 2007, Salter & Wyatt, 1991; Kowalcyk 2002). In further support of this data, over 26 additional ore types have been tested at UBC with good results (Bamber, 2004, 2005; 2006; Mayne, 2005; Simonian, 2005; Stephenson, 2006; Weatherwax & Gillis, 2006; Weatherwax, 2007). Several technologies such as sorting by size, density, colour or conductivity have been emploted. All technologies that have been surveyed are compact, low-cost, high capacity mineral processes showing good metallurgical performance in the application. A combined table of results from the research combined with results from literature is presented in Table 3.1. Table 3.1: Metallurgical Performance of Pre-concentration Technologies on Selected Ores Pre- concentration method Ore Type Feed size (mm) %Wt reject Metal Recovery Reference HMS Cone separator Pb/Zn 38 55.00 97.00 Wright, 1980 DMS cyclone Cu porph 38 33.00 96.00 McCullough et al 1999 DMS Cu porph 38 74.00 81.00 McCullough et al 1999 Shaking table/ Dynawhirlpool Ni 3 90.00 89.00 Ferrara & Guarascio, 1980 DMS Cu 13 75.00 95.00 McCullough et al 1999 Coarse flotation Wits Au 3 60.00 98.00 Lloyd 1979 Coarse flotation Wits Au 4 28.00 97.00 Lloyd 1979 Classification Ni 1 27.00 89.00 Mohanty et al, 2000 Model 19 sorter Pb/Zn 150 26.3 94 Collins & Bonney, 1995 Model 6 radiometric U3O8 150 39 96.5 Collins & Bonney, 1995 Radiometric U3O8 75 73 96 Collins & Bonney, 1995 Magnetic Ni 100 40 96.7 Collins & Bonney, 1995 Radiometric Au/U3O8 50 50 98 Sivamohan & Forrsberg 1991 Conductivity Ni/Cu 25 80 80 Conductivity Ni/Cu 100 60 90 Conductivity Ni/Cu 25 87 61 Conductivity Ni/Cu 70 10.5 99 Conductivity Ni/Cu 70 27.83 82 Model 16 optical Cu 160 98 97 Model 16 optical Cu 100 32.7 96 DMS Pb/Zn 38 39.6 96.3 DMS Fe 38 31.4 98.3 DMS Pb/Zn 19 36.7 93.9 Screening Cu 100 20 99 DMS Cu 38 80 93.8 McCullough et al 99 31 Pre- concentration method Ore Type Feed size (mm) %Wt reject Metal Recovery Reference Optical sorting Cu 100 44.8 94.8 Wilkinson, 1985 Conductivity sorting Ni/Cu 100 54 82 Wilkinson, 1985 Optical + conductivity Cu 100 50.2 97.7 Wilkinson, 1985 DMS Cu 44 55 94.5 McCullough et al 99 DMS Cu 38 35 97 McCullough et al 99 DMS Cu 87.5 55 97 Wright 1971 DMS Pb/Zn 19 37.4 98.7 Jones 2007 DMS Pb/Zn 19 32.7 97.9 Jones 2007 DMS Pb/Zn 19 23 99.4 Jones 2007 DMS Cu/PGE 19 44.3 97.5 Bamber DMS Ni 6.7 14 99 Bamber DMS Ni 75 13 98 Weatherwax 2006 DMS Ni/Cu/Co 75 32 89 Weatherwax 2006 DMS Ni/Cu/Co 75 24.5 95 Weatherwax 2006 DMS Ni/Cu/Co 75 54 98 Weatherwax 2006 DMS Ni/Cu/PGE 75 36.6 97.7 Weatherwax 2006 DMS Ni/Cu/Co 75 25.73 96.7 Weatherwax 2006 DMS Ni/Cu/Co 75 19.52 94 Weatherwax 2006 DMS Ni/Cu/Co 75 25.5 95.3 Weatherwax 2006 DMS Ni/Cu/Co 75 32.37 96.4 Weatherwax 2006 DMS Au 19 17.6 93.2 Bamber DMS Au 19 24 94.6 Bamber Radiometric Wits Au 250 44.1 87.9 Kowalcyk 2002 Radiometric Wits Au 250 29.5 92.6 Kowalcyk 2002 DMS Ni/Cu 300 22 95 Bamber DMS Cu/PGE 250 55 97 Bamber Size Cu/PGE 75 37 99 Bamber Size Cu/PGE 75 35 93.5 Bamber Size Cu porph 31.75 49 76.32 Burns & Grimes 1986 Size Cu porph 31.75 54.37 75.75 Burns & Grimes 1986 Size 500 67.64 64.49 Burns & Grimes 1986 The results suggest that a high degree of waste rejection is possible at a coarse particle size with good metallurgical recoveries on a wide range of ores using the separation technologies selected for the research. Detailed results from testwork on specific approaches from the research are presented in the Chapter. 32 3.2.1 Ore pre-concentration by Comminution and Size Classification 184.108.40.206. Development of the Concept Several citations from literature suggest that concentration by comminution and size classification alone would be effective for the concentration of some ores (Mohanty et al, 2000; Sivamohan & Forssberg, 1991; Jones, 2007; Burns & Grimes, 1986). Evaluation by the size- assay method is required to indicate an opportunity for this on a potential ore (Bamber et al, 2006). Using the size-assay data, similar results had been indicated in previous research with INCO (Buksa & Paventi, 2002; Bamber, 2004). The data indicated that Footwall ores of the Sudbury Basin exhibit the characteristic of both a coarse, barren fraction as well as significantly upgraded sulphide content in the fines fraction (Figure 3.3). 0 5 10 15 20 25 30 35 40 125 .0 0 100 .0 075. 00 53. 00 38. 00 26. 00 19. 00 12. 509.0 0 6.7 0 3.3 0 1.6 0 -1. 60 Size (mm) W t% PavGrd BuksGrd PavWt% BuksWt% Figure 3.3 – Size-assay data for INCO 153 orebody (left after Buksa & Paventi, 2002) It was suggested that this might indicate a case for the concentration of this ore simply on the basis of size alone. The literature suggests that this feature of variable grade distribution by size is common in a number of ores, including massive sulphide ores, base metal ores, as well as gold ores and some platinum ores, where, due to the friability of the high-grade ore, and the high density of the valuable mineral, the majority of metal values appear to be present in the fine fraction and thus this fraction is significantly upgraded compared to the average. Data presented from van der Berg & Cooke (2004) indicates that a characteristic upgrading in the fines of some Bushveld platinum ores could be a basis for pre-concentration of these ores prior to hydraulic hoisting (Figure 3.4). 0 10 20 30 40 50 60 +1 25+7 5 +5 3 +3 8 +2 6 +1 9+9+6 .7 +3 .3 +1 .6-1. 6 Size class (mm) W t% Wt% % Sulphide 33 Figure 3.4 – Pt enrichment of fines in Bushveld Pt Ores (after van der Berg & Cooke, 2002) This feature of ROM ore, if identified to be present, has great potential for the rejection of waste from ore at low cost. The characteristic size/grade relationship suggests one of two possible concentration opportunities: • Rejection of a relatively barren size class • Preferential acceptance of a relatively enriched size class For maximum economic impact, waste rejection through pre-concentration is preferred at coarse particle sizes (Klein et al, 2002), either at the naturally arising ROM particle size distribution, or at a coarse crush size, as close as possible to the mining face. There are some examples of this concept applied in practice, as at Kroondal Platinum Mine in the South African Bushveld. Wide-reef mining methods are enabled in the mine as coarse barren pyroxenite is rejected simply by scalping in the stopes prior to the loading of the ore onto the underground strike conveyors. The ore is then further upgraded on surface by a combination of dense media separation and flash flotation1. A further example of ore pre-concentration by crushing and size classification only is referenced for Rio Tinto’s Bougainville Mine in PNG where it was found that the -38.5mm material in the ROM porphyry ore was significantly upgraded in copper values compared to the average (Burns & Grimes, 1986). A crushing plant was installed in order to enhance this, and the enriched fraction was subsequently screened out and sent to the flotation 1 http://www.aquariusplatinum.com/aquarius_db/pdfs/Kroondal.pdf 34 plant for processing. Similar characteristics were observed during testwork on ores of the Mississippi Bonaterre formation near Viburnum in Missouri (Jones, 2006). A significant upgrading of the fine fraction in the Buick and Viburnum ores can be seen from the size assay data. Concentration based purely on size is indicated in the preliminary assessment on both of these ores. 220.127.116.11 Testwork at UBC During the research, it was observed that comminution may further improve the liberation of the ore as well as alter the distribution of metal values in the ore more favourably (Bamber et al 2007). Coarse particle sizes are desired thus comminution by gyratory crushing, jaw crushing or autogenous grinding was considered. 500kg stope samples were taken from the McCreedy East orebody and subjected to a series of tests. In the first phase of work, size assay of the contact- type ore sample indicated the presence of a coarse barren fraction, but little natural upgrading of the fines fraction. The +125mm fraction was weighed and assayed to assess the impact of removing this fraction by scalping. Waste rejection was 10% with a metal loss of 2% indicating some potential in this regard (Bamber et al, 2006). In the Footwall type ore, the high Wt%, as well as the high Cu grade of the fines indicated that a large portion of the ore value was represented in this fraction and the potential for mass upgrading of the Footwall ore through grinding and classification only was thus investigated (Mayne, 2005; Simonian, 2005). A 20kg full fraction sample of the 153-4550-2 sample, grading 13.2% Cu and 0.22% Ni was taken and dry ground autogenously at over 15min sequential intervals in a 3kW, 1m x 900mmØ tumbling mill. Products of each grinding campaign were screened at 19mm, weighed and assayed. Significant upgrading of the -19mm fines fraction was observed after the first grinding interval. Results are shown in Figure 3.5. 35 - 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0 1 2 3 Grinding Cycles % Cum Wt% Distr Cu% Distr Ni% Cu% Ni% Figure 3.5 –Results from Autogenous Grinding and Classification of 153 ore -19mm Fraction The -19mm undersize material was significantly upgraded compared to the feed sample over each stage of comminution, and was designated as concentrate in each case. A final concentrate comprising 63% of the mass of the sample grading 21.6% Cu and 0.28% Ni was produced at a metallurgical recovery of 99% for Cu and 83% for Ni. In this ore the nickel feed grade is low, and the pentlandite is transitional from the Contact ore down dip towards the Footwall (Naldrett, 1984), and typically remains mineralogically associated with silica in the Sudbury breccias and not the massive sulphides, and consequently metal recoveries are low. Overall the results indicate that 37% of the Footwall type ore could be rejected with a copper recovery of 99% by dry autogenous grinding in closed circuit with a 19mm screen. The results for the ore indicate that the ore is already well liberated at the ROM particle size; subsequent Bond Work Index testing on concentrate and tailings fraction of this ore type indicate that there is a significant discrepancy in the Work Index of the sulphide component compared to the gangue component of the ore (Altun, 2007), and that these two facts in combination indicate potential for this approach in upgrading this type of high-grade, massive-vein sulphide ore. A second campaign of grinding and screening tests were performed on the 153-4550-3 sample grading 8.14% Cu and 0.35% Ni. In this test, assays were taken for all potentially valuable elements in the concentrate and tailings fractions of the ore in order to confirm metallurgical results across all metals. The results indicate a mass rejection of 35% overall to a concentrate grading 11.71% Cu, 0.29 Ni and 27.43 g/t total precious metals (TPM). Recoveries of copper, 36 platinum and palladium to the -19mm concentrate are good, and indicate that these metals are all closely associated in the narrow vein sulphides. Nickel recoveries are also much improved over the first test with an overall recovery to concentrate of 92.3%. Ni and Au recovery is poorer indicating that these may not be associated with the massive vein Cu sulphides. The resultant grade distribution by size indicates a significant upgrading of metal values in the fines component of the ore after grinding and classification (Figure 3.6). Figure 3.6 – Grade Distribution by Size in 153 ore after 60 minutes of grinding Comminution and size classification is the most compact, and lowest cost, of all mineral separation processes, and the results indicate that this methods has great potential for the pre- concentration of this type of high grade massive sulphide ore. A proposed flowsheet for the adoption of this approach is shown in Figure 3.7. 37 Feed bin SAG Mill Scalping Screen RejectAccept Figure 3.7 – Concentration by comminution and classification 3.2.2 Pre-concentration by Sorting 18.104.22.168. Sorting practices in the mineral industry Sorting has been practiced in the mining industry for as long as ores have been mined. Native iron and copper ores have been mined and hand sorted since pre-Roman times. Cornish tin ores were traditionally hand sorted into various products, as were copper ores mined near Clausthal in Germany in the 1800’s. Hand sorting as a means of pre-concentration the ore underground prior to hoisting has also been practiced historically – Agricola also noted the use of hand sorting in mines of Europe in the 16th century, encouraged by the mine owners in order to improve the economics of mining (Agricola, 1556). Automated sorting practices are more recent, with modern electronic sorters having been applied in the industry only since the Second World War. Pre-concentration by sorting can be used to improve the grade of large, low grade heterogeneous deposits by rejecting coarse, barren waste, thus significantly decreasing milling costs (de Jong, 2005). Radiometric sorting of uranium ores is the oldest known application of the technology at Port Radium and Port Hope in Ontario, Canada in 1958 (Collins & Bonney, 1995). Radiometric sorting was widely used in South Africa in the 1980s as a substitute for hand sorting of low-grade gold ores. Diamonds are probably the widest known application of sorting, in particular x-ray fluorescence, but literature on the technology is scarce. There are few examples of the application of sorting at a large scale in the metal mining industry, most recently the installation of an Ultrasort conductivity sorter at Kambalda in Western Australia (Goode, 2006), and the installation of a CommoDas Mikrosort Primary optical sorter at Amplats 38 Rustenburg UG2 Section (Kinver, 2002). Competing technologies to sorting are selective mining methods and gravity concentration; it has been suggested previously that pre- concentration is preferable both technically and economically to selective mining (Bamber et al, 2005) and that sorting as a simple low cost, dry coarse particle separation technology would be preferable to gravity methods as they typically require water, thus investigation and testing of selected sorting methods was considered crucial for this research. The development of electronic sorting technology has largely been driven by the recycling industry and while sorting, and in particular photometric sorting, has found some application in the industrial minerals industry (Wotruba & Junsgt, 2000) the technology has not found as wide application in the hard rock mining industry due to several common misconceptions about the application and benefits of the technology (Salter & Wyatt, 1991). Despite the availability of a wide variety of sensing methods, and the low capital and operating cost of the technology, there are many challenges to the acceptance of sorting in the minerals industry, including a lack of understanding of the basic discrimination principles, opportunities for and applications of sorting; perceptions of high capital costs, high unit operating costs, and poor reliability for the technology; a perception that sorting incurs an unacceptable loss of metal; and the traditionally low capacity of sorters when compared to present throughputs of grinding and flotation plant. The need for specific feed preparation for sorting appears, wrongly, to be seen as a challenge, as other process technologies, for example DMS and heap leaching also require equally rigorous feed preparation in the form of feed sizing and fines removal. The main sorting technologies and their applications are presented in Table 3.2 (Salter & Wyatt, 1991). Table 3.2 – Methods of Discrimination in Sorting Method Application Photometric Coal, sulphides, phosphates, oxides Radiometric Uranium, Witwatersrand gold ores Conductivity Metal sulphides, native metals Fluorescence Metal sulphides, limestone, iron ore X-ray luminescence Diamonds X-ray transmission Coal Electrostatic Salts, halite, slyvite Magnetic Iron ore, andalusite, quartz, kimberlites Regardless of the principle on which the sorter is based, sorters share several common design features. Feed to the sorter must generally be closely sized, the particles are typically 39 individually sensed and ejected, except in cases of whole-ore diversion, thus the feed from the hopper is typically accelerated by the sorting belt to between 3 – 6 m/s, and passed through the sensors. Belt widths vary between 1.2 to 2m for high capacity sorters. Under-belt sensors are used in conductivity and X-ray transmission applications, or in combined applications such as optical/conductivity sorting. The effectiveness of the process is often enhanced when used in combination, for example photometric + radiometric or photometric + conductivity. Sensor signals are passed to the micro-controller and compared to some previously determined threshold value in order to make a sorting decision. All sorters have one of several kinds of particle ejection mechanism, which can be physical (flappers), pneumatic (high pressure air valves), or even magnetic fields in the case of magnetic separators (Figure 3.8). Figure 3.8 – Features of Sensor Based Sorters The maximum throughput of the sorter is typically limited by the speed of signal processing, analysis and comparison that the microprocessor can achieve, thus the maximum number of particles the sorter can handle in a given time is a constant. This is limited to between 10 – 15000 particles per hour for modern sorters2. For optimum performance, the feed size to the sorter should be limited in range to approximately 4:1 (topsize : bottomsize) in order to control the particle count at the sensor within this range. Throughput is thus dependent on the nominal particle size and density of the feed to the sorter (Table 3.3). Recent developments with Ore Sorters Model 13 and 16 optical sorters indicate good sorting results on 20mm particles at throughputs up to 75tph and at belt speeds up to 4m/s (Arvidson & Reynolds, 1995). 2 www.mogenson.com/mikrosort.htm 40 Table 3.3 – Typical Sorter Capacities (after Wotruba & Jungst, 2000) Feed size (mm) Min Max Capacity tph 20 60 20 10 200 100 100 350 300 Sorters are also somewhat limited in application compared to gravity separation methods as a practical lower particle size limit of 10mm appears to exist for most discrimination methods. This, however, is perhaps not as limiting as it might first appear as it has been identified in the research that ores considered optimal for pre-concentration possess a significantly upgraded fines fraction, thus sorting of this fraction is mostly considered superfluous. The limitation on the sorter topsize is driven largely by limitations on the type of ejector mechanism used in sorting (air or mechanical diversion), as well as consideration of the typical decrease in the degree of liberation in the large ore particles. The overall limitation on sorter capacity is irrelevant at low tonnages, however this has been identified as a challenge to be overcome in designing a process plant for tonnages over 15 000 tpd due to the unfeasibly large number of units that would be required (Bamber, 2007). However, despite or because of these limitations, sorters are indicated for particular applications in the industry: Bulk treatment of low value commodities such as construction aggregates and industrial minerals Pre-concentration of low grade or highly diluted ore, in-pit or even underground Re-processing of waste dumps Recycling It has also been identified in the course of the case studies as well as the literature that sorting has a particular capability to produce custom material streams. For sorting to be successful there must exist: Sufficiently large disparity in the physical characteristics of the valuable and non- valuable ore fractions Sufficient liberation of the non-valuable fraction at a coarse particle size The existence of a commercial sensor with the capacity to discriminate between the physical characteristics of the valuable and non-valuable fractions 41 The benefits of sorting when applied to base metal ores, as with all methods of pre- concentration, are manifold and significant and the potential for increased application of ore sorting in the industry is thus significant. Several of the ores sampled in the course of the research have demonstrated clear visual discrepancies between the ore and waste fractions (Bamber et al, 2006). Optical sorting had been previously investigated by INCO at a preliminary level on some of these ores with extremely good results (Buksa & Paventi, 2002). INCO Mines Technology conducted electronic sorting tests on synthetic McCreedy ore samples using a combination of optical and conductivity sensing in a Mogenson sorter, which gave results of up to 77% waste rejection at a recovery of 98% (Schindler, 2001). In the investigation of low-grade ultramafic hosted nickel ores of the Thompson nickel belt, the occurrence of chrysotile minerals present in the ore was visually discernible and identified as a basis for the preferential rejection of this fraction prior to grinding and flotation. Several principal sorting methods are currently in use. By far the most common type of sorters are photometric models (Wotruba, 2006). Next most common are radiometric, fluorescent and conductivity sorters. A typical sorting flowsheet for base metal sulphides would comprise the following stages: feed preparation; bypass of the high grade fines; coarse sorting and fine sorting; product and waste handling. Possibly the biggest challenge for sorting is in the discrimination of optically indifferent low-grade sulphide ores as often there is a variety of colour and texture in the ore, thus photometric methods are ineffective, and the discrimination of metal values by conductivity-based methods below 1% is presently considered a challenge, thus a commercially available method for this class of ores has yet to be developed (Sivamohan & Forssberg, 1991; Wotruba, 2006). 22.214.171.124. Radiometric Methods For naturally radioactive ores, the measurement of the radioactivity is possible using a basic scintillometer, and several sorters including the Ore Sorters Mk IVA and Model 17 are presently commercially available using this technology. For this type of sorting it is important to integrate an estimate of the size of the particle with the strength of the signal in order to obtain an accurate value for the ore grade, thus these sorters are often found in combination with photometric sensors in order to overcome this. Separation results are excellent and comparable to separation by gravity methods, however the application is limited as ores with natural radioactivity or associated with naturally radioactive elements, for example Witwatersrand gold 42 ores, are not common. If ore is not naturally radioactive, radioactivity can be induced by neutron bombardment, however this approach is costly and no commercial machines based on this method are known to exist. 126.96.36.199. X-Ray and Laser Methods X-ray sensing, where the particles to be sorted are subjected to X-ray excitation, and the subsequent decay radiation is then measured has the potential for much finer discrimination. Furthermore X-ray has the advantage of penetrating the particle, thus discrimination on a whole- particle basis is thus possible. Excitation can be by radioactive source or, more recently by laser methods, and the overall reading is a function of the permissivity (μ) and the diameter d of the particle, thus correlation of the reading with a visual estimation of the particle size as in radiometric or conductivity methods is not necessary. X-ray methods are widely used in the diamond industry, as diamonds have extremely low permissivity, while transmission methods are common in the measurement of calorific value in coal due to the low permissivity of coal compared to ash fractions (Sivamohan & Forrssberg, 1991). Like X-rays, lasers can also be used to produce breakdown radiation, such as fluorescence or near visible spectra, in ores. Laser breakdown spectroscopy can be used for grade control on individual ore particles. However, the resolution of the laser is often too fine, (<1mm2) and thus cannot accurately determine bulk properties of an ore. To overcome this, laser induced fluorescence can be used, in the evaluation of bulk ore properties, but this capability comes at the expense of accuracy (Kruuka & Briocher, 2002). Both methods do not give an absolute value and the sensor reading must be calibrated by back-assay for every ore to be measured. 188.8.131.52. Optical Sorting Possibly the most common sorting technique in industry is optical sorting. Sensing is generally either by photocell or more recently by digital line-scan camera, and sorting decisions are typically based on differences in colour, reflectance, or transparency between the particles. Sensing frequencies are generally in the visible range, and in the manufacturing and recycling application, where particle sizes are typically larger than the present lower detection limit of 1mm for the technology, incident light is sufficient for most sorting applications. Optical sorters such as the Ore Sorters Model 16, Gunson’s Sortex Model 612M and 811 operate in such principles and are common in these industries as well as the industrial minerals industry (Schapper 1977), and some application has been investigated in the metals mining sector. However, the method presents challenges as optical methods have no penetration and measure 43 superficial ore features only; also, minerals are typically optically complex and variegated, and more sophisticated illumination sources as well as techniques are often required. Amplification of the light source is the first choice, as well as increasing the number of photo-receptors/ cameras. Alternative light sources and sensors such as UV, fluorescent, or near infra-red excitation and sensing may be used when discrepancies in the ore using the visible spectrum are not apparent. Infra red methods are promising and can be enhanced by differential heating of the ore for example via microwave. Fluorescence is promising, however, according to Wotruba (2006), only a small percentage of minerals are amenable to fluorescent detection and the use of high resolution colour cameras for this type of detection is only now in development. More sophisticated are shape and texture recognition methods, however there are currently no sensors available using this method, and while developments are ongoing (Sivamohan & Forssberg 1991), no commercial sorters are currently available for this. In order to evaluate ore potential, colorimetric analysis of the ore is required. In the course of the research, an optical ore evaluation system has been developed for this at UBC, the details of which are presented in Chapter 4. Mineralization in the samples is discriminated by means of Red Green Blue (RGB) / YES image processing and ‘blob’ analysis of the textures using neural network software. The system delivers useful data on correlations between the colorimetric characteristics and grade of the ore. No sensors are currently based on this principle, but applications for the method are foreseen. 184.108.40.206. Conductivity Sorting Conductivity sensors are most commonly simple metal detectors - conductive copper coils excited by an AC current typically of a frequency between 2-200 kHz. Power and sensitivity are in inverse proportion to each other, and sensing power is also in inverse square proportion to the distance from the coil. Conductivity methods are applicable in the case where there is significant difference in the electrical properties for the valuable and non-valuable fractions in the ore (Table 3.4). Results are improved for discrepancies > 1x102. Conductivity methods are thus generally applicable for discriminating native metal ores such as gold and copper, as well as massive sulphide ores grading between 2-3% metal with good results. 44 Table 3.4 – Conductive Properties of Minerals (from Ford, 1986) Mineral Resistivity Silica 3.8 x 1010 – 1.2 x 1012 Amphibole 107 Alumina 107 Wolframite 102 – 105 Hematite 4x104 Sphalerite 105 Pyrite 10-3 Chalcopyrite 1.2 x 10-3 Pentlandite 10-4 Galena 10-3 Testwork reports on copper porphyry ores from a range of mines in the Montana and Upper Michigan area indicate conductivity sorting delivered up to 50% waste rejection by mass from the ores at recoveries from 85 – 92% (Miller et al, 1978). As in radiometric methods, conductivity is highly particle size dependent, thus a large low grade particle gives a similar reading to a small high grade particle. Due to a combination of these issues, conductivity sensors which can accurately and efficiently quantify the metal content of low grade base metal sulphide ores are currently not commercially available (Wotruba, 2006; Sivamohan & Forssberg, 1991). It is suggested that conductivity methods should be utilized in conjunction with optical sensors in order to compensate for some of the drawbacks of the method, with improved results. In previous ore sorting tests at INCO, tests of optical sorting or conductivity sorting alone on combined ROM ore from McCreedy East did not give acceptable results (Schindler, 2001). Sorting tests on the ore using a combination of optical and conductivity sensing in a Mogenson sorter, gave results of up to 77% waste rejection at a recovery of 98% confirming the potential of combined sorting. A variation on the metal-detector type sensor which is indicated for use with lower grade base metal sulphide ores is the induction balance coil (Sivamohan & Forrsberg, 1991). This arrangement gives good results for native ores and is considered to have high potential for conductive sulphides such as such as sphalerite, galena and covellite in VMS orebodies where there is a significant conductivity differential between the ore and gangue minerals. In this method, an inductive coil is excited by a high frequency AC signal, generating an electromagnetic field around the coil. The coil has a natural frequency f due to the unique inductance and capacitance arising from its construction. Inserting a conductive lump of ore into the field changes the inductance of the system and the change in natural frequency is measured. 45 The method is appropriate for native and sulphide ores between 1-3%, however, is inaccurate for low grade disseminated sulphides < 1% and any particle < 1mm. The accuracy of the method is improved by introducing a second balancing coil, and measuring the difference in signal between the disturbed (sensing) and undisturbed (balancing) coil. A conductivity sensor based on this principal has been developed for the characterization of relevant ore properties at UBC, the details of which are presented in Chapter 4. No sensors are presently commercially available and it is believed there is significant future application for the technology. 3.2.3 Pre-concentration by Dense Media Separation Dense media separation is the principal process technology used in surface pre-concentration plants. It is effective in removing coarse waste from a high-grade ore stream at low operating costs and high metallurgical recoveries. DMS can be used to produce a final product, as in coal washing, chromite and iron ore applications, or to prepare feed for downstream flotation or leaching such as on lead-zinc ores at Mt Isa Mines in Australia or for UG2 platinum ores at Kroondal Platinum Mines. DMS has also been used historically to pre-concentrate gold ores of the Witwatersrand prior to cyanidation (Adamson, 1972). DMS plants typically comprise a feed preparation section, the dense media vessels themselves (either HM drums or cyclones), together with related product and discard areas. DMS units are compact, low cost, and high capacity. A typical DMS cyclone arrangement for processing 100tph is shown in Figure 3.9. Note that the overall dimension of the processing module is 10m x 5m x 13m, which is of a size considered appropriate to be housed in an excavation in good rock, thus making DMS attractive in an underground pre-concentration application. However, there are a number of challenges to integrating dense media separation into the underground environment. These include accommodating the required height of the plant, introducing and managing dense media in the underground environment and maintenance of the plant. DMS plants are, however, high capacity, and flexible in terms of feed tonnage and conditions, although variations in process efficiency across the vessel can be experienced when the feed tonnage varies. 46 Figure 3.9 – 100tph DMS Cyclone Module (Drawing courtesy Bateman Projects, South Africa) A focus of the design effort would be in simplifying the media preparation and recovery circuit, both to reduce the size of the plant, but also to minimize media losses to the underground environment, as this would make DMS cost-uncompetitive when evaluated against other technologies. However, several DMS operations, such as Sullivan Mine in BC, have used coarse galena recovered from the flotation section as heavy media (McCullough et al, 1999), which can substantially reduce media costs in the process. It is thus believed that these challenges can be overcome, and benefits in terms of process efficiency and improved mineral recovery would be enjoyed when compared to sorting or gravity concentration in the underground environment. As previously mentioned, there is good potential for this technology underground as evidenced by the many examples of surface pre-concentration plants as well as AMT’s proposal for the installation of DMS underground on a copper porphyry orebody (AMT Annual Report, 1997). Pre-concentration by DMS is presently employed widely in Bushveld precious metal operations such as Kroondal and Marikana Platinum, where due to the narrow (<800mm) and partitioned nature of the UG2 chrome- and platinum bearing seams, up to 30% waste can be included in the mining cut, which decreases the average ROM grade to below the economic cutoff grade in many cases. Pre-concentration typically occurs in three stages: coarse barren pyroxenite waste is removed by scalping underground prior to hoisting. Further oversize is removed on surface by 47 scalping. Crushed UG2 feed reports to the pre-concentration plant where -1mm fines are treated by flash flotation and +1mm -38mm is treated via DMS3. The grade of the ore is increased to above the economic cutoff, and tonnage to the grinding and flotation stages is substantially reduced, leading to lower surface plant capital and improved PGM recoveries overall. Several other operations using pre-concentration have identified using DMS including Impala Platinum’s Morula and some Lonmin operations such as Karee. The most recent cited example of pre-concentration by DMS would be the plant built for Lion- Ore at Tati Nickel in Botswana: Phase I is a 1.7Mtpa DMS plant treating low grade base metal sulphide ore, and a Phase II study just completed is for the installation of a multi-stream, 12Mtpa DMS plant at a cost of ~US$70m (Lion Ore, 2006). Potential also exists for the pre- concentration of other ores such as copper porphyrys; although no known porphyry pre- concentration plants are in operation, testwork in this regard is ongoing. Bench and pilot scale metallurgical work has been done previously for AMT Copper in Arizona at Mountain States R&D, demonstrating that, in the case of relatively coarse copper porphyry mineralization, pre- concentration by DMS was effective in improving the grade of the copper ore at an acceptable recovery (AMT, 1990). Separation is typically by static baths, such as the Wemco Drum, and although high capacity DMS cyclones are a possibility, drums are unlikely to be replaced entirely by cyclones in the near future. However, there are alternatives to DMS by drum or cyclone that have been developed that must be considered. These include the development of a new centrifugal separator by British Coal, the LARCODEMS , which has expanded the size range treated in dynamic DMS from 38mm (cyclone) too 100mm. A further alternative to the cyclone is the Tri- flo Separator which allows separation to occur at two densities allowing for either a scavenging or cleaning effect without the need for a second washing plant. However, the Tri-flo, like the Dynawhirlpool on which it is based, is prone to short circuiting of sinks to floats and further development is required for applications on high value ores. Feed size to the Dynawhirlpool/Tri- flo is similar to that of cyclone, thus careful evaluation of the performance of each option in the application is recommended. Typical DMS process designs are compact, high-throughput processes operating on tonnages between 300 – 1000 tph. The application of DMS as a pre-concentration technology is not 3 http://www.aquariusplatinum.com/aquarius_db/pdfs/Kroondal.pdf 48 considered research in itself. However, the technology has huge potential in the underground application, and several studies have been conducted in order to quantify and evaluate this potential (Bamber, 2004; 2006; 2007). The metallurgical and geotechnical results from these studies are excellent and indicate a wide application for the technology; however, the results do not preclude the use of any of the available DMS technologies, and the most appropriate separator must be chosen for the application in each case. 3.2.4 Coarse-particle flotation techniques for the pre-concentration of base- metal ores 220.127.116.11. Development of the Concept Flotation is considered to have great potential for the pre-concentration of base as well as precious metal ores on surface as well as underground. Conventional flotation is typically undertaken at particle sizes between 10 and 200 μm in order to optimize recoveries, although conventional flotation has been utilized at particle sizes up to 600μm (Kallioinen & Niiti, 1997). Achieving these feed particle sizes can consume a substantial amount of energy as the particle size distribution of the ROM ore depends principally on the geotechnical characteristics of the rock, the blast design and the mining method employed (Laing, 2002), and can range from >1m for open pit or caving type operations, slightly finer in underground open stoping methods to approximately 300mm for mechanized underground methods. The particle size distribution is reduced in order to improve the liberation of the ore, creating sufficient exposed mineral surfaces to enable efficient flotation or leaching. As has been previously discussed, the dilution in the ROM ore is typically siliceous gangue, harder than the valuable metal bearing sulphides, and thus grinding of the ore to these fine particle sizes requires a significant amount of energy, up to 50% of the total milling energy required (Pitt & Wadsworth, 1989). Fine particle flotation has been used in some Witwatersrand gold operations to pre-concentrate low grade ores prior to leaching, and in this light can be considered a pre-concentration technology in its own right (Adamson, 1972). However, conventional flotation of sulphide ores usually entails the grinding of the ore to extremely fine particle sizes, typically -149µm for primary flotation, -75µm for secondary floatation, and more recently even as fine as -30µm in tertiary regrind circuits, and is thus complex and costly, and most often is employed to produce a final concentrate and is thus not considered as a suitable approach for ore pre-concentration for the purposes of this thesis. The flotation of sulphide ores at a topsize of between 1 and 3mm is suggested as a method of pre-concentration, which, if commercially viable would be a 49 massive improvement in terms of size and cost over the conventional approach, specifically in the underground application. Traditionally the term coarse particle flotation has been used to indicate approaches for improved recovery of coarser particles of between 300 – 500µm in conventional flotation. New developments in coarse-particle flotation have demonstrated that good recoveries can be obtained at particle sizes up to 3mm in the flotation of apatite (Hui & Achmed, 1998; Leppinen et al, 2003). Such coarse separation sizes are also common in flotation practice in the potash industry. Separation in Froth (SIF) is considered effective with coarse particles but is not an efficient process; conventional and flash flotation is efficient, but mass pull to concentrates is limited. However the literature indicates that even coarser particles can be separated by means of froth flotation - in the flotation of minerals such as potash, barites and phosphates feed topsize can be as high as 3-5mm, although unconventional low-capacity techniques such as Separation in Froth and Froth Flotation, where the feed is passed by gravity through a stable, previously established froth layer, are typically used in these instances. Furthermore, the grade and degree of liberation in these ores is typically high and the relative density is low, which increases the potential of successful separation by floatation at these particle sizes. However, the literature does indicate that coarse particle flotation of gold ores may be possible. The pre-concentration of Witwatersrand gold ores by means of coarse particle flotation followed by gravity concentration of the flotation tailings was researched by the South African Chamber of Mines Research Organization (COMRO) (Lloyd, 1978, 1979; Lloyd et al, 1986). Flotation of the ore at a topsize of 3mm, followed by gravity concentration of the flotation tails resulted in the production of a 40% by mass concentrate at an overall Au recovery of 98% (Lloyd, 1979). The testwork was undertaken in support of strategic research into underground pre-concentration and waste products were to be utilized as fine aggregate for backfill. Great potential for this technology in the underground application is thus indicated. In designing flotation cells for the underground application, the liberation characteristics of the ore are critical in developing the machine, thus established mineralogical techniques must be used to determine the maximum particle size distribution at which flotation can occur for the ore thus determining the size requirements for preparation of the ore as feed by means of grinding (Morizot et al, 1991; McIvor & Finch, 1991). For the ores tested in the course of the research, good liberation of the sulphides at coarse particle sizes, and an α-log normal grade distribution has been observed, which appears to maximize the chances of success of pre-concentration. It is posited that the presence of these characteristics would also apply to improve the potential of a 50 separation method based on surface chemistry as well, such as flotation, as opposed to physical means such as density or colour discrimination, in order to efficiently separate the gangue from the sulphides at coarse particle sizes. In addition to establishing good liberation and flotation conditions, appropriate hydrodynamic conditions must be established for the flotation of base metal ores at the particle sizes under consideration. A flotation arrangement comprising preparation of a coarse (-3mm) slurry feed, by HPGR or SAG milling, introduced high in the cell, with high specific energy and air inputs and a high degree of froth removal is envisaged as a starting point. Such an arrangement would be compact and high capacity and integrate well into the underground environment. It is suggested that if a more sophisticated flotation arrangement is required that the process should then be undertaken conventionally on surface. 18.104.22.168. Theoretical Basis A significant increase in the feed particle size to the flotation cell would require a high energy density for the suspension of the slurry in the cell. As ore grades decrease and the mineralogy of the ore moves from massive through less massive to disseminated, grind sizes have decreased, tonnages have increased and the size of grinding and flotation plants have generally increased. Modern tank flotation cells have become continually larger; the maximum presently known cell sizes, as installed at the new Potgietersrust Platinum Concentrator in the Bushveld of South Africa, are of the order of 160m3. Cells of 200m3 and larger are also being considered for other future operations4. However, grinds in flotation are now typically finer, and the size of the drive has not increased in proportion with tank sizes, and these large tank cells, with energy densities of 1.2-1.6kW/m3 are typically lower in power intensity than the preceding Wemco- or Denver type flotation cell designs with energy densities of approximately 3.6 kW/m3. In order to overcome this limitation, flash flotation devices such as the Skimair, with smaller tank sizes, oversized drive units (thus increased energy density) and improved froth launder designs have recently been introduced to improve performance particularly in the coarse-particle flotation application (Table 3.5). However, the energy density of flash flotation cells is still low compared to the potential maximum, and the coarsest known application of flash flotation is for feed sizes of -1mm (Bushveld Platinum) or up to 2mm in cyclone underflow applications in gold milling circuits (Lloyd, 1981). Alternatives to flash flotation cells such as the Aeromix 4 www.outotec.com 51 ‘Flo-triter’ can be used in applications where the ore must be further cleaned during the flotation process5, however, these are small machines with no rougher flotation applications. Table 3.5 – Energy density comparison for commercially flotation cells Machine Size (m3) Installed Power kW Power Draw kW Installed Power kW/m3 Power Draw kW/ m3 SkimAir®-1200 49.00 132.00 74.58 2.69 1.52 SkimAir®-500 23.00 55.00 31.08 2.39 1.35 SkimAir®-240 8.00 22.00 12.43 2.75 1.55 SkimAir®-80 2.20 11.00 6.22 5.00 2.83 SkimAir®-40 1.30 5.50 3.11 4.23 2.39 SkimAir®-15 0.30 2.20 1.24 7.33 4.14 Wemco #6* 6.00 22.00 12.43 3.67 2.07 Wemco #30* 30.00 55.00 31.08 1.83 1.04 OK-50 50.00 110.00 62.15 2.20 1.24 OK-38 38.00 90.00 50.85 2.37 1.34 OK-16 16.00 45.00 25.43 2.81 1.59 OK-8 8.00 37.00 20.91 4.63 2.61 Denver D12 0.03 0.20 0.11 6.67 3.77 From: http://www.outotec.com/20846.epibrw *Dworzanowski et al, 2007 Flotation of base metal ores with particle sizes of greater than 2mm presently appears to be out of the specification of present flotation equipment. The concept can, however, be tested at the laboratory scale, as laboratory and pilot cells are extremely energy intensive, with much higher installed power/unit volume than conventional field units, and higher energy intensity than even flash flotation units. However, there is a limit to the degree of agitation that can be successfully applied in flotation. For the successful recovery of the sulphide particles to the froth launder, a fine balance between the dynamic forces in the cell must be maintained, thus a good understanding and close control of the energy intensity and agitation speed in the cell is required. The hydrodynamics in the cell must be sufficiently vigorous to suspend the largest mineral particle, yet conditions must conversely be still sufficiently quiescent that the hydrodynamic forces do not overcome the electrostatic forces attaching the sulphide minerals to the air bubbles in the froth (Figure 3.10). 5 www.aeromix.com/flo-triter.htm 52 Figure 3.10 – Hydrodynamic Zones in a Flotation Tank Cell For this, the effective agitation intensity must be evaluated, and the concept of Critical Speed is suggested (Equation 1). - (1) Where: Njsg – critical agitation speed Ks, Ka - empirical suspension constant (dependent on tank diameter, impeller diameter and impeller height in the tank) T – tank diameter d – particle diameter in microns B – solids content (%) υ – kinematic viscosity g – gravitational constant 53 ρs – solid SG ρl – liquid SG Qgv – air volume addition in m3/min At the critical agitation speed Njsg, all slurry particles are considered to be in suspension off the bottom of the tank, particles are evenly suspended in the mixing zone, there is no settlement of particles in the separation zone, and the chance of exposure and attachment to air bubbles, and delivery to the froth zone maximized. Below the critical agitation speed, settlement occurs in the separation zone, insufficient particles are suspended in the mixing zone and thus the exposure rate of the mineral particles to the air bubbles is low. The minerals have a lower chance to become attached to the air bubbles. Above the critical speed, the solid particles in the slurry acquire velocities superfluous for suspension and thus the likelihood of remaining attached are substantially decreased. It is thus important in this situation to determine the critical agitation velocity and maintain conditions as close to the critical point as possible in order to achieve this, thus maximizing the potential to recover large, high-grade mineral particles. Using the parameters from the literature, a value for the empirical constant Kg can be determined. Through previous experimentation it was determined that for a flotation cell of diameter 0.5m, with a 0.15m Ø impeller, suspending silica at a particle size of 90μm,with air addition of 1.1m3/min, the critical speed for solid suspension is 782 rpm (van der Westhuizen and Deglon, 2006). For the suspension of sulphide minerals with an SG of 3.1 and a topsize of 3mm, an agitation speed of 1500 – 1750 rpm is indicated from the equation. Subsequent observation of the hydrodynamics in the laboratory cell during the testwork has confirmed the functioning of this phenomenon in practice. 22.214.171.124 Scoping Testwork Results Footwall Ores The Footwall ores of the Sudbury igneous complex were an obvious candidate for evaluation in this context, as they are extremely high grade, with massive mineralization and good liberation. The footwall deposits consist of networks of massive sulphide veins associated with Sudbury breccia and felsic gneiss. On the basis of data acquired from both the Strathcona and Clarabelle mills, Footwall ores appear to be fast-floating ores (Kerr et al, 2003; Verdiel, 2006). A series of trial rougher flotation tests were undertaken on a sample of the 153-4550 ore ground to 100% - 300µm and floated under standard conditions in order to determine the flotation kinetics, and grade recovery characteristics of this type of ore in a pre-concentration application 54 (Nunes & Malkuuz, 2005). Ore was prepared by crushing and grinding by rod mill prior to flotation. Flotation was undertaken in a 3dm3 Denver D12 laboratory flotation cell with conditions for minimum suspension for this cell indicated at an agitation speed of 900rpm. These ores are typically treated at INCO’s Clarabelle Mill in Sudbury, and conditions were selected to match flotation conditions at the mill. Flotation was undertaken under alkaline conditions at 1200 rpm and 5dm3/min air addition, and thus conditions for full suspension of the particles were exceeded in this test. Collector and frother were potassium ethyl zanthate at 90g/t and Dowfroth 250 at 20 g/t respectively with 3 min. of collector conditioning plus 1 min. of frother conditioning. Concentrates were taken at 2, 7 and 12 minutes during flotation. Four rougher flotation tests were performed; in tests 3 and 4 a fourth concentrate sample was taken after 20 minutes of flotation time in order to evaluate the effect of additional residence time on the flotation kinetics. Flotation results indicate that on average mass pull to concentrates was 50%, with metal recoveries of 97.4% (Cu), 89.13% (Ni), 91.8% (Pt), 95% (Pd) and 82% (Au) after an average flotation time of 16 minutes. Nickel recovery was poor due to the association of the pentlandite with silica at fine particle sizes in this ore. Gold recovery is also lower than other minerals, however comparable recovery was achieved using gravity methods, indicating the possible presence of gold as native gold in the sulphides. The results indicate that rougher flotation at a feed size of 100% -300µm is possible with good metallurgical results on this ore. The results compare well to results from previous pre-concentration testwork on this type of ore using sorting as well as dense media separation methods (Bamber, 2004, Weatherwax, 2006). Furthermore recoveries are superior to the reported values for the mill (Kerr et al, 2006; Holmes et al, 2000), and it would be recommended to consider adopting a custom flotation circuit for the treatment of these ores at the mill. However, flotation of the ore even at these particle sizes still requires substantial feed preparation in terms of grinding, and the results do not appear conducive for consideration of this arrangement for the underground application, and thus it was determined to do further work at coarse particle sizes in order to investigate the minimum comminution/flotation arrangement which would give acceptable metallurgical results. As it has been determined that the 153 ore is unusual among the Sudbury ores in its extremely high grade and high degree of liberation, it was chosen to continue the work on low grade, and less well liberated ores. 55 Matrix/Breccia Ores The Lake-Granite-Breccia (LGBX) ores of Xstrata’s Craig Mine in Sudbury are pentlandite- and chalcopyrite rich transitional ores of the Sudbury Igneous Complex, situated down-dip between the principal pentlandite-rich Contact ores and the Footwall ores (Binney, 2007). Compared to other contact type ores, LGBX ores exhibit a high sulphide content and medium tenor, with massive- to matrix type sulphide mineralization, and were thus considered to be good candidates for coarse particle flotation trials. 12 flotation samples were prepared in order to investigate the performance of these ores during rougher flotation (Table 10). Flotation was performed at nominally 30% solids using a 0.2kW Denver D12 laboratory unit with a 6dm3 cell. In order to standardize flotation conditions as much as possible, excess reagent additions were planned, and kept constant over the course of the tests. KaX dosage was increased to 200g/t feed from the previous testwork, and frother addition was via Dowfroth 200 at an initial dosage of 300 g/t feed, with additional frother added towards the end of the test in order to maintain froth stability. Samples were crushed to nominally -8mm prior to testing. Sample feed size distribution was modified for selected samples by rod milling at 50% solids for specified intervals of 5, 10 and 15 minutes respectively. The various feed size distributions of the samples used in flotation testing are shown in Figure 10. The topsize of the samples varied between 8mm (CL1,2,10), 5.6mm (CL4,5,6,7,11,12) and 4.76mm (CL9,13) with a feed D50 of 2.4mm, 1.5 and 0.42mm respectively. A critical agitation speed of 1750 rpm was indicated by Equation 1, however this was found to be impractical in the D12 cell, and a standard of 1500 rpm was adopted. CL2 was attempted at an agitation speed of 1200 rpm, but results were poor due to complete saltation of the flotation sample on the bottom of the cell. Flotation tests were undertaken with 3 minutes of collector conditioning and 1 minute of frother conditioning prior to feeding the samples into the cell. Concentrate was collected continuously to avoid froth choking, and flotation of each sample was sustained up to 20 minutes or until the froth loading was negligible. Results ranged from the recovery of 6.26% by mass to concentrate with 36% metal recovery in sample CL4 to 12% recovery by mass with 55% and 58% metal recovery to concs for Cu and Ni respectively for sample CL11. Concentrate recovery in the coarse size fractions was excellent, with particles of up to 1mm being recovered in the CL13 test. The recovery of concentrate by size was also observed to be in inverse ratio to the feed size of the sample, thus the d80 of the concentrate increases from 0.12mm, to 0.15 and ultimately 0.18mm for feed samples with a d80 of 4, 3 and 2.5mm respectively (Figure 3.11). 56 0 20 40 60 80 100 0.0 0.1 1.0Size (mm) W t% CL 10 Conc CL 11 Conc CL 12 Conc CL 13 Conc Figure 3.11 –Comparative Size Distributions – Flotation Concentrate The results support the observation that recovery of coarse particles (between 0.6 and 1mm) to concentrate is improving in each test with finer grind. Good potential is thus indicated for coarse particle flotation of base metal ores, particularly in the pre-concentration application. 3.3. Waste Disposal Technologies 3.3.1 Background A challenge in the consideration of this approach is the appropriate disposal of the waste products. The waste products from a number of case studies were available for this stage of the research, and it was decided to investigate several key aspects of waste disposal in order to identify a suitable technology by which to dispose of the waste. As it was desired to maximize the degree of waste rejected from the ROM ore and disposed of as fill, the use of classified tailings from the surface mill as a source of the fines component for the mix was also considered essential. The science of backfill is largely empirical, and rigorous treatment of the topic has only recently been undertaken (Hassani et al (Eds) 1989; Potvin et al (Eds), 2005). For the purposes of this discussion the following principal types of backfill will be defined (after Grice, 1989): o Rockfill (RF, CRF) o Hydraulic (classified tailings) fill (CHF) o Paste fill o Composite fills (e.g. ‘rocky’ paste fill) 57 Fills as described above can be both cemented (using either cement or pozzolanic binders) or uncemented. Mines utilize such fill for several reasons: o Localised roof support o Long term regional geotechnical stability o Limiting excavation exposure o Disposal of mining waste underground In addition to these benefits, there are a number of secondary benefits of fill that have been documented. These include a 45% reduction in post-mining closure, and a reduction in the Energy Release Rate (ERR) during rockbursts to below the critical value (Kamp, 1989). Kamp also notes increased ore extraction through a reduction in pillar size, a 45% increase in ventilation efficiency, improved fire control and improved safety and productivity leading to reduced unit working costs. The use of fill also leads to a reduction in the heat transferred to ventilation air through a reduction in the surface area available for conduction, as well as reduced air leakage past filled areas (Matthews, 1989). Cemented backfill has been successfully substituted entirely for engineered support in deep-level stopes at a competitive direct cost, and has been shown to substantially reduce rock stresses, plastic rock deformation and rock bursts when utilized at depths below 2000m (Patchett, 1977; Lloyd 1979). Sources of backfill material are numerous. Coarse development waste as well as mining waste that can be clearly identified for resuing is utilized as uncemented rockfill, however this type of fill does not impart a significant degree of local or regional roof support. Cemented rockfill (CRF) comprises coarse fill which is consolidated by the application of a dilute sand/cement spray for the generation of additional compressive strength. Falconbridge’s Strathcona Mine in Sudbury employs CRF comprising -125mm rockfill combined with 6.5% water and 6.5% Normal Portland Cement (NPC), although other fill systems have been considered in order to further alleviate rockbursts (Swan et al, 1993). Surface mill tailings is also used as a backfill material, and is typically combined with 2- 5% NPC by mass, generally to reduce slumping and improve pumpability rather than for any strength requirements; NPC can easily be replaced for this purpose by fly-ash, granulated blast furnace cement or some other pozzolanic material. This fill is typically delivered to the stope by means of gravity lines, or pumped to the stope, either from a surface facility, or less commonly from a backfill plant located underground. Mill tailings is not an ideal backfill material as the particle size distribution is too narrow, and contains too little fines (<25% < 10µm), in addition to a paucity of coarse material; fill strength 58 development is thus severely limited and in order to generate compressive strengths > 1 MPa a substantially higher ratio of cement addition is required, which is usually uneconomical. In addition to this, the ratio of water in the fill can be as high as 55%, leading to slumping and drainage problems in the filled stopes (Blight, 1979). The size distribution of mill tailings is often modified through hydrocyclone classification to improve the fill characteristics (classified tailings fill). Water content in hydraulic tailings (classified or unclassified) is high, and drainage of excess water post fill is often a challenge. These problems have been largely overcome through the development of high-density ‘paste’ backfill systems, where a superior fill is prepared with a broader particle size distribution and a decreased water content of between 10 and 25% (Brackebusch, 1994). A dense paste is produced from mill tailings and pumped or gravitated underground. Moderate quantities of coarser aggregates up to 25mm can also be added to the dense paste without significantly impairing its pumpability (Cooke et al, 1992). Mill tailings are dewatered in a conventional thickener in order to preserve the ultrafine fraction, and then filtered to approximately 13% moisture. The particle size distribution of the tailings can be modified by classification, and by the introduction of significant amounts of sand to the mix. The paste, typically comprising the tailings with 15% water, 2-4% cement and some additional aggregates, is made up on surface and delivered to the stopes by a combination of concrete pumps and gravity pipes of between 100 – 150mm diameter. The maximum realistic strength paste fills can achieve in situ is 1- 3MPa regardless of the degree of cement addition; fill strength development is limited by the narrow size distribution of the tailings, and the addition of a high proportion of cement is generally uneconomic in most backfill situations. Some movement towards the design of fill using more conventional concrete-type mixes have been made in order to maximize potential strength of the fill. Target Gold Mine in the Free State of South Africa has been experimenting with an underground concrete batching plant fed with aggregate prepared from crushed mining waste (SA Mining News, 1995). A high strength of fill is generated which is used for construction purposes underground. Typical concrete mixtures comprise a mixture of -19mm crushed aggregate, -1mm sand, cement and water in a pumpable mixture of 6:2:1:1 by mass, and can achieve compressive strengths of typically 20 – 40 MPa. Concrete strength varies with the overall particle size distribution in the specification, cement content and water:cement ratio in the mix (Talbot & Richart, 1927). Minefill is not generally required to achieve such strengths, however, and less competent mix ratios and lower cement 59 contents should result in adequate backfill strength. More recent developments include the addition of aggregates to paste or hydraulic fills to create composite fills with higher strength. Composite fills include conventional cemented rock fill (CRF), rock fill with cemented sand fill, and more recently ‘rocky’ paste fills which are typically a blend of paste fill with up to 30% coarse aggregate by mass. While there is by necessity a focus on aggregate-based fills, the consideration of backfill in the context of pre-concentration requires consideration the complete range of fill materials currently in use in order to exploit the opportunity of combining coarse rejects from underground with fine tailings from surface in order to maximize the use of solid mine waste in the fill. The use of high-strength backfills such as cemented aggregate fill or rocky paste fill can only further improve the benefits of utilizing backfill as documented above, and such benefits can outweigh the additional cement cost of these fills (Quesnel et al, 1989). This is borne out in observations by COMRO in South Africa (Adams et al, 1989): residual ore pillars were entirely substituted by 20MPa concrete at a depth of 3800m in an innovative mining system designed for use at extreme depth. Ore extraction was increased by 43% and the ERR was reduced from 50MJ/m2 to below 40MJ/m2. Benefits noted elsewhere include improvements in post-yield pillar stability and a 90% increase in residual pillar strength in sandstone (Yanaguki & Yamatomi, 1989). However, the cost of preparing and delivering these fills is prohibitive, especially for deep and more extensive mines, and it would be advantageous if the economics of producing such high strength fill could be improved. It is thus proposed to consider the use of the waste rejects from the proposed underground pre-concentration process to potentially increase the strength as well as decrease the cost of utilizing such fill. 3.3.2 Development of A ‘Rocky’ Paste Fill For Use With Underground Pre- Concentration Systems Significant economic potential has been identified for pre-concentration underground simply arising from the rejection of large quantities of the barren fraction in the ROM ore directly underground. There is also potential for the application of pre-concentration simply on surface, as this does not preclude the disposal of the waste rejects fill underground. However, in either case it is critical to dispose of these waste products in an appropriate and environmentally sound manner. A suggested use is as an aggregate for fill, should the process deliver a suitably sized product. This seems possible as the naturally arising topsize of the waste products from the processes under consideration is expected to range between 3 and 200mm , which with further 60 crushing and screening can be modified to produce a suitable size distribution for addition to fill as aggregate. The addition of aggregate to hydraulic or paste fills to form a composite fill has several advantages. These include improving the strength of the fill, increasing the binder efficiency and reducing the quantity of water required in the fill mixture (McKinstry & Laukkanen, 1989; Annor et al, 2003). Optimum fill strength is indicated for the addition of 25% by mass of -10mm aggregate (Quesnel et al, 1989). A composite fill mixture with the optimal aggregate size distribution would be dilating by nature during curing (Kuganathan, 2005), which is desirable in a confined environment, and will lead to a higher fill strength in situ than indicated by the ASTM test procedure. Incorrectly graded aggregate results in a contracting fill which can be prone to failure of the fill mass even prior to loading. Strength in composite fill mixtures derives either solely from the cement bond, interlocking of the aggregate particles or a combination of both mechanisms. Fill strength is determined principally by binder content of the fill, however, aggregate fills are expected to generate higher strength than hydraulic fills due to a higher maximum Proctor density due to the decreased coefficient of uniformity and thus higher cohesion of the mix. Aggregate fills are expected to generate higher strength than cemented hydraulic fills or paste fills due to the low coefficient of uniformity and increased natural cohesion of the mixture. Incorrectly graded aggregate results in a contracting fill which can be prone to failure of the fill mass even prior to loading. For hydraulic and paste fills: σmax = 27 [ c ] 1.57 v and for aggregate fills: σmax = 63 [ c ]1.54 (Henderson & Lilly, 2001) v Where: c = cement content in % v = void ratio of the particle size distribution Physical characteristics of the fill material such as UCS, mineralogy and particle geometry also have an additional influence on the final strength of the fill (Lamos & Clark, 1989). A comparison of fill mixtures comprising full spectrum plant tailings, a classified plant tailings and a mixture utilizing –9mm crushed waste are presented for comparison (Figure 3.12). 61 0 1 2 3 4 5 6 0 2 4 6 8 10 12 Cement content (%) U C S (M Pa ) Full Tailings Classified Tailings Crushed Waste Figure 3.12 – Comparison of Backfill Strength by Aggregate Type Note that full tailings in this case exhibited a higher UCS than classified tailings, and that comminuted waste generates a higher strength than either tailings product. A high-strength backfill mix design has also been pioneered at the Cannon Mine in the U.S. (Brechtel et al, 1989). Several aggregate mixes were designed and tested using a coarse (-38mm) and fine (-9mm) aggregate fraction. Fill strength of 9MPa at 6% cement addition was achieved for the coarse aggregate fill in the tests. Also, the confined strength of the resultant composite fill was found to be 25MPa at 27% porosity in situ. Several potential mix designs for cemented backfill using rocky aggregate are thus suggested in the literature. Cemented hydraulic fill would typically be 100% -100um with < 10% -10µm fines combined with 3% OPC and 3% fly ash or similar (Grice, 1989). Various cement ratios for fill are suggested by Udd (1989) in order to optimize cement usage for different applications: o 2.5% - 3.5% for cut-and-fill stopes o 2.5% - 6.25% for bulk pours in blasthole stopes o 12.5% for sill mats and containment structures The use of aggregate in the fill mix is expected to reduce these binder requirements significantly. However, sourcing aggregates for mine fill is not usually economic as this requires special preparation of the non-valuable development waste, or in extreme cases, quarrying of a suitable aggregate on surface. A cost effective source of such aggregates would be through pre- concentration, and Mt Isa Mines in Australia have been a pioneer in the development and use of composite fills using DMS rejects. Pre-concentration by DMS is employed on surface, 62 producing a coarse, rocky waste, which has been used to prepare fill for the underground blasthole stopes . A number of lessons can be drawn from their experience. Typical fill compositions in use at Mt Isa, and the typical fill strengths, are tabulated in 3.6 below (Kuganathan & Shepherd, 2001): Table 3.6 – Mt Isa Backfill Mixes Component Mass in Mix (kg) CHF CRF RF + CHF (3:1) Aggregate 0 2116 1725 Tailings 1365 0 471 Cement 45 46 34.5 Pozzolan 90 138 69 Water 135 184 103.5 UCS (MPa) ~1 5.84 1.77 As can be seen, potential aggregate content varies widely, thus there is a definite potential to incorporate the waste rejects from all the processes under consideration in the study into an appropriately designed backfill for improved fill performance. Mt Isa has been innovative in the use of these waste rejects for the production of composite fills for difficult applications underground, such as steeply dipping blasthole stopes, as well as the replacement of sill mats and pillars. The open-stoping method at Mt Isa requires a high-performance fill with free- standing ability for rapid re-entry and high strength for good regional support. Fill designs based on these parameters are recommended for consideration in the context of the recommended approach in this thesis. 3.3.3 Fill Preparation and Delivery Systems The fill preparation system at Mt Isa is a good model for an appropriate system in terms of the research, (McKinstry & Laukannen 1989) . Approximately 2Mtpa –70mm aggregate is sourced annually from the reject stockpile of the DMS plant and combined with cemented hydraulic fill in a typical ratio of 1:4. The composite fill is made up on surface and gravitated 220m underground via 300mmØ fill raises. A schematic of the ‘rocky’ pastefill preparation and delivery system at Mt.Isa is shown in Figure 3.13. 63 Figure 3.13 – Rocky Paste Fill Backfill System at Mount Isa Mines Two methods of creating the composite ‘rocky’ paste fill have been researched in the course of the work. The first method requires the preparation of a conventional cemented paste fill, and the subsequent mixing of the rejects with the paste underground prior to placement. The second approach requires the simultaneous preparation of fill using all fill components, which is preferred for maximum dissemination of the binder onto the aggregates in the fill. The design of a conventional paste fill preparation system is presented in Reschke (2000). Metallurgical tailings are classified and the undersize fraction is delivered to the tailings dam. Suitably classified tailings are passed to the high rate, high density mechanical thickener through the addition of suitable flocculants. The thickened paste is passed to a high shear mixer where the paste is combined with typically 5% cement slurry. Cemented pastefill is pumped to the pug mixer where the aggregate is added and mixed. Fill can be pumped to the stopes or hauled and placed by means of push cars if not suitably viscous. However, this means of preparing the composite fill is considered ineffective as it is not possible to guarantee adequate coverage of the aggregate surface area with cemented paste, and more sophisticated batching and mixing methods are recommended in order to achieve a consistent, high strength composite fill. A suggested approach is to adopt batching and mixing technologies from the construction industry, 64 where a range of concrete mixes can be efficiently batched and prepared using generic batching plant technology (Figure 3.14). . Figure 3.14 –Composite Fill Preparation System In this method, the fill is prepared in the same manner as a ready mix concrete with positive impacts on fill quality and fill quality control. An infinite variation of mixes can be provided by such a plant depending on the application. Batching plants are compact and high capacity and would be easily accommodated in suitable excavations in the underground scenario. Backfill is typically made up on surface and delivered to the stopes via a suitable delivery and emplacement system, although several mines have been experimenting with underground batching plants (SA Mining News, 1995; Kuganathan & Shepherd, 2001). Emplacement systems for backfill include gravity-based systems, pumped systems, conveying or batch haulage by LHD. Coarse rockfill systems may even require physical handling of the fill. Backfill emplacement can be direct to stopes via dedicated pipe ranges, or placed in 65 intermediate storage dams underground prior to final placement (Kamp, 1989). Placement can be unconstrained if stope geometry permits, via constructed paddocks or in pre-filled geotextile bags, although the latter method is unwieldy and ineffective in delivering good post-fill support. Pumped systems are common on hydraulic, paste- and sandfill systems. Pumped fill systems are less common on coarse backfill applications, although positive displacement pumps are used extensively in the concrete industry, and show good potential for use in the underground fill batching scenario. A pumpable composite fill mixture is preferred as this fill has superior material handling properties over rockfill. Composite fill thus batched can be pumped to the fill stopes using heavy-duty concrete pumps. A schematic of a typical positive displacement concrete pump is shown in Figure 3.15. Such pumps are capable of delivering -38mm aggregate concrete mix at 52 m3/hr (100 tph @ 50% solids) over 400m horizontal, and 240m vertical distance. Figure 3.15 – Thomas Katts BS 907A Concrete pump schematic (Bamber, 2004) There are thus several potential backfill strategies arising from the integration of a pre- concentration facility into the underground mining environment which are attractive. 3.4. Interfacing the Technologies with the Mining Activity The integration of these pre-concentration and waste disposal technologies with the mining activity is expected to impact significantly on the economics of mining and processing and require some changes to the design, layout and planning of production stopes, backfill arrangements and other supporting infrastructure. The adoption of the integrated mining and processing approach is thus expected to impact significantly on the future design and planning 66 of mines. The major interface with the mining method concerns the physical accommodation of the equipment required to perform the pre-concentration and waste disposal step, as well as the means of accommodating the waste material generated in pre-concentration in the stopes. There are an almost infinite number of variations of mining method presently employed in the hard rock mining industry. For the purposes of the thesis, methods will be split into 2 groups: mining methods which depend on backfill, non-backfilling methods which can accommodate fill. Methods which cannot accommodate fill at all such as sublevel caving are not considered. The integration of the envisaged processing and waste disposal steps with the mining activity will thus be discussed for the following methods: • Resue Methods • Mining methods which depend on fill o Overhand Cut & Fill (incl. drift & fill, cut & fill, post-pillar cut & fill) o Underhand Cut & Fill • Mining methods which can accommodate fill o Open Pit methods (incl. open cast and strip mining) o Block Caving o Open Stoping methods (incl. shrinkage stoping / blasthole and longhole) o Room & Pillar The integration of automated pre-concentration and waste disposal systems is thus considered for both surface and underground methods. For deep deposits, integration of the pre- concentration and waste disposal step underground is recommended, necessitating additional stable excavations for ore storage, the pre-concentration plant as well as the storage for waste rejects and the fill preparation and delivery system. Pre-concentration and waste disposal plant must thus be compact, efficient and high-capacity. Pre-concentration should be undertaken as early as possible in the mining cycle and at as coarse a particle size as possible (Klein et al, 2002). The process design, equipment selection and block layout for two of the potential pre- concentration processes has been previously undertaken for an 1800 tpd underground facility. This plant design has been used to develop an estimate of the excavation size required for the individual process areas making up the system (Figure 3.16). The plant design requires a series of excavations, linked by standard 5m x 5m haul drifts, each of which require designing for the specific setting of depth, stress and rock conditions. Surge requirements require additional 67 excavations which would be typically found underground such as ore passes and access drifts. The maximum expected excavation size based on these layouts is 12m x 7m x 15m for the process module (Bamber, 2004). Figure 3.16 – 3D Layout of Excavations for DMS- and Sorting Based Pre-concentration System The envisaged size and arrangement of excavations for the underground application is not considered impractical in the context of existing large and deep excavations. Hoek and Brown (1994) present a bibliography of large excavations which can be used for comparison (Table 3.7). From the table it can be seen that excavations of the order of that indicated are both possible and practical at depth. Significant challenges in the consideration of the underground processing scenario lie in the areas of health, safety and environment, as well as in the sourcing of skilled operators for such a facility. Safety and environmental challenges for the underground installation include additional heat, noise and dust generated during ore and waste processing. While the processes considered are generally inert, or reagent free, significant additional services such as water, power and ventilation will be required to maintain process integrity and safety. 68 Table 3.7 – Data for Existing Large and Deep Excavations (from Hoek & Brown, 1994) Location Date Excavation size (m) Depth (m) Rock Type Aura Power Station, Norway 1953 18 x 17 x 123 + 17 x 15 x 95 250 Moderate Gneiss Tarbela Dam, Pakistan 1972 19m Ø tunnels 270 Weak Gabbro Ralu I Power Station, PNG 1975 15 x 24 x 51 200 Marble / diorite Vaal Reefs sub shaft hoist, RSA 1966 10 x 12 x 13.5 1700 Strong Quartzite Western Deep Levels sub shaft hoist, RSA 1974 16.6 x 12 x 32 2750 Strong Quartzite Stonfors Power Station, Sweden 1958 18.5 x 24 x 124 200 Strong Gneiss Hartebeestfontein 6# Refrigeration Plant* 1986 16 x 12 x 30 2000 Strong Quartzite McCreedy East Crusher Station* n/a 20 x 30 x 20 1100 Strong Norite Nevada Test Site, USA 1965 24 x 43 x 37 400 Moderate Sandstone Technology transfer and particularly the transfer of technology and skills to the underground scenario is seen as a serious challenge to be overcome. Evidence of chronic cultural resistance to the adoption of pre-concentration in general can be noted in the literature (Wotruba, 2006; Salter & Wyatt, 1991) and resistance to the concept of underground pre-concentration in particular has been continuously observed during the fieldwork stage of the research. Resistance of mine operators to the concept of introducing mineral processes into the underground environment was noted as well as an expressed reluctance on the part of mineral processors to consider transferring their operations to the underground environment (Verdiel, 2006). It is recognized that the underground environment can be highly aggressive towards both personnel and equipment and that substantial work is needed in these areas to address this if the approach is to be successful. For shallower deposits, the pre-concentration and waste preparation step can be either on surface or underground, however for maximum positive impact, the rejects should be disposed 69 of underground. For open pit methods, the pre-concentration and waste preparation step are situated on surface, obviating the need for excavations, and solid waste is by necessity disposed of either on surface or in the pit. In integration with surface methods, there is no requirement for process excavations, thus there is no limit on the size or location of the pre-concentration or waste disposal facility with these methods. 3.4.1 Cut-and-Fill Mining Both overhand and underhand cut and fill methods will be considered in this section. In cut and fill methods a small tranche of the overall orebody is mined using mechanized methods and then backfilled with either cemented or uncemented fill prior to the removal of the next tranche of ore. The method is most commonly applied in narrow, moderate to steep dipping orebodies with weak wall rock where high recovery as well as high selectivity is required. Disadvantages of the method include the typical high cost of using fill, challenges in scheduling fill and typical high overall cost and low productivity. Also, in typical wall rock conditions, the method requires the retention of ore blocks in the form of sill pillars for stability. In these cases, underhand cut and fill is suggested as a means to eliminate sill pillars, although for extensive orebodies, the requirement for a sill pillar remains even in underhand cut and fill operations. The integration of the ore pre-concentration and waste disposal step is expected to impact significantly on the design of the cut and fill based mine. A basic approach as suggested for the McCreedy East Mine (Bamber 2004) as well as for Xstrata’s Onaping Depth Mine is a centralized underground pre-concentration and waste disposal facility located between the orebody and the shaft (Figure 3.17). Ore is mined as in conventional cut and fill and can be transported to the pre-concentration facility by haul vehicle or alternately for simpler layouts, by conveyor. Pre-concentrate continues up the shaft via the hoisting system or alternate methods such as hydraulic hoisting. Waste from the pre-concentration plant is prepared together with tailings from surface and gravitated back to the fill stopes on demand. 70 FILLCUT ROM Or e Backfill P re -c on ce nt ra te Figure 3.17 – Integration of Pre-concentration and Waste Disposal with Underhand Cut and Fill An alternative approach would be to integrate a small pre-concentration and waste disposal facility designed for the production of one mining section only at the exit to the section; however, concerns about weak wall rock for orebodies requiring this method in the first place may preclude the creation of a stable excavation for this in the stope. Impacts on the method are significant (Bamber et al, 2005): selectivity is reduced, larger equipment can be employed, the costs of preparing and placing backfill, are reduced, thus the productivity of the method is improved and costs are reduced. Fill would typically be a composite-type fill, thus the total quantity of solid waste disposed of on surface is reduced for this method compared to the conventional approach. Composite fills are also generally higher strength than conventional cemented fills, and applications for the use of this fill to reduce or even eliminate sill-pillars are being considered. Additional impacts are projected for underhand cut and fill operations. Pre-concentration facilitates the consideration of the use of composite fill, which is more efficient in the use of binder, more cohesive, higher strength and quicker setting than comparable conventional rock- or paste fills (Weatherwax et al, 2007). This is expected to greatly facilitate re-entry times in underhand cut-and fill operations, thus greatly improving productivity. As in overhand methods, there is potential to reduce or even eliminate sill pillars. 71 Challenges lie in the scheduling of what is now a continuous source of fill back into the stopes, although modeling and simulation on ExtendTM indicates that a cut and fill section comprising 4 to 5 stopes (3 working, 1 fill and 1 spare stope) possesses sufficient flexibility to accommodate a properly scheduled continuous fill sequence (Morin et al, 2005). 3.4.2 Open Pit Methods The integration of pre-concentration with open pit methods is seen as a natural extension of this method of mining for increased extraction of the orebody. Several impacts and benefits with regard to open pit operations were identified in the course of the study for the Pipe II open pit in Thompson Manitoba. Firstly, pre-concentration impacts directly on operating costs, which impact directly on cutoff grade and the size of the ore reserve. Evidence from the literature, as well as from the outcome of the study indicates that in the case of applications to large, open pit operations, cutoff grade can be significantly reduced, if not eliminated for the pit (Lion-Ore 2006). This implies that, with the integration of pre-concentration as appropriate, the entire mineralized zone is now mined. A new definition of cutoff grade must be introduced as the cutoff grade of the deposit, to all intents and purposes, becomes the grade of the rejects from the pre-concentration plant. Grade control decisions in the pit are eliminated, and the stripping ratio is massively improved. Secondly, the introduction of pre-concentration, particularly with the use of sorting methods (optical and conductivity), in addition to producing an upgraded ore stream, creates the opportunity to simultaneously analyze and classify the wastes into reactive (acid generating) or non-reactive waste prior to disposal. Both optical and conductivity methods can be used to distinguish between valuable ore components, barren rock, and mineralized waste of various grades. Thus the integration of pre-concentration technologies with open pit methods is expected to massively increase extraction of the deposit, as well as improve waste management at these operations (Figure 3.18). 72 Open Pit Pre- concentration Plant Product Open Pit Non -acid gen erat ing Metal leaching Acid generatingReturn to pit ? Figure 3.18 – Process Integration for Waste Management in Open Pits 3.4.3 Block Caving Block-caving is a method gaining increased application and importance in the industry for deposits where open-pit methods are considered marginal due to increasing depth or where increased selectivity is required over the open pit method. Block caves are finding particular application in copper porphyrys, which by reason of their typically uniformly low grade must be mined at high mining rates, and according to Cross (2006) will be the principal source of primary copper by 2014. Block caving relies on the natural proclivity of mineralised rock to be fractured and weaker than the host rock, and hence its natural tendency to cave under gravity only when subjected to vertical stress. Ore blocks in block caving can be up to 200m high and extend the full width of the orebody. The block to be caved is prepared by undercutting the ore with a series of haul drifts accessing the orebody via a large number of regularly spaced drawpoints in a grid (Figure 3.19). Ore fractures and flows naturally under gravity and is delivered ultimately by gravity to the drawpoints where it is removed by large capacity LHDs and delivered to the ore handling system for hoisting. 73 Caving Block Mucking level Drawpoints Zone of potential subsidence Figure 3.19 – Representation of Block Cave Stope Block caving possesses a number of advantages and disadvantages which are presented for comparison in Table 3.7. Table 3.7 – Advantages and Disadvantages of Block Caving Advantages Disadvantages High capacity Relies on presence of inherent geotechnical properties of ore for flow which may vary during the cave and stop production e.g. presence of clays Low cost Long pre-production lead times for development of drawpoint level Low ventilation requirements Unselective Minimal support requirements underground Management and maintenance of drawpoints is critical to maintain production No lost blasts due to poor grade as grade control is executed directly at the drawpoint Dilution can be unacceptably high at end of cave sequence Large block caves present a high risk of surface subsidence There is some potential for pre-concentration at these mines and research has been undertaken into the pre-concentration of copper porphyry ores in particular (McCullough et al 1990; Burns & Grimes, 1986) with positive results. Pre-concentration is expected to facilitate the extraction of the valuable component of the porphyry while rejecting a large amount of barren material, the disposal of which on surface could be used to compensate for any expected surface subsidence 74 during the cave, massively benefiting the operability as well as profitability of the block cave operation. There are several motivations for considering the integrated mining, processing and waste disposal model for copper porphyry ores. Costs per ton are low, however ROM tonnages are high, and costs per ton of copper produced for these methods are consequently high (Pitt & Wadsworth, 1980). Copper porphyry deposits, for example at Candelaria and North Parkes mines, are typically located in remote, arid regions where low water consumption is critical, and the application of pre-concentration at these mines would seek to reduce water consumption considerably. Studies are being considered for these mines, however, the integration of the appropriate processing and waste disposal technologies into a block caving scenario has not yet been examined in detail and is considered a significant challenge. It is obvious that waste rejects cannot be disposed of directly underground, and thus the option to pre-concentrate underground is not immediately attractive, although this option facilitates the opportunity to hoist valuable material only at peak times, while scheduling the hoisting of waste only from the cave at more opportune times. Pre-concentration and waste disposal on surface is a possibility, with rejects from the pre-concentration facility deposited in the area of potential subsidence as a means of managing the subsidence process. A typical block cave mining sequence would be thus discovery, design, development, preparation of the caving area, pre-concentration during operations and reclamation, landscaping and closure of the cave area post operations (Figure 3.20). It is recognized that significant further research is required to develop this concept. Figure 3.20 – Integrating Pre-concentration with Block Caving Operations 75 3.4.4 Open Stoping Methods Open stoping is used in narrow, steeply dipping deposits where both ore and wall rock are relatively competent. Open stoping methods include shrinkage stoping, blasthole and longhole methods, and the following discussion is considered generically relevant for each of the methods. Open stoping is normally restricted to large, thick, steeply-dipping orebodies where both ore and wall rock are considered competent. Considerable development is required to prepare for the typical blasthole stopes, however much of this development can be in ore. Stopes are large, typically between 10 and 30 000 t, and barrier pillars are typically left at regular intervals within the orezone to enhance post mining stability (Figure 3.21). Dilution can also be high in more irregular orebodies, especially through wall sloughing when this rock is also less competent, thus the overall method, while high in productivity and low in cost, results in high levels of dilution and low overall extraction in unfavourable conditions (Canadian Mining Journal Sourcebook, 2007). Figure 3.21 – Typical Open Stoping Method As in the cut-and-fill scenario, integration of the pre-concentration and waste disposal facility into an open stoping environment will require the creation of excavations in competent rock, and located between the upper horizon of the orebody and the shaft. In shallower scenarios, the pre-concentration facility would be located on surface, as at Mount Isa, however the waste preparation and disposal facility should ideally still be located underground. For the successful 76 integration of the pre-concentration and solid waste disposal steps with the generic open stoping method, the method would be adapted to the modified AVOCA type method, as already in use at one of the case study mines, Musselwhite Gold mine in Northern Ontario. The AVOCA method is a shrinkage, or open stoping method modified for retreat mining with fill, for situations where the wall rock is less competent and the stope must be filled prior to the removal of the following block to aid in ground control and limit dilution. The ore zone is typically mined from the bottom up, and from left to right or right to left as the case may be with the longhole/ blasthole stope face retreating sequentially from the fill face stope by stope (Figure 3.22). Figure 3.22 – Adaptation of Open Stoping to Modified AVOCA with Composite Fill Several of the mines investigated during the research, such as Montcalm in Timmins ON, and the Thayer Lindsley Mine in Sudbury utilize open stoping methods, and conversion of these stopes to fill-based methods is recommended under the pre-concentration scenario. 3.4.5 Room and Pillar Several adaptations of the room and pillar method are required to accommodate the concept of integrated mining and processing. Room and pillar mining is typically undertaken in extensive, flat to shallow dipping orebodies in shallow depth or where hangingwall strength is poor, for example as at the Doe Run Company in Missouri. Mining is highly mechanized and low unit cost. Dilution is low, however extraction is also low due to the requirement to leave remnant pillars of ore for roof support. Adaptation of the room and pillar method to accommodate fill is expected to convert this method to the post-pillar cut and fill method. Several additional impacts and benefits are expected. Dips are typically shallow, which is a barrier to simple filling 77 operations. However, orebodies are typically thick, presenting several opportunities. Firstly the ore could be mined at a steeper apparent dip to facilitate the gravitation and confinement of fill. Also, in thick seam situations, rooms could be mined in a series of overhand passes, with previous passes to be filled in sequence, thus facilitating the fill cycle. Ore extraction in post- pillar cut and fill is higher than that of basic room and pillar mining, thus the economics of adapting to this method in the pre-concentration scenario are expected to be favourable, and has been recommended in for Doe Run in this instance. 3.5. Conclusion The motivation for integrating pre-concentration and waste disposal systems into the mining cycle is maximized in situations of extreme depth, low grade, remote geography, or a combination of these issues. Underground pre-concentration has the potential to reduce the logistical costs of handling large quantities of low-grade ore over great distances and depths underground, as well as delivering an increased value product to surface. There are positive environmental and economic benefits to the adoption of this concept as well in the reduction of waste delivered to surface and reductions in the size of the mill required on surface. The individual technologies for the development of integrated mining and processing, and waste disposal systems already exist at some level of implementation in the industry, and particular precedents for the practical application of these technologies also exist. Through the research they are projected to integrate well into the underground mining environment, and a successful application is envisaged to be found. However, it remains for these integrated technologies to be implemented in the field to create the envisaged system comprising mechanized mining, underground pre-concentration, automated preparation and placement of the waste as backfill. 78 4. Experimental Methods for the Geo-metallurgical Evaluation of Ores for Pre-concentration and Disposal of the Rejects 4.1 Introduction For the successful design of the pre-concentration and waste disposal system, as well as an accurate quantification of the impacts and benefits thereof, a detailed characterisation of the Run of Mine (ROM) ore is required. This Chapter outlines the tools and procedures that have been developed to assess the mineralogical properties of an ore with respect to its amenability to pre- concentration, evaluations of the characteristics and potential uses of the valuable and waste products, as well as the quantification of additional impacts and benefits such as savings in transport and milling costs. Examples are presented for selected ore types. The methods presented in this Chapter have been shown to be applicable to a wide range of ores. 4.2 Mesotextural Evaluation, Fragmentation and Liberation The first step in the design of the pre-concentration system is to assess the physical and mineralogical characteristics of the ore, including elementary visual observation of the presentation of the minerals in the rock. Models of mineral liberation often assume a random fracture pattern between particles, thus visual observation of the un-liberated ore is not typically considered; however it has been observed that for many ores that the response to breakage is largely controlled by the mineralogy and discrepancies in the comparative mechanical properties of the mineral components, such that preferential fractures tend to form along grain boundaries (McIvor & Finch, 1990; Olubambi et al, 2006). These grain boundaries can often be visually distinguished in the ore unaided, and have been shown in previous research to have a significant influence on the processability of the ore (Bocjevski et al, 1998). Furthermore, it has been noted in the literature that fragmentation and liberation can be successfully determined directly from this mineralogical texture, as behaviour between mineral phases in the rock are essentially structurally controlled (Schneider et al, 2003). Meaningful data can therefore be gleaned directly from visual and near-visual observation of the macro mineralogical texture, or mesotexture, in the unbroken rock. Thus in the research, study of the mesotextural characteristics of the ore is used to determine the potential for liberation of the valuable mineral, which in turn determines both the optimum feed size and performance of the selected coarse particle separation process. While microscopic examination as well as procedures such as QEM/SEM and Mineral Liberation Analyser (MLA) are useful for liberation analysis at finer particle sizes, this requires evaluation of fragmentation and liberation from ROM type particle sizes (~500mm) down to the 79 lower limit of the separation processes under consideration (~1mm). 2 dimensional visual observations of the ore texture in situ are thus of primary importance; photographs can be taken prior to blasting and collection of the samples (Figure 4.1). Should an exposed face not be available, similar data can be obtained in one dimension only from observation and photographs of the core (Figure 4.2). Good pre-concentration potential is clearly indicated when there is a clear visual discrepancy between valuable and non-valuable components in the ore, and good liberation of the valuable components at a coarse particle size. In the visual evaluation of liberation and separability, several principal mesotextural classes, in order of increasing complexity, can be defined. Similarly the mineralogical micro-texture is noted, and the combination of mesotexture and microtextural data is used as a preliminary indication of the potential for pre-concentration of the ore (Table 4.1). For good pre-concentration potential, a combination of the simplest meso- and micro-textures (massive + massive, or massive + matrix) is expected to give the best separation results. The visual information can be correlated with actual measurement of physical separation potential by size fraction using one of the available bases for discrimination. Table 4.1 – Mesotextural and Microtextural Classifications for Visual Evaluation of Ore Types (after Bocjevski et al, 1998) Mesotextures Micro-textures Massive vein Massive Massive sulphide Matrix Massive discrete sulphide Sieve textures Banded sulphides Simple intergrowth Disseminated sulphides Fine grained 80 Figure 4.1 –In-situ observation of Massive-Vein Massive Sulphide Mesotexture at Fraser Mine Figure 4.2 – Mesotextural Evaluation of Musselwhite Core For the pre-concentration of coarse ore, as distinct from fine ore processing, the degree of liberation of coarse barren rock is as important as the liberation of metal bearing rock. The extent of liberation of barren rock will indicate the potential amount of waste that can be rejected without significant metal loss. For underground pre-concentration this criterion is practically limited to circa 60% waste rejection by mass arising from the volumetric bulking effect of broken vs. in-situ ore (Bamber et al, 2005). Thus an important target to be determined is the particle size distribution at which a coefficient of liberation of 60% with reference to the gangue mineral can be achieved. For this measure, the size-assay of a crushed sample of ore is 400mm 100mm 81 recommended to determine the metal distribution and liberation data by size class in order to identify size classes with potential for direct rejection as well as the optimum liberation size for waste rejection. Measurement of further visual data of value in the evaluation of the concept includes RQD data from the cores which is useful as an indication of the fragmentation behaviour of the ore; measurement of the Work Index of the various ore, pre-concentrate and waste fractions is required in order to quantify impact in the milling and flotation plant. Based on a particle top size of approximately 500mm a representative sample size would weigh several tons; due to practical constraints the actual samples are restricted to between 500 and 1500 kg each. Samples are weighed and split into 4 sub-samples by the cone-and-quarter method. Samples are wet- screened into a series of size fractions selected based on the indicated top size, and a screen series of √2 (Figure 4.3). Figure 4.3 – Screened Sample Fractions for Size Assay Screened sub-samples are washed and visually inspected for characteristics such as colour, lustre, texture, and degree of liberation, which are to be noted. A series of photographs are taken of each size fraction and particles representative of the ore and waste are identified and taken for mineralogical examination by polished section. Coarse fractions can be hand sorted into ore and waste products based on observed colour, texture, and density discrepancies. The densities of the ore and waste fractions identified are measured by the volumetric displacement method. A liberation evaluation is essential to be performed. Particles in each size fraction are sorted into 82 groups according to an estimation of the degree of liberation. The Liberation Index, based on an evaluation of the proportion of fully liberated particles (particles with >80% sulphides) to the proportion of partially liberated particles, is evaluated according to the relation: Li = n(fi=90)*0.9 - (1) Σn*fi Where: n = number of particles fi = degree of liberation of particle The overall liberation coefficient of the sample is then calculated based on the weighted average of the liberation indices of each size fraction. A further characteristic indicator of the presence of pre-concentration potential is found through size assay of the fragmented ore. Size assay gives data on the particle size distribution of the ROM ore, as well as the metal distribution by size. Samples from each size fractions in the tests are split, crushed to –6mm, wet ground to –150 µm, oven dried and pulverised for assay. Representative 100g sub-samples of each size fraction are analysed typically for Cu, Ni and Co, as well as Pt, Pd and Au (TPM) and Mg content. In the course of the research, good liberation of the sulphides at coarse particle sizes, and an α-log normal grade distribution has been observed for ores which demonstrated good pre-concentration results (see Figure 4.4). It has been found that a normal distribution of grade by mass in the ore indicates poor pre-concentration potential. Figure 4.4 – Footwall Ore Size Assay Showing α-log Normal Grade Distribution Should a stope sample not be available, meaningful data can still be determined from core samples. Mesotextural information from the core is directly comparable to the in-situ 83 mesotexture. Size assay can then be performed on the core to determine a nominal particle size distribution as well as metal distribution by size. While the size distribution of the core sample will not be the same as the PSD of the ROM ore, core data can be used to determine the ROM size distribution by means of the Rock Quality Designation (RQD). RQD is determined for the core by measuring the total length of core pieces > 100mm and comparing to the total length of the core run: ⎥⎥ ⎥⎥ ⎦ ⎤ ⎢⎢ ⎢⎢ ⎣ ⎡ ≥ = ∑ ∑ i i li mmli RQD 0 0 100 - (2) The results of the RQD evaluation can be used together with data on the compressive strength of the rock, and information about the blast design, to correlate the size distribution of the core with the expected nominal size distribution of the ROM ore (Cunningham, 1983). See Figure 4.5. Blast Fragmentation - Pipe Open Pit 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.01 0.1 1 10 Size (m) W t% Figure 4.5 – Kuz-Ram Model Results for Pipe II ROM Ore The metal distribution by size in the core can then be adjusted proportionally to the modelled size distribution of the ROM. Once mesotexture, fragmentation, liberation and metal distribution characteristics have been determined for the ore sample, several methods of discrimination can be evaluated based on measurement of the discrepancy between values for relevant physical properties such as colour, texture, lustre, density, conductivity or magnetic properties of the discrete ore and waste fractions. 84 4.3 Separability Testwork 4.3.1 Dense Media Separation In the separability testwork, sink-float analysis is first investigated. Selected mineralised and non-mineralised particles are subjected to densimetric analysis by the volumetric displacement method. An indication of the separability of the ore can be determined by estimation of the gravity concentration criteria. The concentration criteria can be expressed as: Cc = (ρm – ρl) - (3) (ρl – ρg) Where: ρm = mineral density ρs = heavy liquid density ρg = gangue density The ore is then subjected to density separation tests at the range of SG’s indicated in the densimetric analysis. In the course of the testwork, ores with good pre-concentration potential have demonstrated an enriched fines component, and for base metal sulphides this is typically below 9mm. A novel static-bath DMS unit has been built in the lab to perform these tests (Figure 4.6). The unit has an effective separation range of -53mm + 9mm, thus the -9mm fraction is typically screened to concentrate, with substantial positive impacts on the metallurgical balance for these ores. Samples can be tested in batches of up to 10kg at a time using the rig. Typical results are shown in Table 4.2. Table 4.2 – Heavy Media Separation Results for 153 Ore (at SG 3.0) Cu Ni TPM Wt % % Wt % % Wt % g/t Wt% Feed Sink Float 100 45 55 13.26 29 0.5 100 98 2 0.39 0.8 0.06 100 91 9 14.8 35 1.4 100 93.3 6.7 85 Figure 4.6 – UBC DMS pilot testing unit The size distributions of the reject (float) and accept (sink) fractions can also be analysed (Figure 4.7). The presence of a bimodal size distribution in these samples reflects the different fragmentation behaviour of the sulphides and gangue minerals. Figure 4.7 – Footwall ore sink and float fractions by size showing bimodal distribution of ore & waste fractions Results can be been compared to current mine data to determine estimated dilutions for the samples. 4.3.2 Optical Characterization Optical data in terms of RGB, reflectance, colour balance and YES balance are measured for selected ore and waste samples using the National Instruments Machine Vision Station that has 86 been assembled at UBC. The NI Machine Vision station was specifically developed for the testwork due to the lack of commercially available optical sorting evaluation systems. Several mineralogical analysis packages are available, however these are typically for microscopic analysis and inappropriate for samples with particles larger than 10mm, and do not provide software for image comparison and discrimination, thus at present, investigation of this potential technology usually requires samples to be sent to one of the various sorter manufacturers, who are based either in Europe or Australia, which has been a financial barrier to the research. Therefore, in order to investigate optical sorting in more detail, an optical analyser has been developed and tested on range of ores at UBC. System Design Several options for this were available, and a functional specification based on previous observations was developed. The requirements for the system to be assembled were: • System should be capable of analysis and discrimination • Optical analysis was to be on the basis of colour, lustre, brightness, and texture • Field of vision would be suitable for samples between 10mm – 100mm • Images would be high resolution, digital colour images • Colorimetric analysis would be by absolute Red-Green-Blue (RGB) value as well as colour balance (‘gamma’), textural analysis and ‘YES’ colour comparison Automated digital image analysis is a specialized field and only a few vendors worldwide offer hardware and software for this application. National Instruments is a leading vendor in the field, and offers educational versions of its Labview data acquisition and analysis suite to UBC, thus it was decided to investigate a solution using this avenue1. In digital imaging, colour, colour balance and overall intensity are automatic components of the image data set, thus analysis by these features is relatively elementary once a suitable image is captured. Camera and lighting specification and setup are important in order to optimize this however, thus a specially designed image capture enclosure with standardized lighting was designed. Texture analysis was initially intended by means of the standard pattern recognition features offered by NI/Labview. Pattern recognition is a method for the accurate and efficient recognition and categorization of an object based on a characteristic subset of the object’s 1 http://www.ni.com/analysis/vision.htm 87 features. Traditional pattern recognition requires comparison of the sample pattern area to every area in the image which is very computationally intensive. Advanced pattern recognition involves two steps – pattern learning and pattern matching. The learning phase involves identifying common characteristic features on a series of objects by repeated exposure of the image analyser to the object. If found, these characteristic features can be used for efficient and rapid pattern matching between new objects with similar features. Probably the most common application of pattern matching is in quality control for circuit boards for example the checking of capacitors on an amplification circuit board (Figure 4.8). Figure 4.8 –Pattern matching application for electronic components Labview features a standard pattern matching algorithm which incorporates pseudo-random sampling, neighbourhood checking, and pattern edge detection2. Pattern matching is independent of pattern orientation, which in conjunction with the above features suggested huge potential in mineral image analysis due to the irregularity of the typical mineral object. For the initial development of the system at UBC, Labview 7.1 was used and a National Instruments system integration specialist, Pishon Software of Surrey, Vancouver, was engaged to specify the final hardware setup and configure the Labview software. The final optical analysis setup is shown in Figure 4.9. It soon became apparent that the pattern recognition method is limited in this application, as the routine requires an almost exact match for success, while the pattern of mineralization even within the same sample was found to be insufficiently consistent to generate an accurate result. Pishon Software was again engaged to overcome this, and it was decided to use fuzzy logic in the pattern recognition routine in order to obtain a distribution of matches instead of an exact pattern match. 88 Figure 4.9 – UBC NI Machine Vision Optical Analysis System Schematic (left) and Actual System (right) ‘BrainCom’ is a freeware artificial neural network routine using back-propagation in the network to define the coefficients for the neural net (Wittnaum, 2001), and was obtained and configured for use with the UBC image analysis system. Results are encouraging and the system is now able to be trained to recognise the presence a range of typical sulphide minerals such as chalcopyrite, pentlandite and galena in a sample. The system calculates the total are of mineralisation in the sample and an approximate grade for each particle can be calculated. Experimental Procedure A typical test procedure involves the selection of several coarse rock samples for evaluation. It is important that the selection includes a full range of the styles of mineralization identified geologically. Rocks are weighed, numbered and split into ‘A’ and ‘B’ fractions. ‘A’ samples are subjected to image capture and analysis. ‘B’ fractions are prepared and sent for assay. Results of the optical evaluation are compared to the assay results and the basis for a sort using optical parameters can be established from the data. Selected samples of concentrate and rejects from the testing of the Xstrata ores were subjected to optical evaluation using the sensor setup. Reflectance, and colour measurements were taken and correlated with the grade of the samples. Results are presented in Table 4.3. RGB data from the images can also be analysed spatially to determine textural characteristics of the ore sample. Details of a complete photometric evaluation of ores from Xstrata nickel is presented in Appendix H. 2 http://zone.ni.com/devzone/cda/tut/p/id/3763 89 Table 4.3 – Correlation of Photometric Measurements to Grade for Xstrata Ni Ores Sample Red Green Blue Red Green Blue Craig 8112 Ore: 1.26% Ni, 0.57% Cu Waste: 0.19% Ni, 0.12% Cu Ave 66.94 66.94 108.28 107.68 100.95 105.32 Sdev 38.84 38.84 50.56 52.83 47.76 41.10 Craig LGBX Ore: 3.5% Ni, 0.38% Cu Waste: 0.21 % Ni, 0.18% Cu Ave 109.41 104.76 98.22 104.27 101.15 87.44 Sdev 51.33 58.43 56.70 56.85 58.53 53.38 Fraser Cu Ore: 0.84% Ni, 22.01% Cu Waste: 0.03% Ni, 0.4% Cu Ave 107.51 92.23 75.94 99.01 89.37 68.77 Sdev 64.10 61.10 47.79 37.49 42.16 21.20 Fraser Ni Ore: 1.97% Ni, 2.8% Cu Waste: 0.20% Ni, 0.12% Cu Ave 95.73 97.45 105.38 107.68 100.95 105.32 Sdev 36.03 40.95 41.55 52.83 47.76 41.10 Montcalm H Ore: 2.12% Ni, 0.82% Cu Waste: 0.15% Ni, 0.18% Cu Ave 116.27 120.28 117.55 95.062 111.05 125.78 Sdev 56.22 54.05 45.42 48.84 52.22 49.99 Montcalm L Ore: 1.4% Ni, 0.56% Cu Waste: 0.15% Ni, 0.1% Cu Ave 88.63 83.23 76.34 86.82 81.76 75.52 Sdev 51.96 47.47 39.35 46.23 49.25 43.62 T-L 15 Ore: 1.83% Ni, 10.79% Cu Waste: 0.076% Ni, 0.4% Cu Ave 92.33 104.66 109.00 129.74 121.60 104.08 Sdev 48.54 51.71 46.92 48.13 50.72 47.11 TL-80 Ore: 1.70% Ni, 1.11% Cu Waste: 0.114% Ni, 0.15% Cu Ave 131.28 117.04 90.41 74.98 82.54 82.40 Sdev 47.12 48.35 43.33 34.41 38.10 32.62 T-L 670 Ore: 0.82% Ni, 0.45% Cu Waste: 0.16% Ni, 0.15% Cu Ave 130.40 138.00 124.40 119.09 112.05 92.27 Sdev 44.65 47.55 44.22 46.10 49.66 33.40 For the Xstrata ores, the RGB data correlates well with the assay results and the basis for a colorimetric sort has been established for each ore. Based on the results, a synthetic grade- recovery curve can be constructed retrospectively from the data by correlation of the sensor response curve with the assayed sample grades. A synthetic sort can then be performed on the data to indicate potential sorting results through the selection of an appropriate cutoff grade. 4.3.3 Conductivity Evaluation It was also desired to evaluate various samples for their amenability to conductivity based separation. Conductivity methods are generally applicable for native metal ores such as gold and copper, as well as massive sulphide ores grading between 2-3% metal with good results. Testwork reports on copper porphyry ores from a range of mines in the Montana and Upper 90 Michigan area indicate conductivity sorting delivered up to 50% waste rejection by mass from the ores at recoveries from 85 – 92% (Miller et al, 1978). It is suggested that conductivity methods should be utilized in conjunction with optical sensors in order to compensate for the variance in response by particle size for improved results. In previous ore sorting tests at INCO, tests of optical sorting or conductivity sorting alone on combined ROM ore from McCreedy East did not give acceptable results (Schindler, 2001). Sorting tests on the ore using a combination of optical and conductivity sensing in a Mogenson sorter, gave results of up to 77% waste rejection at a recovery of 98% confirming the potential of combined sorting. The method is appropriate for native and sulphide ores grading between 1-3%, however, is inaccurate for low grade disseminated sulphides < 1% and any particle < 1mm. A variation on the metal-detector type sensor which is indicated for use with lower grade base metal sulphide ores is the induction balance coil (Sivamohan & Forrsberg, 1991). This arrangement gives good results for native ores and is considered to have high potential for sulphides such as such as sphalerite, chalcopyrite, millerite, galena and covellite in VMS and igneous intrusive orebodies where there is a significant conductivity differential between the ore and gangue minerals. In this arrangement a second balancing coil is introduced, and the difference in signal between the disturbed (sensing) and undisturbed (balancing) coil is measured. Laboratory equipment for the evaluation of an ore to this method is not commercially available, and as in the case of optical methods, testwork by the vendors is expensive and prohibitive to the research. On this basis, other avenues for the development of capabilities in ore characterization by conductivity methods were pursued. 126.96.36.199 Testing of the INCO ‘B2’Sensor A sensor based on these principles, and using a coil design taken from the field of geophysics, had previously been developed at Ecole Polytechnique (Boucher, 2005), and employed in testwork at INCO ITSL in Sudbury. Sensor design is of the ‘pancake’ induction coil type comprising a single-layer coil of signal wire (Figure 4.10). 91 Figure 4.10 – ‘Pancake’ type inductance coil (after Boucher, 2003) The sensor had been brought to INCO Technical Services Limited in Sudbury for further testing and development. This INCO ‘B2’ sensor was tested against a commercially available sensor from Applied Sorting Technologies of Sydney, Australia on selected Ni ore and slag samples at INCO Technical Services with good results (Boucher, 2003). The sensor was given by INCO to UBC for further testing versus the AST conductivity sensor on low grade ultramafic nickel ores from Thompson, Manitoba. The arrangement of the ‘B2’ sensor setup is shown in Figure 4.11. Figure 4.11 – INCO ‘B2’ Sensor setup Prior to the testwork, the sensors were taken to Thompson and calibrated on selected core samples (Stephenson, 2006) using the half-rock assay procedure described under the section on optical sorting. Core conductivity tests were performed on NQ core halves selected from previously logged drill holes from various locations around the Pipe deposit in order to evaluate 92 the conductivity response of the sensor to the core pieces, and compared to Ni assay data from INCO Sheridan Park in Mississuaga for each piece of core (Figure 4.12). BH1410 - EM Reading and % Ni 0 1 2 3 4 5 6 7 0 20 40 60 80 100 120 140 Sample Number % N i 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 E M R ea di ng % Ni EM Reading Figure 4.12 – Correlation of B2 reading with Ni Grade (from Stephenson, 2006) A good correlation between the EM reading and % Ni was thus established for the samples using the sensor. The AST sensor was erratic and inaccurate at low Ni content, with accuracy improving with increasing Ni grade. The ‘B2’ sensor was shown to be unresponsive in the lower Ni grades, erratic in the cutoff range, but demonstrated an increasingly accurate response above this figure. From the tests, it was concluded that the current B2 sensor design had potential in this application, but embodied a number weaknesses. The sensor was: • unresponsive in the lower ranges, and erratic around the cutoff grade • unreliable and highly sensitive to ambient interference • not of robust construction and unsuitable for an industrial environment • the pancake sensors are large, and highly sensitive in the centre, but signal strength falls rapidly with increasing distance from the centre of the coil • the setup comprises 10 individual system components, making the arrangement unwieldy and complicated to set up and use • there were a number of redundant functions in some of the system components which would be eliminated with improved component selection 93 188.8.131.52 Development of the MineSense‘B2’ MkII Sensor On further examination, it was found that the B2 sensor incorporated a number of further fundamental design and construction errors. Several components, such as the amplification and bridge circuits had been custom designed and possessed a number of flaws when compared to good practice in instrumentation design. The sensors were not robust, were imperfectly shielded, poorly grounded; furthermore the signal wires and connectors on the sensor supplied were damaged, possibly contributing to the poor sensor performance. Also, according to the software specification, it was intended to perform Fast Fourier analysis on the incoming signal, followed by calculation of the phase and amplitude components of the signal vector (Bamber & Houlahan, 2007). However, it was found that the vector analysis calculations in the software were incorrect and that the final sort decision was being made on a false basis. After the testwork campaign, the ‘B2’ and AST sensor setup were returned to INCO for further testing at Sheridan Park. However, it was decided to proceed with the design and construction of an improved MkII unit at UBC in order to overcome the issues with the ‘B2’ that had been identified. Private sector funding was obtained from BC Mining Research Ltd and an electrical EIT was engaged. The sensors, bridge and amplification circuits were redesigned in accordance with good instrumentation and electrical engineering practice; additional design work involved integrating the power supply into the amplification and bridge box, and integrating the signal generation, signal acquisition, and sort signal processing into a single PCI data acquisition card. The primary improvement in the design is an improvement in the power and sensitivity of the sensing coils. The single layer pancake coil has been replaced by a smaller double layer coil incorporating a solenoid air gap which maintains the field strength across the majority of the coil diameter (Figure 4.13). The B2 sensor-bridge and signal amplifier was identified as an area for improvement, and the bridge and amplifier arrangement was redesigned according to conventional instrumentation principles. Principal design changes were the introduction of a balanced Wheatstone bridge, and the use of variable amplifiers on the input and output signals to and from the sensors. The modified sensors, bridge and amplifier network have been integrated with a new PCI data card and PC. The new PCI card is capable of generating the base signal as well as analyzing the attenuated signal from the sensors. The card allows up to 4 input channels, thus it is now possible to expand the sensor arrangement into a sensor array. Signal analysis software written in C++ was provided with the Chico PCI card and has been modified to analyze the attenuated signal and calculate the magnitude and phase of the sample impedance. 94 Figure 4.13 – Plan and Section Plot of Field Strength vs. Coil diameter (B2 Mk1 (top) vs B2 MkII (bottom) (Bamber & Houlahan, 2007) Using these modified components, an integrated inductance-based ore evaluation station has been built and tested at UBC with significantly improved results over the ‘B2’ prototype. The final B2 MkII sensor setup is shown in Figure 4.14 & 4.15. 95 Figure 4.14 – MineSense ‘B2’ MkII Conductivity Sensor Schematic Figure 4.15 – MineSense ‘B2’ MkII Sensor setup showing data capture PC, bridge amplifier (centre) and sensing (right) and balancing coils (left) The sensor gives a calibrated response for conductivity (in mV) as well as magnetic susceptibility (Χm). The sensor was tested on a sample set of Xstrata ores previously calibrated by half-rock assay for the optical characterization tests. A correlation of sensor response with metal grades is presented in Figure 4.16. 96 y = 0.0045x + 0.5927 R2 = 0.6124 0 5 10 15 20 25 -200 0 200 400 600 Conductivity (mV) Cu , N i % Cu % Ni % Cu + Ni Figure 4.16 – Preliminary Correlation of ‘B2’ MkII Conductivity Readings with Grade for Xstrata Cu-Ni Ores Initial results show a good correlation between the conductivity response and Ni grade. Correlations with Cu grade are not apparent, although an initial correlation between Cu grade and magnetic susceptibility has been observed. As in the optical evaluation procedure, data from the grade/sensor response curve can be used to generate a synthetic sort for evaluation of the potential for waste rejection prior to undertaking actual sorting testwork. The sensor is recommended for use as a laboratory tool for evaluating Ni sorting in this application. As with the original B2, it is felt that this sensor array, with increased power and sensitivity has a number of additional applications including core logging, grade analysis and down-the-hole sensing. However, based on initial tests, it appears that the new coil is also highly sensitive to changes in the dielectric characteristics of the coil environment, and due to the complexity of the attenuated signal arising from the coil it appears possible to discriminate more than just a scalar value of grade, and thus is potentially able to distinguish between the highly non-metallic samples passed across the sensor as well. This observation is supported by references from the literature (Kieba & Ziolkowski, 2002) and the application will be developed through further research and testwork using the MkII sensor. Further developments, including improved signal conditioning and the option to expand the sensor setup for belt sorting by adding additional sensors in an array are pending. 97 4.4 Grinding Work Index Testing The pre-concentration of ores is expected to result in power savings at the mill in three principal areas. Firstly, the rejection of waste results in a reduction in tonnage reporting to the mill. Secondly, the waste is primarily hard siliceous rock; removing this material leaves a product enriched in relatively soft metal bearing sulphides which will have a lower grinding work index. Thirdly, the pre-concentration product will be crushed to a finer size than the present plant feed. This projected impact on grinding power requirements can be calculated from the Bond equation as follows. ΔP = Δt10ΔWi [1/√P80 - 1/√F80] - (4) Where: Δt = waste rejection (tph) ΔWi = Change in Bond Work Index P80 = 80% passing size of the product (µm) F80 = 80% passing size of the mill feed (µm) For the evaluation, the work index of a 2kg sample is determined through a full Bond Work Index test and used as a reference. The sample is crushed to nominally -6.7mm and subjected to grinding in a standard Bond Mill with a closing screen of 149µm. The Work Indices of the other ores and ore concentrate and waste products are then determined by comparing the product size distribution of the samples to the reference sample in comparative rod mill grinding tests under identical conditions: WI2 = WI1 * √p802 -(5) √p801 4.5 Additional Geotechnical, Geo-metallurgical and Rheological Testwork 4.5.1 Geotechnical Characterization of Rejects A range of waste products re produced during the pre-concentration testing. Waste products are to be characterized geotechnically in preparation for their use as potential underground fill material. Appropriate geotechnical properties were selected for evaluation and compared to values recommended in the appropriate ASTM standards (Talbot & Richart, 1923): • Particle size distribution • Particle shape • Adsorption • Specific gravity • Void space • Strength • Chemical composition 98 Particle size distribution is evaluated by means of screening and weighing the screen oversize. Particle shape is evaluating by comparing the typical ratio of the longest and shortest particle dimensions respectively. Particle density is measured on a dry, weighed sample by the displacement method in a 1l volumetric flask. Absorption is also measured in the same test - aggregate is left in the volumetric flask for 15 minutes; the sample is then removed and dried on burlap. The difference in mass between the soaked and dry aggregate gives the absorption in % of the sample (Talbot & Richart, 1923). The size distribution of the rejects is analysed and compared to the recommended ASTM aggregate size distribution curves (Talbot Curves) for fill aggregates. The coefficient of uniformity is determined by calculating the quotient of the d80 and d20 of the products. The coefficient of uniformity also determines the void ratio in the aggregate sample. Void ratio is a measure of the free space within a mass of broken material, and is thus used to estimate the bulk density of such a sample. This information is collectively used to determine the optimum density for the mix design. The size distribution of the rejects is ultimately controlled by the method of feed preparation and concentration and thus the coefficient of uniformity, void ratio, and ultimately the optimum mix design can vary widely. The ores tested typically demonstrate a characteristic high grade fines fraction and removal of this fraction prior to concentration increases the relative size distribution of the rejects when compared to the ROM ore, increasing their compatibility in terms of aggregate uses. Evaluation of the characteristics of the Xstrata rejects indicated the rejects measured generally coarser than a typical ASTM fill aggregate of similar topsize, but were generally within the envelope of the Talbot specification and are nevertheless considered acceptable for use as a fill aggregate. Rejects were further geochemically characterised by ICP whole rock assay in order to evaluate any potential acid or neutral rock drainage potential. Acid based accounting is derived from the chemistry of the minerals as evaluated in the ICP whole rock assay. Rejects can be classified as either non-acid generating, neutral or acid-generating depending on the stoichiometry of the sulphur assay in the samples. Results confirm that the rejects are generally non-acid generating, with a minority of potentially acid-generating rejects. The results confirm the potential shown in preliminary mix design and testing (Bamber et al, 2006), and it can be concluded that the rejects of the pre- concentration process under consideration can typically be considered acceptable for disposal on surface as non-acid generating waste or use as aggregate in fill underground. 99 4.5.2 Mix Design and Testing Once the rejects are geotechnically and geochemically characterized, they can be evaluated and tested in terms of an appropriate fill mix design. Several mixes by mass are typically to be considered for the testing: • A rockfill comprising typically 95% rejects, 5% OPC with 1:1 water cement ratio ‘Maximum density’ composite fill with typically 70% rejects ‘1:3’ composite fill with typically 36.5% rejects ‘1:7’ composite fill with typically 20% rejects 5% cement content by mass is suggested as a standard for the composite mixes in order to generate comparative results. Mixes are prepared using the data from the characterization and tested according to ASTM standards for backfills. Sample moulds were industry standard plastic cylinders typically 200mm long x 100mm Ø. Moulds are typically half filled with mix and left covered in polyurethane for 28 days to cure at room temperature. UCS testing was done at the UBC Geomechanics laboratory on an MTS 815 ‘stiff’ UCS testing machine (Figure 4.17). Figure 4.17 – MTS 815 UCS Testing Machine The uniaxial stress/strain curves from the UCS Testing of the samples are then analysed to determine the peak stress at failure as well as evidence of any residual strength in the fill post- failure. The strength and efficiency of the mixes in terms of UCS : Wt% cement is also analysed (Table 4.8). The binder efficiency generally increases significantly with increasing reject content, thus indicating the positive benefits of utilizing these rejects as fill material. Ultimate 100 cement content in the mixes as well as water cement ratio can vary widely, so care in preparing the mixes is recommended. 4.5.3 Rheological Tests As it is the intention to use the rejects as a material for composite fills in underground backfill applications. Thus the transport of the fill becomes an issue and it is felt important to investigate the basic rheological properties of the fills for the purpose of determining an appropriate rheology for transportation by pumping. For this, the slump test method of Hu (1995) was chosen. The appropriate rheology for fill placement is determined by three characteristics, the shear stress τ0, plastic viscosity μ and shear strain rate γ’ (strain gradient) of the mix (Tattersall, 1991): . 0 μγττ += - (6) The standard slump test returns a value for the slump of the concrete which can be used to determine the yield stress of the mixture in the following form: ( )s−= 300 2700 ρτ - (7) Where ρ- density of the mix s – standard slump in mm For full determination of the parameters of the Bingham equation (5), use of the modified slump test to determine the plastic viscosity of the mix is proposed where the rate of slump in s is also measured (Chiara et al, 1998). Slump rate in s has been shown to empirically determine the plastic viscosity in mixes with low slump according to the relation tx ρμ 31025 −= - (8) Where t is the time of the slump in seconds For the slump testing a cylinder with a diameter of 100 mm and height of 150 mm was utilized. The slump cylinders were filled as for the UCS testing, then drawn slowly over the slump sample. Slump was measured from the top plane of the mould to the upper surface of the extruded material. In combination with the physical observations, the slump test and the resultant rheological analysis allow for general discussion of the potential to pump the composite fill mixes. Figure 4.18 shows indicative slump testing results for selected fill samples 101 composed of Fraser Copper rejects and full tailings. Figure 4.18 Slump test Photographs for Fraser Copper Fill Mixes showing qualitative differences in rheology of the mixes For a pumpable fill mix, a τ value of between 0.15 – 0.5 is suggested; mix optimization to achieve maximize strength and binder efficiency while maintaining a low yield stress in the fresh mix is thus recommended. 4.6 Conclusions A procedure for the evaluation of ores with reference to the mineralogical, metallurgical, as well as additional geotechnical and geometallurgical characteristics of the ore and waste fractions has been presented. Site visits for geological and mineralogical information gathering, examination of the in-situ meso- and microtextures have been shown to be important. Size-assay, liberation and fragmentation characteristics, and methods for potential particle separation must be determined. Geotechnical characteristics such as particle shape, strength, hardness and Work Index have also been determined to be important in the evaluation of the waste products. A range of fill mix designs have been developed and procedures for evaluating the relevant geotechnical and rheological properties of the mixes have been identified and used. The procedure generates data for the evaluation of the potential for pre-concentration of the ore, suggests potential separation methods and provides data for the evaluation of waste disposal strategies. The results from the evaluation also provide data for the parametric evaluation of the opportunity, impacts and benefits of the concept as discussed in Chapter 5. 300mm 102 5. A Parametric Impact Valuation and Evaluation Method for the Application of Pre-concentration and Waste Disposal Systems to Hard Rock Mining 5.1 Introduction The results generated from experimental evaluation of the candidate ores for pre-concentration and waste disposal as described in Chapter 4 allows the subsequent evaluation of various impacts and benefits which have been identified in the course of the research. Pre-concentration reduces the quantity and increases the quality of ore arriving at all stages subsequent to the pre- concentration step by rejecting between 30-60% of the coarse, barren siliceous material from the ROM ore. Impacts and benefits arising from this that have been identified through the research include (Klein et al, 2002; 2003; Bamber et al 2005, 2006): • Facilitation of lower unit cost mining methods through a decrease in selectivity • Facilitating an increase in mining rate without increasing the capacity required of downstream facilities such as haulage, hoisting, transport and milling • Increasing the mining rate without increasing the capacity required in the concentrator • Increase in the grade of ore delivered to surface without significant loss of metal • Reduction in the cost of engineered support through the generation of large quantities of fill material close to the mining face • Improvements in geotechnical properties in the fill and thus ground control due to an improvement in post-mining rock mass stability • A reduction in ventilation required through an reduction in the number and thus surface area of unfilled voids • Reduction in overall mining, plant and infrastructure capital costs; • Higher overall metallurgical recovery in the surface grinding and flotation plant compared to unsorted ore (specifically on well liberated, highly diluted low grade ores); • Reduction in overall material handling costs subsequent to the pre-concentration step; • Reduced energy consumption in transport and comminution; • Reduced waste disposal costs of coarse dry waste compared to fine, saturated waste; • Reduced water consumption in processing and waste disposal; 103 Negative impacts are also to be noted, and include in particular: • Additional capital for the pre-concentration and waste disposal plant; • Additional unit operating cost for pre-concentration • Additional metallurgical penalty across the pre-concentrator; • Additional excavations required for the underground situation • Additional heat, noise and dust in the underground environment Quantification of these impacts is not an elementary task, as it involves estimation and quantification of impacts along the entire mining value chain. The valuation and evaluation of impacts such as these typically requires a multi-disciplinary team of professionals, an option which is not available at this early stage of concept development. Using the expected impacts and benefits as a guideline, and data from fieldwork and laboratory testwork as a basis, a comprehensive framework for defining, valuing, and evaluating this approach in comparison to a more conventional exploitation approach has been developed and is presented in this Chapter. 5.2 Evaluation Methodology Earlier studies conducted under this research programme utilized a spreadsheet-based approach incorporating fixed grade, tonnage and metal prices with cost savings arising from pre- concentration estimated by proportional adjustment of the activity-based costs (Bamber, 2005). However, this method did not include for the evaluation of capital cost impacts, or other indirect impacts of the suggested approach on the feasibility of mineral extraction. Parametric estimation and evaluation methods are suggested as a means of overcoming this limitation for the purposes of this thesis. The evaluation and valuation of the extraction and recovery of a mineral deposit is a special class of investment analysis. Mining projects are capital intensive and are characterized by high fixed costs, high risk (geological risk, political risk, market risk, operating risk), and an irreversible depletion in the major asset – the resource (Gentry & O’Neil, 1984; 12). Evaluation requires a methodology for quantifying and estimating impacts to these variables. Most variations in methodology relate to geostatistical methods used in determining the resource, cost estimating methodologies and the method by which the project value is determined. Such methods do exist, but are presently manual, paper based, of limited applicability and accuracy, and lack detail particularly in the estimation of capital costs for mine, surface plant and infrastructure. The economics of the project are largely determined by 104 the price and cost regime at the time of evaluation, coupled with some forecast of how input costs and market prices will vary over time. Project evaluation can be by replacement cost, total sales value, earnings based (NPV) or by option pricing techniques such as CAPM (Barnett & Siorentino, 1994). The basic deposit evaluation methodology broadly incorporates the following features: Input of basic deposit parameters such as ore reserve, metal grades, geotechnical and geographic factors Automated calculation of capital costs and activity-based working costs for the pre- concentration scenario based on a user-input operations flowsheet Adjustment of the overall mineral extraction and metallurgical recovery based on the 2- product formula Re-calculation of the cutoff grade based on revised working costs and recoveries Re-calculation of the size and grade of the mineral reserve based on an idealized grade- tonnage model for the deposit Calculation of the Net Present Value Calculation of the comparative asset utilization efficiency between the proposed approach and the base case as a tool for the final evaluation of the concept The inputs for the model are a data set which must be developed through site visits, sampling, and the results of the geo-metallurgical evaluation procedure described in Chapter 4. Additional data comprises the size and grade, and thus the value of the deposit; the selected mining rate, documented operating costs, data from existing milling operations; financial parameters such as metal prices, discount rates, and any smelter tolls or additional taxes which may be levied. The system cost estimate is factorized based on data from the literature and capital estimates from previous studies. Operating costs are activity based and are recalculated according to the degree of waste rejection achieved in pre-concentration and the location of the pre-concentration step in the overall value chain. Revenue losses are calculated according to the recovery data for the pre- concentration process and metallurgical performance at the existing mill. Recovery improvements due to the calculated increase in feed grade to the mill are also accounted for by means of the two-product formula. The result of the analysis is used to recalculate the cutoff grade for the deposit, which in turn is used to re-evaluate the size and grade of the economic mineral reserve to be exploited. The parametric project evaluation process for a given deposit is thus an iterative process as depicted in Figure 5.1 (after Gentry & O’Neil, 1984; 4). 105 Profit and Value Analysis Ore Reserve Model Capital costs Production rate Operating costs Resource Economic Cutoff Grade Market Analysis Figure 5.1: Mineral Deposit Evaluation Flowchart A basic data set for mineral deposit evaluation is proposed for the purposes of the evaluation as follows: 1. Ore reserve – the size of the ore reserve to be mined in tonnes 2. Mining rate – the mining rate in tonnes per day and production profile if relevant 3. Operating days/annum – (0 < x < 365) 4. Metals of Interest – copper, nickel etc. 5. Metal grades – contained metal grades in % or in grammes/tonne as the case may be 6. Mill recoveries – existing or predicted % metal recovery in the surface mill 7. Metal prices – current metal prices in $/tonne or $/oz as the case may be 8. Operating costs – activity-based breakdown of the total (fixed + variable) operating costs of the mine (drill, blast, muck, haul, hoist, surface haul, mill and tailings) 9. Capital costs: • Mine development, mill development, infrastructure and pre-concentration capital 10. Processing parameters 10.1. Throughput 10.2. Recovery 10.3. Concentrate specs 106 11. Pre-concentration parameters (projected or derived from metallurgical testwork) • Waste rejection – the percentage of total ROM ore (including planned and unplanned dilution) rejected in the pre-concentration step. • Metal recoveries – recovery in % feed content (by element) achieved in the pre- concentration process 12. External factors • Geography – location, elevation, surface conditions, geotechnics, climate 13. Financial parameters • Smelting toll - % toll (including metal losses) levied on concentrate feed to the smelter • Discount rate – expected rate of return on investment / risk factor • Smelter tolls, penalties and recoveries • Technological factors • Market analysis – a price forecasting function is included • Source & cost of capital • Extent of pre-production design, development & construction • Shipping, Marketing, Freight & taxes • Expansion options 14. Other factors • Geographical parameters – location (urban, rural, remote, arctic) • Geotechnical parameters (impacts the evaluation of surface, hangingwall and footwall conditions) – good, fair or poor Aspects such as environmental, legal & permitting, and exploration or production drilling costs are assumed to be sunk and are not accounted for. Data from the data set is used to configure the model, and evaluate a decision to invest or not to invest in the pre-concentration option. The model is sensitive to the date and relevant market conditions under which it is run, and evaluation results may change as these factors change over time. The model evaluates the NPV of a particular ore reserve, and compares it to the projected NPV for the exploitation of a similar deposit with the implementation of pre-concentration on surface or underground. The model has been configured for copper, nickel, cobalt and three platinum group elements (3 PGE’s) – platinum, palladium and gold - although other metals can be considered. The spreadsheet is protected for input of only the critical parameters. All variables are in S.I. units, and costs are in present-value US dollars. 107 5.3 Model Assumptions It is suggested that the pre-concentration of ore, combined with the appropriate disposal of the waste rejects underground as backfill can address a number of the core areas impacting the operating and capital costs of a mine. A cursory examination of previous study results indicates that up to 55% of the ore can be rejected through sorting or dense media separation techniques at a metal recovery of 97%, with a resultant increase of 100% in the grade of the ore delivered to surface (Bamber et al, 2004). The rejects are suitable for disposal underground in a suitably designed rocky paste fill (Bamber, 2005; Kuganathan & Shepherd, 2002). The model assumes a degree of impact on based on the expected impacts of these physical aspects. Pre-concentration rejects a significant proportion of the ROM tonnes mined from the ore at high metallurgical recoveries, indicating a potential business case if savings can be derived therefrom. From previous scoping studies and evaluations, cost savings through underground pre-concentration are derived in a number of ways: • potential savings in mining costs are derived through an increase in productivity at the mining face through a reduction in selectivity required, the facilitation of more productive mining equipment at the face and thus a potential increase in the minimum mining width allowable and a consequent decrease in mining costs (Bamber et al, 2005). For the model, mining productivity is assumed to improve by 10%. • savings in fill costs are possible through the generation of a large quantity of cheap backfill material close to the stopes as a by-product of the pre-concentration process. The balance of the backfilling cost constitutes the batching of waste rejects with water, sand and cement, and costs to place the cemented backfill. Cost impacts such as a potential reduction in rockbursts or greater stope- and drift reliability through improved ground control are not accounted for here. • savings in hoisting are directly proportional to the proportion of waste rejected. The total tonnage reporting to the shaft is reduced, shaft utilization and therefore operating time is reduced, possibly by up to 1 shift for waste rejection > 30%. • hoisting and material handling constitutes more than 65% of all energy usage in underground mines. A reduction of 60% in the quantity of material to be hauled, hoisted to surface and delivered to the concentrator results in an estimated 30% savings in the overall power cost for the operation. Additional power savings are derived from the 108 elimination of a large quantity of hard siliceous waste in the feed, with a consequent reduction in grinding effort, as well as overall tons to be processed in the mill. • up to 60% of the ore has been rejected prior to surface processing. This results in lower unit cost/ROM t for processing and tailings disposal. Savings due to a reduction in Bond Work index and thus grinding effort relating to the reduction in tons reporting to the mill, and the reduction in BWI are thus accounted for in the model. • Additional operating costs – underground pre-concentration constitutes an additional processing cost, which must be accounted for. The metallurgical processes under consideration are typically cheap, high capacity coarse-particle processes such as sorting, dense media separation, and coarse particle grinding and flotation, for which good estimates of operating costs on surface exist. In previous studies, pre-concentration operating costs have been estimated at an additional 2.5% of the base cash cost. This is included in the evaluation. The cost of waste disposal is accounted for within the residual backfill cost as described above. Previous investigation indicates that the increased volume of broken ore over in-situ material places a constraint on the volume of waste material that can be practically disposed of underground (Bamber, 2004). Bulking factor for broken ore can vary between 40 – 60%, thus it is calculated that the maximum quantity of ROM ore that could be returned to the mining void is 60% of the in-situ material and thus the model is only valid up to this limit. 5.4 Cost and Revenue Impacts 5.4.1 Operating cost Impacts The operating cost of a mine and mill can be broken down into several categories of cost: • Mine: drilling, blasting, loading, haulage, hoisting • Surface transport • Mill: ore receiving, crushing, grinding, concentration, tailings disposal Costs such as labour, supervision, power, water, reagents, consumables and general services are accounted for but included proportionally into the activity-based costs. Raw operating costs included in the estimate generally include the following: • Power • Water • Labour (incl supervision) 109 • Consumables • Maintenance & spares Power is a function of the total installed capacity of the plant in question, and can be estimated using load and diversity factors as appropriate. Water is generally drawn from groundwater and is accounted for in pumping and storage costs. Labour and supervision is generally proportional to the size of the plant and can be factorised from the throughput. Consumables and maintenance (incl spares) are allowed for at 3.5% of the estimated capital required for the plant (see section 4). Operating costs, while thus influenced by the value of the initial investment, are influenced principally by the following factors (Baxter & Parks, 1957:142): • Orebody geometry and depth – increasing depth increases unit costs by the square of the depth • Mining method and mining rate – affects the total cost of production • Grade – affects the cost of production per unit metal • Geotechnical factors o Improvement in hangingwall and footwall conditions decreases costs o Increase in hardness of ore increases labour, energy, steel & explosive costs • Processing method required (ore dressing, flotation, leaching, SX/EW) • Level of mechanization • Infrastructural issues (power, water, roads) It is generally accepted that increasing the mining rate decreases the unit cost of production by spreading fixed costs over a larger tonnage base. However, increasing the mining rate inappropriately for a given deposit can cause an increase in variable unit costs, which may be masked by the overall impact of increasing production. Several factors limit a simplistic increase in the mining rate: • Financial constraints (development capital) • Geological, geometric, geotechnical and geographical constraints • Mining method • Technological constraints • Raw labour supply • Supervision and logistical constraints on labour • Shaft, hoist & haulage capacity • Ventilation, heating, cooling and pumping capacity • Concentrator and smelter capacity • Market constraints Over and above simply increasing the mining rate, mining companies have a number of other more immediate means to lower costs in the face of tougher competition or market conditions. 110 Such methods include improving productivities through increased mechanization, improving process efficiencies either through improved recoveries or waste management, temporary capitalization of operating costs, and limiting or even suspending maintenance activities. However, some of these techniques, particularly the capitalization of costs, and the suspension of maintenance activities can actually increase unit costs in the long term. It is assumed that the impacts as described above impact proportionally on the operating cost for that process area. The total cost of an operation per ROM tonne can be defined in terms of the sum of the individual operating costs: C = 1n Σ ci ti - (1) T Where C= total operating cost in $/ROM tonne and T = ROM tonnage ci = unit cost of a single process t = tonnage reporting to the process n = number of processes in value chain Thus, for the mining value chain selected for the model: CTotal = cdrilltdrill + cblasttblast + cmucktmuck + chaulthaul + choistthoist + chaulthaul + cmilltmill + ctailsttails And, for a given unit process: c1t1 < c1t2 Where t1 < t2 and ci does not vary significantly over variations in tonnage to the process. In such a relation, costs can be reduced in two ways. The unit cost of the process can be reduced, or alternately the tonnage reporting to the process can be reduced. If the tonnage reporting to a process is reduced in terms of ROM tonnes, the total cost in terms of $/ROM tonne will be reduced. Conservatism is built into the model by the use of a cost adjustment factor, where: c2 = [t2]n c1 - (2) t1 And n is a value from 0-1. In this way, adjustment can be made for the impact of varying fixed operating costs, and small variations in the cost/t over variations in tonnage. For this model, n is set to 0.5 which accounts for the unit cost varying by the square root of the tonnage. 111 5.4.2 Estimation of Capital Costs for Hard Rock Mines Capital costs for mining are dependent primarily on the size, depth, and grade of the deposit and estimation typically requires the establishment of a basic system design and manual and precise establishment of at least the mechanical cost component of the system. However, several parametric capital cost estimation methods are found in the literature for use in such cases. Two models that were identified for the estimation of mining and processing costs are the O’Hara method for underground deposits (O’Hara, 1980), and the model for small underground deposits developed by CANMET (Mular et al 1986). While these methods also include factors for the estimation of surface plant, including pre-concentration type plant, and surface infrastructure, they are considered inaccurate due to capacity and time issues. In the light of this, a more detailed surface plant and infrastructure capital estimation model was developed for use in the research using estimate factors derived from previous capital projects conducted by the author. 184.108.40.206 Estimation of Open Pit Mining Costs The O’Hara method (1980), which is intended for open pit operations has been adapted for use in the open pit model. The estimated cost is dependent on the capacity of the mine and is of the general form: C = ΣCiTpn - (3) Where C = total capital cost Ci = unit cost factor for item Tp = Throughput in ore tons n = scaling factor The overall capital cost for the pit capital can be estimated using the basic O’Hara method as: C = 600Tp0.7 + 5000Tp0.5 - (4) However, this function is considered inaccurate, and more resolution on the costs was desired. O’Hara indicates that this overall cost factor can be broken down into: Pre-production stripping of the pit: C = 8500Tp0.5 - (5) 112 The cost of haul trucks: C = 9000t0.85(0.2Tp0.8 ) - (6) t Where C = cost of truck; t = capacity of haul truck The cost of shovels/scoops is: C = 230000s0.73(0.0525Tp0.8 ) - (7) S Where: S = capacity of scoop in m3 Installation costs for overall surface infrastructure can be calculated as C= 30000T0.6 - (8) 220.127.116.11 Estimation of Basic Underground Mining Costs A more detailed method for the estimation of underground mining costs is presented by Mular et al (1986) in CANMET’s ‘Estimating the Pre-production and Production Costs for Small Underground Deposits’. The model can be used to predict the capital and operating costs of mining and processing based on the geological, geotechnical and geographical setting of the orebody. Similarly, the total cost of exploitation, thus cutoff grade and ore reserve can be calculated directly from this parametric estimating system. Historical data from a range of mining and processing projects undertaken by J.S. Redpath prior to 1986 have been used to develop the model. However, there are a few shortcomings - the estimation model is mining- focused and estimation parameters for plant and surface infrastructure are bulk numbers only, and more sophisticated methods are required. Furthermore, the model is manual and paper- based, comprising tables of values, functions and graphical data, base dated 1986, and is also only valid for mining rates up to 500tpd. The CANMET capital estimation model has been simplified and computerized and used as a basis for this parametric estimation model. The factors presented are for underground development only as a superior plant and infrastructure cost estimation model has been developed through previous work. 113 Table 5.1 Cost Estimating Factors for 500tpd Underground Mine (1986 US Dollars) Quantity Unit Cost US$ Feasibility, Reserve Estimation & Design Feasibility & design 25% $217,391 Environmental & permitting 10% $144,927 Surface Drilling 15000 m $706,521 Cover Drilling 3600 m $117,391 Assay 150 ea $ 1,304 Preliminary & General Roads 20 km $ 1,811,594 Bridge 1 ea $ 72,463 Site clearance 20000 m2 $ 163,043 Infrastructure & Services 1 ea $ 57,971 Power line (33kV; 2200kW) 50 km $ 1,086,956 HT substation 1 ea $ 144,927 Compressed air 2 m3/s $ 181,159 Main office & store 1 ea $ 166,666 Mine access Preliminary & general 1 ea $ 192,028 Shaft collar 1 ea $ 235,507 Shaft & equipping 1500 m $ 8,152,173 Shaft stations 4 ea $ 217,391 Loading pocket 1 ea $ 65,217 Shaft bottom 1 ea $ 10,869 Hoist (2.44m 1000kW) 1 ea $ 554,347 Headgear 1 ea $ 253,623 Loading bin 1 ea $199,275 Conveyances 4 ea $ 202,898 U/G services Ventilation 1 lot $ 25,362 Heating 1 lot $ 32,608 Vent raises (1.8m dia) 1500 m $ 1,347,826 Sumps & pumps 1 lot $ 57,971 Rockbreaker & grizzly 1 lot $ 68,840 U/G substation 1 ea $ 26,811 Miscellaneous 1 lot $ 18,115 Pre-production development 4x3m drifts 308.6111 m $ 322,028 Orepasses (2.2m dia) 200 m $ 191,304 Ore pass controls 1 lot $ 14,492 Mining equipment Mechanised Mining Suite 1 lot $ 688,405 Subtotal $ 21,209,565 Contingency 10% $ 2,120,956 Total $ 23,330,521 114 18.104.22.168 Estimation of Plant & Infrastructure Capital Costs Using the O’Hara method capital costs for the process plant can potentially be estimated from capacity factors: Table 5.2: Capacity Based Factors for Process Plant Capital Cost Crushing and screening 45000T0.5 DMS / Sorting 45000T0.5* Grinding and Flotation 72000T0.5* *Estimated Concentrator building 30000T0.5Fcl Where Fcl = 1 – mild climate/conditions 1.8 – poor conditions 2.5 severe / arctic conditions Outputs of the O’Hara Method can be used directly to estimate costs, thus calculate the economic cutoff grade and ultimately the reserve or pit limits as the case may be (Akbari et al, 2005). In Mine Investment Analysis, Gentry & O’Neil (1984) present more detailed capital cost parameters for a range of operations where: C = Σct - (10) Where C = total capital cost c = unit capital cost t = throughput in tpd And c is taken from a table of values from historical operations. Typically c depends on the nature of the ore and the metal extraction method: Table 5.3– Process Plant Capacity Cost Factors (after Gentry & O’Neil, 1984) Type of Operation C ($/tpd – 1984 terms) SX/EW 10 000 Cyanide leach 9 000 Bulk sulphide flotation 8 000 Dense Media Separation 5 000 However, these methods lack accuracy and more detail is required. In order to overcome these drawbacks, estimates for plant and infrastructure capital from recent studies by the author hae been used to both calibrate the model estimates as well as providemore detail in the estimation of surface plant and infrastructure. Previous estimates used in the calibration are presented in Table 5.4. 115 Table 5.4 – Previous Estimates of Mine & Mill Capital Project Location Base Date Through- put Mine Capital Mill Capital Rhino Minerals Richards Bay SA 1996 750 US$7m Dwarsrivier Steelpoort, SA 1999 3100 US$15m US$20m Kroondal Rustenburg, SA 2000 5000 US$27m Mimosa Zimbabwe 2001 3000 US$35m US$27m Morula Steelpoort, SA 2002 10000 US$100m Voskhod Khromtau, KZ 2006 3000 US$66m US$36m A detailed cost breakdown for plant & surface infrastructure for the typical 3000 tpd scenario is presented in Table 5.5. Table 5.5 – Cost estimation factors, 3000 tpd pre-concentration, grinding and flotation plant 2006 US Dollars. Area Estimate US$ % of Total Crushing & Screening $4,491,469 8.5 DMS Plant $5,505,164 10.4 Product Handling $2,893,412 5.5 Tailings Handling $995,975 1.89 Milling, flotation, reagents and concentrate handling $6,185,000 11.7 Plant Conveyors $2,576,918 4.9 Plant services $6,205,670 1.18 Plant piping system $1,320,551 2.50 Plant power system $5,087,907 9.64 Freight and transport $1,256,102 2.38 Site preparation, bulk earthworks, roads and terraces $2,484,116 4.7 Infrastructure & Buildings $4,865,665 9.22 Electrical buildings $693,295 1.31 Engineering procurement and construction $1,1903,690 22.5 First fill & Spares $654,080 1.24 Contingency $1,256,848 2.38 Total $52 791 398 100 116 22.214.171.124 Adjustment of Capital Cost Estimates for Variations in Mine Capacity and Cost Escalation with Time The above basic estimation parameters are either static in terms of the time value of money, or static in terms of the capacity of the asset for which costs are estimated. Hence the capital estimates produced by the model must be adjusted for 1) variations in throughput and 2) the time lapsed between the configuration of the model and the time of use of the model. For this, two methods are used. For capacity variations, the industry-standard exponential capacity adjustment method is used where the capital cost and date of estimation/construction of a previous operation is known in some detail: Ca = (ta)n - (11) Cb tb Where Ca = cost of operation a Cb = cost of operation b Ta = capacity of operation a Tb = capacity of operation b And 0.5 < n < 0.9 is an exponential scaling factor, typically 0.67 for most operations. This method has the advantage of discriminating between mining, plant and infrastructural capital, where the total capital required for an operation is: Cm + Cp + Ci - (12) And Cm2 = Cm1(t2/t1)0.67 etc. The base date for the factors in the O’Hara pit model is 1980, thus estimates must be adjusted for accuracy using an average escalation factor of 3% for every year between 1980, i.e. O’Hara estimation factors have been adjusted by an overall factor of 2.22 to bring the estimate into 2006 terms. The base date for the CANMET underground estimate is 1986 and a similar adjustment is made for this base date. The revised plant and infrastructure parameters are based- dated 2003. The model then accounts for variations in the future value of money between 2006 and the present by means of a user-input escalation value. Impacts to overall capital costs are in proportion to the change in tonnage reporting to that process area in accordance with equation (11). 117 5.4.3 Revenue Impacts 126.96.36.199 Impacts on Recovery The impact on revenue of the pre-concentration step is derived from the impact on the recovery of metals in the entire process. Metallurgical recovery is defined as the percentage metal recovered from a particular metallurgical process. Non-recovered metal reports to tailings. Thus in a surface mill, for a given metal, the material balance according to the 2-product formula is as follows: mfgf = mcgc - mtgt - (13) And mf = mc + mt - (14) Where mf, gf – mass and grade of feed mt, gt – mass and grade of concentrate mt, gt – mass and grade of tails Recovery across the plant is the ratio of metal in the concentrate to metal in the feed, thus: R = mcgc -(15) mfgf Pre-concentration introduces an additional recovery penalty on the beneficiation process in accordance with equation (15). There are also inherent metal losses to be accounted for in the performance of the existing mill. However, for a given mill, metallurgical recovery would improve with improving feed grade based on the assumption of a constant grade of tailings. This assumption is borne out in several places in the literature, and has been included in the model in order to evaluate the impact on overall metal recovery of the pre-concentration step. Thus actual overall recovery is Rt = Rp R’m, where R’m is the improved recovery in the mill based on the new feed grade (Figure 5.2): 118 Pre-concentration Surface Mill 100% Feed Preconcentrate 80% Distr 96% Concs 5% Distr 93% Tails 20% Distr 4% Tails 75% Distr 3% Figure 5.2 Synergistic Impact of Pre-concentration Step on Overall Metal Recovery Other user-input models are found in the literature, for example the variable recovery model for Musselwhite Mine presented by Blower & Kiernan (2003), that can be substituted for the 2-product equation. The phenomenon can be represented in an equation of different form where: R = X – (Y + 100*gt ) - (16) gf Where X = maximum achievable recovery in the plant Y = recovery factor (shape of grade/recovery curve) And gt and gf are feed and tails grade in g/t as defined previously Substituting for values in the equation gives the same general form as the 2-product equation as shown in Figure 5.3: 119 Musselwhite Mill Performance 0 20 40 60 80 100 0 2 4 6 8 Mill feed grade g/t M ill re co ve ry % Recovery 2P Recovery Eq Figure 5.3 - Variations in plant recovery with varying feed grade Different models are indicated where the ore is low grade, highly diluted, or contains additional deleterious elements to flotation or leaching such as talc or silica. CVRD-INCO’s Thompson Mill in Thompson, Manitoba, accounts for mill recovery in terms of the feed grade according to the relation (Penswick, 2007): R = 93% – (16.1/g)*96.1% - (17) A variation in the feed grade of 0.7 – 1.1% Ni thus results in a recovery variation of 71% - 79% respectively. This relation also follows the general form of the 2-product formula presented in (13), however the variation of mill recovery with feed grade is greater than in the 2 previous models. The variation of mill recovery with feed grade has been built into the model as a general case. Metal losses in smelting are to be accounted for; however, smelting tolls generally include for the cost of smelting and refining as well as metal losses (which are usually negligible in terms of ROM grades and tonnages), and are paid on the basis of contained metal in the concentrate. Thus, on the basis of a 20% toll smelting fee, the actual recovery penalty from pre-concentration is of the order of 80% of that calculated in terms of the raw recovery, which positively offsets the raw cost savings (as calculated in 5.2 above) against the value of lost metal. Values for smelting losses as well as toll fees and NSR can be input into the model. 120 188.8.131.52 Variation in Metal Price Variations may also occur in the evaluation based on variations in the metal price. This can be accounted for by dynamic metal price functions which will allow the user to use variable metal pricing models throughout the life of mine. Generally, commodity markets in general, and base metal prices in particular show a long term decline in real price* (Krautkraemer, 1998; Slade, 1982). However, metal markets, and thus prices fluctuate continually in the short term based on market perceptions, demand cycles, inventory levels and producer responses. The market cycle has inherent delay and inertia resulting in often confusing trends, and difficulty in prediction. Precious metals, and certain base metals sometimes respond rapidly to external influences, and while present market conditions indicate an extended bull run for the general suite of commodities, present commodity price levels would be reasonably expected to decline at some point in the future. Short term mineral commodity cycles are thus a real and present issue at this time of historic high metal prices, and previous cyclicality indicates that a downturn at some stage in the near future would be expected over a life of mine of 20 years, and thus consideration of this factor could be considered important in the evaluation. In order to illustrate this, Figure 5.4 (opposite page) shows a depiction of a general commodity market cycle with a description of salient features (after Baxter & Parks, 1957; 139). In the low cycle, the industry is characterized by several key behaviours – costs are often expensed to capital, management and labour effort is high, and innovation is required. In the high cycle, commodity prices are high, profits are maximized and management and labour becomes lax – the pressure to innovate is low; however, input costs are also high, thus unit profits are lower than at other times, thus innovation can still be required to maintain profit margins even in times of boom. Several simplistic models including price constant, constant decrease, constant increase can be considered. More complex models are fully dynamic models of a particular commodity market, involving interactions of commodity price, producer capacity and production cost inventory and market demand. * http://tin.er.usgs.gov/mrds/ 121 Figure 5.4 - Characteristics of the mineral commodity cycle (after Baxter & Parks 1957) 122 Such a model has been constructed using the StellaTM Systems Analysis (Bamber & Dunbar, 2005; O’Regan & Moles, 2001) to evaluate these dynamics over the life of a typical mineral project. The model generally indicated sinusoidal price dynamics as indicated by the observations of Baxter & Parks (1957). Production costs appear to track increases in commodity price with a small delay, indicating a relatively constant long term cost-to-metal price ratio, thus in the parametric model the metal price and cost margin is assumed to remain constant over the evaluation period. 5.5 Impact of Cost and Revenue Variations on the Cutoff Grade, Mineral Reserve and Mineral Resource The impact on the economic cutoff grade of a deposit with pre-concentration is assumed to vary in direct proportion to the impacts on working cost and the overall recovery of the valuable mineral. An important measure of cutoff grade is the measure of the quantity of the valuable mineral extracted to metal vs. the total amount of metal contained in the resource, the extraction. This is a very important measure of sustainability for mineral resource exploitation, and calculation of this is considered an important outcome of the model. Firstly the ore resource must be defined and quantified for a given cutoff grade. Secondly the cutoff grade for the calculation of the ore reserve is defined, and the ore reserve estimated. Estimation of the ore reserve allows for validation and/or recalculation of the relevant cutoff grades. 5.5.1 Estimation of the Cutoff Grade and Ore Reserve Ore resource and reserve estimation are essentially estimates of the probability of occurrence of the ore, coupled with an estimate of the economic feasibility of extracting the ore, to increasing levels of confidence. In Canada, National Instrument 43-101 gives definitions for the physical and economic definition of mineral resource probabilities (Jensen & Bateman, 1981; 4). A mineral resource is the definition of probability of occurrence of the ore, and can be classified as proven or probable. The mineral reserve is the portion of the mineral resource which is considered economically feasible to extract, and can be classified (in order of decreasing confidence) measured, indicated or inferred. Inferred reserves are portions of the probable resource which may be economic to extract. Assuming that a drilling programme has commenced, and that a resource has been defined, the first step in valuing and evaluating a mineral property is to establish parameters for the size and grade of the ore reserve. An ore is composed of several components which make up the 123 total volume of material to be mined. The metal and associated elements such as sulphur make up the mineral fraction. The mineral fraction plus the gangue minerals such as silica make up the ore itself (Jensen & Bateman, 1981;41). A metal may be associated with many minerals (eg chalcopyrite, chalcocite), or conversely many metals can be present in a single mineral fraction e.g. stannite. The ore reserve is that fraction of the ore resource which is above cutoff grade and can be economically extracted. For the model, the cutoff grade relation selected follows the general form (Wheeler & Rodrigues, 2002): gc = (civar + (Cifixed+Copp)/T) - (18) Price x R Where Civar - variable operating costs/t Cifixed – total fixed costs/annum Copp – opportunity cost in $ T – ROM tonnage/mill throughput Price – metal price in $ per unit mass R – overall metal recovery For an operation with a low proportion of variable to fixed cost, and a low opportunity to total cost ratio, this reduces to: Gc = (F+V) -(19) PR Where G - grade in g/t or % F - the fixed cash cost /t V - the variable cash cost/t P - metal price in $/g, $/kg as appropriate R - overall metal recovery The cutoff grade thus varies in direct proportion with the magnitude of cost saving and the impact on overall metal recovery (Wheeler & Rodrigues, 2002). The tonnage factor is used to establish the total tons of rock which must be extracted to obtain the metal value in the ore. This is the mineral resource. Table 5.6 – Ore Tonnage Estimation Factors SG mineral Ore SG ‘tight’ Rock SG Ore bulk SG SG gangue + porosity (%) + moisture (%) xBulking factor (%) 124 We must thus define the overall tonnage factor in order to define the mineral resource: TF = Ore SG - (20) Rock SG The rock has a certain average metal content, which is the grade of the economic ore. The total metal contained is the mass of the resource multiplied by the grade of the deposit. There is a certain basic cost of extracting and treating one ton of the resource into a saleable product. This cost will determine the basic cutoff grade for determining the resource. The proportion of the resource above cutoff grade which can then be profitably mined under varying market conditions is the mineral reserve. It must be noted that not all ore in the reserve can be feasibly extracted. Ore is often sterilized by geological features such as faults, dykes and ‘potholes’ – pinching of the width of the ore to uneconomic widths between drill holes. Geological losses can be as high as 15% of the reserve tonnage. Ore can also be sterilized through the mining method chosen – ore is lost in pillars left in the stope for roof support, shaft and sill pillars, as well as losses arising from constraints in the layout of ramps and drifts. Mining losses can be as low as 10% for open stoping methods and as high as 20% and more for room-and-pillar methods. Once the mineral reserve is defined, the total amount of ore that can be extracted from the reserve must be estimated from geological and mining factors. For the purposes of estimating the reserve, the level of dilution of the ore must also be defined. As previously mentioned, a certain proportion of gangue minerals are already included in the definition of ore. Additional gangue material may be inadvertently mined from sloughing of the hangingwall and the footwall, thus the grade of ore defined in the reserve model becomes diluted through actual mining of the ore. Thus the total amount of dilution in the deposit is: %100×⎥⎦ ⎤⎢⎣ ⎡ +=∂ toretwaste twaste - (21) However, dilution may simply be ore in the resource that is below the cutoff grade of the reserve, and thus still contains metal, the recovery of which must be accounted for in the economic analysis. The loss of metal in fines in the ore must also be accounted for. Often the percentage dilution is overestimated to compensate for this, which is erroneous. Alternately, companies account for this directly with factors such as the ‘Mine Call Factor’ (per Anglo American) or ‘Mill Cuts’ (per INCO) to account for this loss in metal, which is considered a more correct approach than discounting these losses. An example of a complete ore block 125 definition for the South Deep Mine based on the above discussion is presented in Table 5.7 for clarity: Table 5.7: Parameters for Hypothetical 30 000m3 Gold Ore Reserve Variable Unit Value Volume of ore (from orebody model) m3 30 000 Ore SG t/m3 3.1 Tonnage factor % 10 Rock SG t/m3 2.9 Ore grade g/t 7.7 Cutoff grade g/t 3.6 Metal price $/g 15 Mineral reserve t 87 000 Total metal content g 669 900 Total metal value $ 10 048 500 Geological losses 10% Mining losses 10% Total reserve tonnage 70 470 Value of ore reserve block $ 8 139 285 F/W + H/W Dilution % 20 Waste grade g/t 2.9 Total tons mined from block t 84 564 Grade of block g/t 6.9 Mine Call Factor % 5 Head grade of ore g/t 6.555 Value of ore to surface $ 8 314 755 Let us assume that 50% of the total ore mined can be rejected to waste at a recovery of 95% in the pre-concentration plant. The following assumptions pertain: • The impact on economic cutoff grade is proportional to the decrease in operating costs • The decrease in ore reserve grade arising from a reduction in cutoff grade is smaller than the decrease in cutoff grade • The proportional increase in size of mineral reserve is expected to be greater than the decrease in cutoff grade • More ore is thus extracted from the overall resource • Gangue material included in the definition of ore, as well as gangue material included in the definition of dilution will be rejected • Metal values contained in the dilution will be retained 126 • Loss of metal in fines usually lost through conventional mining and handling of the ore to surface are expected to decrease With pre-concentration the tonnage of ore delivered to surface from the block is reduced to 42 282t, at an increased head grade of 13.66 g/t. Cutoff grade is now the grade of the waste rejects from the pre-concentrator. The size of the reserve block is increased to include material previously defined as waste and thus, the overall extraction of the mineral resource, as well as the recovery of metal values from the mineral reserve is expected to increase with the application of underground pre-concentration. Cutoff grade thus varies in direct proportion with the magnitude of cost saving and the impact on overall metal recovery. This value is calculated in the model and used to calculate the increase in ore reserve for the purpose of revision of the life-of-mine estimate based on an idealized grade-tonnage curve for massive sulphide deposits, although variations to the idealized grade-tonnage curve can also be accounted for in the model. 5.5.2 Cost-Cutoff Grade Interactions It has been shown in the testwork that pre-concentration can typically reject up to 60% of the ROM ore by mass at recoveries in excess of 95% (Bamber et al, 2006; Munro et al 1999, Schena et al, 1999). This can be undertaken on surface with significant impacts, however undertaking such pre-concentration underground facilitates the use of higher productivity mining methods and also rejects a substantial amount of material prior to hauling and hoisting and milling with significantly increased impacts and benefits. Metal recoveries are maintained through high recoveries achieved in the pre-concentration step, as well as increases in the recovery at the mill arising from the increase in the feed grade. Based on previous research and testwork, the model indicates that costs can thus be reduced by between 20-30% through a pre- concentration of the ore 30-60% by mass. From the basic relationships presented previously in the Chapter, we can examine the impact of such a change in operating cost and recovery on the cutoff grade and size of a number of real deposits for which we have data. Mining is typically high in capital intensity, thus fixed operating costs are high and necessitate high production rates to sustain the high cost of fixed capital. The combination of fixed costs, variable costs and the unit value of the ore give an indication of the mining rate at which an orebody will be profitable (break-even): 127 $ Throughput F ixed cost Var iabl e co st Re ve nu e Figure 5.5 – Break-even Production Rate An increase in fixed costs can penalize an operating mine by increasing the margin of loss for no production, and raising the minimum tonnage at which the mine breaks even. Similarly an increase in variable costs also raises the minimum tonnage at which a mine breaks even. $ T h ro u g h p u t F ix e d c o s t V a r iab l e co s t V a ria b le c o s t 2 Br ea ke ve n 2 Figure 5.6 – Change in Breakeven Production Rate A combined decrease in fixed and variable costs can result in substantial operating improvements at the mine, lowering the operating loss during periods of no production, lowering the tonnage at which the mine breaks even, and increasing the profit margin per ton of production. Pre-concentration of the ore underground has been shown to lower variable costs directly as well as reducing fixed costs through a reduction in shaft and surface capital requirements for a mine of the same production rate. Impacts on the cost profile of the mine are significant: 128 $ T h ro u g hp u t F ixe d co s t V a r iab l e co s t F ixe d cos t 2 V a ria b le c os t 2 R e ve nu e 2 B re a ke ve n 2 Figure 5.7 – Impact of change in Fixed and Variable costs on Breakeven Rate The impact of lowering the cutoff grade reduces the break even mining rate as well as increasing the profit margin for all tons produced above the break even point, and is thus projected to have a massive impact on the exploitation of a mineral deposit. 5.5.3 Cutoff Grade and Grade-Tonnage Interactions for an Idealised Mineral Resource Consider a generic mineral deposit of particular mineralogy, metal grade, grade distribution and arbitrary tonnage (after Lacy, 1969): t g Figure 5.8 – Idealized Grade Tonnage Curve The curve has several interesting features for analysis. It indicates that small resources are generally high grade and that the converse is also the case. An alternative indication, or co- observation is that there are a very low number of small, high grade deposits an increasing 129 number of large, low grade deposits, and an almost infinite number of mineral deposits at zero economic cutoff. The curve is also exponential, indicating that a unit decrease in cutoff grade represents a disproportionate increase in the mineral resource: ta > ga - (22) tb gb Similar interactions would be seen for an economic mineral reserve. As cutoff grade is proportional to the operating costs for the mine, it can be seen that, all other things being equal, a unit decrease in operating costs will lead to a substantial increase in tonnage, and therefore mine life (Clark & Harper, 2000). Grade-tonnage data from the literature can be used to construct a generic grade tonnage curve in order to confirm this model. Figure 5.9 is a graph of grade-tonnage data published for the Navidad Hill Resource in Nevada as at March 2006 (Patterson, 2005): Navidad Hill Grade Tonnage Curves 0 50 100 150 200 250 300 350 400 450 500 0 20000 40000 60000 80000 100000 Tonnage (kt) G ra de (g /t) Calcite Hill Navidad Hill Connector Zone Galena Hill Total Figure 5.9 – Real Grade-Tonnage Relations for the Navidad Hill Deposit Regression analysis of the data indicated that the grade tonnage curve can been fitted to an exponential equation of the form: y = Ge-tx - (23) Where y = the grade of the deposit for a given tonnage; G = the maximum estimated grade within the resource; and t’ = the slope of the grade/tonnage curve where t’ ~ 1 x 10-4 for the smaller, high grade Connector deposit and t’~1x10-6 for the larger low grade Galena Hill deposit. A high correlation (R2 > 0.96) between actual data and the model is indicated. Similarly, corroborating data from several global base- and precious metal deposits has been 130 taken from the literature and statistically analyzed and is presented in figure 5.10 (Jensen & Bateman, 1989; Singer, 1995): y = 14.44e-4E-08x 0 5 10 15 20 25 0 1E+07 2E+07 3E+07 4E+07 5E+07 6E+07 7E+07 Deposit size (t) G ra de g /t y = 1.2821e-4E-10x 0 1 2 3 4 5 6 0 1E+09 2E+09 3E+09 4E+09 5E+09 6E+09 7E+09 Deposit size (t) G ra de % Figure 5.10 – Grade/Tonnage Curves for Selected Global Gold (above) and Base Metal Deposits (below) Again, the data fits an exponential model of similar form, where t’ ~ 10-7 for the gold deposits and t’~10-9 for base metal deposits. Using such a generic model, the impact of changing cutoff grade on reserve tonnages can be estimated for any given resource given a single data point for resource size and cutoff grade. At this stage of development, this simple grade-tonnage model 131 for large low grade and small high grade deposits is used for the model. The combined impact of a projected operating cost reduction is shown on the generic grade-tonnage curves thus developed: a t b c1 c2 ta1tb1 tb2 ta2 Figure 5.11 – Cost Impacts on Grade and Tonnage Curve Curve a is for a large, low grade disseminated orebody with an exponent t of 10-9. A decrease in cutoff grade resulting through a decrease in operating costs massively increases the mineral reserve. Curve b is a smaller, high grade deposit with an exponent of 10-7. Increases in the mineral reserve are less significant; however, the operating margin is substantially increased, massively increasing the present value of the orebody. Using this mechanism the impact cutoff grade of the ore reserve by means of these curves is evaluated in the parametric model. Scaling factors and exponents are user-adjustable in order to explore different grade-tonnage scenarios for improved accuracy. 5.6 Impact Valuation and Evaluation Methodology The flowchart for the parametric evaluation model developed for the research thus has several key elements. Basic data for the deposit to be evaluated is obtained and input into the model. Operating parameters for the present mine and mill if pertinent are also input. Results from the pre-concentration testwork are input to evaluate the impact of pre-concentration versus the base case. Activity–based operating costs are input by the user. Capital costs, and impacts to capital and operating costs for the base- and pre-concentration scenario based on the degree of waste rejection are automatically evaluated. Revenue impacts are calculated as described, and overall impacts on the size of the mineral reserve, life of mine, and Net Present Value of the deposit are automatically calculated. It is assumed that variables such as depth, grade, etc. of a 132 given mine are catered for within the cost structure of a given mine i.e. extreme depth translates into a commensurately high hoisting cost component in the input model. The cost saving is translated into an equivalent cutoff grade which is used to calculate the increased ore reserve. Mining rate is considered constant for the purposes of the model, thus the life of mine is extended. An alternative model would increase the mining rate to accommodate the increased reserve, although this would impact negatively on the sustainability of exploiting the deposit. A flowsheet of the valuation and evaluation process incorporated in the model is presented in Figure 5.12. Situational Parameters: location, topography, infrastructure, project start date Drilling/exploration parameters: depth, shape, strike, dip, grade of ore, geotechnical data Market parameters: metal prices, interest rates, risk factors, capital supply base date Database: - dimensionless operating cost factors dimensionless capital cost factors previous mining estimates previous plant estimates previous infrastructure estimates tonnage scaling factors base dates Scaling relations: - operating cost functions - capital cost functions - situational functions -depth, grade, dip strike - geotechnical functions Parametric Estimator Operating cost model Capital cost model Project Design Parameters: - access method - mining method - extraction/losses - processing strategy - metallurgical results - recoveries Figure 5.12 – Preliminary Parametric Valuation and Evaluation Flowchart As has been indicated, the preliminary size, grade and thus value of the ore deposit has now been established, and the capital required to initiate mining and processing and the operating costs to continue mining and processing have been estimated. The value of the ore is typically determined in US$/t, and operating costs, which can be broken down in to fixed and variable plus distribution, sales and marketing costs and a contingency, typically also in US$/t. The overall cash flow for a mining operation, which is the basis of all the succeeding evaluations 133 can be determined from basic accounting principles as shown in Table 5.8. As can be seen 8from the table, the net operating cash flow is influenced by every aspect of the operation, from ore value, to process efficiency, the nature of toll smelting agreements as well as initial and ongoing capital requirements for the mine. Table 5.8 – Net Operating Cash Flow Breakdown Ore value/t Less Recovery/losses Times Production rate Equals Revenue Less Royalties Less Operating costs Less Smelting toll Less Depreciation & amortization Equals EBIT Less Tax Plus Depreciation & amortization Equals Operating cash flow Less Working capital reserve Equals Net Annual Cashflow All parameters required for the cash flow evaluation are calculated by the model and the resulting cash flow analysis on an annual basis over the life of the mine is calculated giving the basic undiscounted cash flow profile of the mine. This is used as a basis for all subsequent evaluations. The Net Present Value of the ore reserve is calculated using operating costs and metal recoveries for the conventional mining scenario. The equivalent NPV for the deposit considering the integrated mining and pre-concentration approach is calculated using the revised operating cost (calculated on the basis of the tons rejected), and the revenue impacts calculated from the recovery data for the pre-concentration process, and a capital estimate for the pre-concentration facility as described above. The estimated increase in ore reserve and thus life-of-mine is also accounted for in the calculation. Capital estimates for the mine and or shaft development, surface mill and infrastructure as well as the pre-concentration facility is calculated using the methodology described above. The NPV of the scenarios are compared parametrically to evaluate the impact of pre-concentration on the deposit. Once the size, location, orientation and value of the deposit is established, metallurgical performance as well as capital and operating costs have been determined, it is possible to make an economic valuation of the potential mining operation that is envisaged. Mining operations 134 are characterized by high capital intensity, long pre-production lead times, high risks (e.g. geological, political, technological and market risk) as well as an irreversible depletion of the major asset, the mineral resource. Thus the valuation of mineral deposits is a specialized form of general investment analysis which must take these particular characteristics into account (Gentry & O’Neil, 1984). There are many techniques available to do this, both historical and current, including the simple replacement cost method, market (sales) value methods and earnings-based methods: Net Present Value (NPV), Internal Rate of Return (IRR), and payback. The model incorporates a probabilistic NPV estimation based on pre-programmed variations in capital requirements, operating cost, metal price (grade) and a range of discount rates in order to produce the NPV spreads. The asset efficiency of the comparative scenarios is also examined. The economic efficiency of an asset is a measure of the comparative return on investment on the asset (ROA) to the size and capacity of the asset (Russell, 2003): ROA = Net_Operating_Income x Revenue - (23) Revenue Net_Assets And Net_Operating_Income = Revenue - Costs - (24) Sales Thus ROA = (Revenue – Costs) x Sales -(25) Net_Assets The measure evaluates whether the asset is performing efficiently or not. It is expected that the pre-concentration of ore underground will increase the efficiency of the mining asset as the operating margin is vastly increased for a marginal increase in asset value. From analysis of the equation, there are a number of ways to increase ROI: Increase revenue Increase production volume and/or utilization of the asset Decrease costs Reduce the asset base The impact of the pre-concentration and waste disposal scenario on the overall cashflow is thus determined in the model. The sum of these cashflows must then be evaluated in terms of their Net Present Value (NPV) in order to determine their present value and thus decide whether an investment in the concept is to be made. A suitable method of evaluation is the 135 NPV ‘profile’ method, where the robustness of the project cashflows are tested against scenarios of varying capital and operating cost, metal recovery and metal price. This method highlights the benefit of small, long-term cashflows with a small initial investment as well as large initial investments with large short-term cashflows, and exposes ‘vulnerable’ projects for which the NPV rapidly diminishes with a small increase in initial investment, capital costs or a decrease in metal price or recovery (Barnett & Sorentino, 1994). It is also considered a good method of accounting for project risk by use of these risk-adjusted values for recoveries, capital and operating costs. All other project variables appear to have a second order effect on the viability of the project. A screen plot of a typical NPV ‘profile’ is presented for the Pipe II deposit (Figure 5.13). This process is undertaken for the project base case as well as the pre- concentration case and the NPV and NPV profile of the two options is be compared. Further comparative analysis of the various project investment options, and the timing of these options can be subsequently undertaken by objective analysis or by some more sophisticated method such as option-exercise pricing, or real options theory (Winsen, 1994; Samis, 1993). Details of two previously developed parametric models for a surface and underground scenario can be found in Appendix 1a and 1b. -150000 -100000 -50000 0 50000 100000 150000 200000 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20C um N P V ($ '0 00 ) Years Preconcentration Cost +-10% Capital +-10% Metal Price +-10% Discount Rate +-5% Base NPV Figure 5.13 – Examination of the Robustness of the Project using the NPV Profile Method 136 5.7 Conclusion A spreadsheet-based model for the financial evaluation of pre-concentration compared to a base case of no pre-concentration of the ore is presented. The size and grade of the deposit must be chosen, and a mining rate selected. Expected operational data for the deposit as well as any mill testwork and results from the pre-concentration testwork are entered as input variables on the sheet, and cost and revenue impacts are calculated in the ‘Output’ sheet. A capital sum for the initial mine and mill development as well as the pre-concentration facility is calculated. The net saving in operating cost due to pre-concentration is calculated, and impact on revenue through changes in metal recovery are also calculated. These outputs are used to calculate the overall impact on the cutoff grade and thus the size of the mineral reserve using a generic grade-tonnage model. The NPV and NPV profile of the two scenarios is calculated, with variations in basic parameters such as operating cost, capital requirements, metal price or grade as well as discount rate. Cost and revenue impacts are evaluated and compared to the baseline to establish a positive or negative impact overall. Different scenarios of grade, tonnage, mining rate, pre-concentrator performance, capital and operating cost structure can be explored using the model. For the case studies presented in the thesis, estimates and outcomes projected by the model have been calibrated against internal estimates by the sponsoring organizations and found to be accurate to within 30%, confirming the utility of the model in practice. 137 6. Integrated Mining, Pre-concentration and Waste Disposal Case Studies 6.1 Introduction UBC has been involved with research into Mine-Mill Integration and underground pre- concentration since 2000. Fundamental and speculative papers highlighting the potential benefits of the approach were previously published (Peters et al 1999; Scoble et al 2000; Klein et al 2000, 2003). In 2001 UBC Mining entered into a research agreement with INCO Ltd with the objective of investigating appropriate technologies for pre-concentration, and considering the application of these technologies in a conceptual underground facility. Substantial work has since been completed, including an extensive literature review and the investigation of over 26 case studies for the application of pre-concentration, both on surface and underground for a variety of mining operations and mineral deposits. Since the initial case study work was completed in 2004, research in the field has been ongoing with an expanded team of researchers: the scope of the research has been further expanded to include the development of advanced mineral processing techniques as well as the investigation of waste disposal aspects. Metallurgical and modeling results from earlier case studies at INCO’s Sudbury operations have been previously published (Schindler 2001; Bamber, 2004; Bamber et al 2004, 2005, 2006; Scoble et al, 2006), however, selected results are presented in this Chapter for a more comprehensive analysis of the concept. In 2006, the initial underground pre-concentration work at INCO (now Vale) was followed up by a detailed ore sorting study for the Pipe II deposit in Thompson, Manitoba. Additional case studies have now been undertaken for Falconbridge (now Xstrata), and Placer Dome. Work at Placer Dome has not continued since their take over by Barrick Corp. However, following encouraging results, the Xstrata study was expanded to include investigation of pre-concentration and waste disposal strategies for 9 ore types. The study has now been continued with detailed investigations for two of Xstrata’s mines, Onaping Depth and Nickel Rim. The scope of the research to date has thus included the sampling and testing of over 25 ores, including, inter alia: massive vein sulphides; massive and disseminated sulphides; banded sulphides; volcanogenic gold-pyrite deposits; paleo-placer gold deposits of the Witwatersrand Basin associated with both pyrite as well as uraninite; lead zinc ores of the Mississippi Valley type (MVT), as well as preliminary investigations into the pre- concentration of Cordilleran-type copper porphyry ores. Preliminary system design, configuration and layout of the generic integrated mining, processing and waste disposal 138 system as presented in Chapter 3 has also been conducted for several of the cases, allowing a more detailed examination of the impacts and benefits of the concept for these deposits. This has enabled the evaluation of a complete suite of potential impacts and impacts and benefits of the concept in terms of mining method, capital costs, operating costs, cutoff grade, mineral reserve extraction and surface environmental footprint to be evaluated. A summary of the results and outcomes of these studies, as well as additional supporting data from the literature where relevant, are presented in this Chapter as further evidence in support of the proposed approach. Details of the case studies are presented in Appendices D-F. It is suggested from the extent of these results that the potential for the pre-concentration of ore and the subsequent disposal of the waste in the mining void is a general case and should be considered as an option for the exploitation of a deposit as early as possible in the development cycle. 6.2. Gold Ores Gold-pyrite and Witwatersrand/Paleo-placer Deposits An underground pre-concentration study was conducted for selected operations at Placer Dome Group in 2005 (Bamber, 2005). Operations studied were the South Deep Mine in Carletonville, South Africa, and the Musselwhite Mine in Ontario. Testwork was undertaken on selected ore samples. The initial evaluation model was developed for this project and used to estimate potential impacts and benefits at each operation based on the results of the testwork. Previous research had indicated that up to 60% of the Witwatersrand type gold ore could be rejected through coarse particle flotation and gravity concentration at a gold recovery of 98% (Lloyd, 1979). Results from testwork by Placer Dome during the project indicated that 50% of the ore by mass could be rejected at a metal recovery of 88% through radiometric sorting. A modification of the process design resulted in the ultimate rejection of 15% of the ore at 98.7% recovery (Kowalcyk, 2002). Capital cost for the sorting plant and excavations were estimated at US$90m (2005 terms). Using the evaluation model, these results translate into a decrease in operating cost from $78 to $71/t, and a commensurate lowering of cutoff grade at the mine and an increase in the ore reserve (Bamber, 2005), and an increase of $104m in the NPV of the mine over 20 years. For the Musselwhite Mine, the testwork indicates that 24% of the ore could be rejected by dense media separation at 94.6% Au recovery. An estimate for a dense media separation plant for the mine was made at US$30m (2005 terms). Modelling of the results indicate a decrease in operating cost from $61.50/t to $55.03/t; however, metal losses, at 5.4% are of the order of this cost saving, and the impact on NPV predicted by the model was 139 marginally negative overall: Musselwhite Mine is essentially a shallow, high-grade, low cost producer, and the impact of pre-concentration was marginal for the mine, indicating a very close dependency between ore value and operating cost structure for the success of the concept (Bamber, 2005). Meta-sedimentary gold Testwork was been conducted on samples of gold ore from International Wayside’s operations near Barkerville, BC. The material tested was near-surface, sedimentary-hosted, high talc gold- pyrite ore at nominally 14 g/t, and was not initially considered an obvious candidate for pre- concentration. However, the mine intended direct shipping the ore to an existing mill some 50km away in Quesnel and dense media separation testing on the ore at -19mm indicate that up to 48% of the ore by mass could be rejected at a gold recovery of 97.4%. A preliminary estimate of the feasibility of implementing ore pre-concentration at the mine on this type of ore was positive (Weatherwax & Gillis, 2006). 6.3 Mississippi Valley Type Lead Zinc Ores In 2007, a preliminary study into the pre-concentration of Mississippi Valley type (MVT) lead zinc ores was conducted for the Doe Run Company in Viburnum, Missouri. A combination of hand samples, core samples and ROM samples were taken for analysis. The results of previous Dense Media Separation and basic size assay data were also obtained for the evaluation. Doe Run is situated on the Viburnum trend in SE Missouri, which is a Pb-Zn deposit hosted in the Mississippi Valley ‘Bonaterre’ formation. Doe Run currently operates a complex of 6 underground mines and 4 mills along the trend, with a future planned mine at the Higden deposit near Farmington, MO. Total present reserves are estimated at 60Mt @ 6% Pb, 1% Zn and 0.15% Cu at a cutoff of 5% Pb equivalent. Mining is generally by sequential room and pillar mining with the balance of production by planned extraction of the remnant pillars. Mineralogy is generally massive galena, sphalerite and lesser occurrences of disseminated chalcopyrite in flat-lying seams hosted in dolomitic sediments of the Bonaterre formation. Pb is associated either with Zn or Cu but rarely both. Gangue minerals are CaCO3, pyrite and marcasite, with a typical combined density of 2.86 (Paradis & Hannigan, 2007). Two major ore types are identified, high grade ‘banded’ ore comprising perhaps 20% of reserves and lower grade ‘breccia’ type ore comprising the balance. Higden ore is situated lower in stratigraphy than Doe Run’s present operations and is hosted in weak sandstone and clays (reported 5-15% 140 clay). Doe Run Co. is currently forced to mine high-grade reserves including high-grade remnant pillars in order profitably meet the requirement of historical Pb supply contracts. Doe Run thus desires to increase production of reserves below the present 5% Pb eq cutoff. Options for the Doe Run expansion are: • to change the mining method to a more selective method on lower grade sections, • increase hoisting and milling capacity • increase ore production by means of pre-concentration. Furthermore, as Doe Run is situated in the Mark Twain National forest, the operations are also constrained in terms of coarse and fine tailings disposal, as well as any further surface infrastructure, hence the option of underground pre-concentration was considered attractive. Hand samples of both breccia and banded ore types were taken during the underground visit. Photographs of several stope panels, as well as close up photographs of the ore were taken for mesotextural evaluation. From these images, it appears that the galena and sphalerite mineralization is clearly visually distinguishable at the meso-textural as well as micro-textural scale (Figure 6.1, 6.2 & 6.3). Figure 6.1 – Banded ore of the MVT type at Doe Run Company showing horizontal orientation and clear mesotextural characteristics 6000mm 141 Figure 6.2 – Close up of breccia type ore in-situ showing sharp contact with the dolomites Figure 6.3 – Close-up of Doe Run breccia-type ore showing both massive and disseminated sulphides Initial indications from previous sink-float testwork results on ore from the Buick and Viburnum properties (Jones, 2006) include a clear upgrading effect in the finer fractions, as well as good metal recovery to the sinks fraction in each case (Table 6.1). Preliminary work on a 5kg hand sample of Brushy Creek ore confirms the pre-concentration potential of this ore. Concentration by means of dense media separation as well as concentration based purely on size is indicated in the preliminary assessment, with potential rejects of between 32.7 – 64.6% of the ore by mass on average with an expected metallurgical recovery of 98% Pb. 20mm 142 Table 6.1 – Results of Doe Run Ore Evaluation (after Jones, 2006) Product Wt% Pb % Zn % Cu % Pb Wt% Zn Wt% Cu Wt% Viburnum Ore 100.00 3.31 0.50 0.17 100.00 100.00 100.00 +9mm 72.90 1.77 0.31 0.11 38.92 45.26 47.39 ‐9mm 27.10 7.46 1.01 0.33 61.08 54.74 52.61 2.81 sink 64.60 5.05 0.72 0.26 98.56 93.02 98.80 Buick ore I 100.00 3.89 2.09 0.01 100.00 100.00 100.00 +6.7mm 88.80 2.73 1.99 0.01 62.40 84.41 66.40 ‐6.7mm 11.20 13.06 2.91 0.03 37.60 15.59 33.60 2.87 sink 35.40 9.84 4.62 0.02 89.55 78.25 70.80 Buick ore II 100.00 2.19 0.40 0.02 100.00 100.00 100.00 +13mm 65.10 1.36 0.28 0.03 40.56 45.91 82.55 ‐13mm 34.90 3.73 0.62 0.01 59.44 54.10 17.45 2.85 sink 67.30 3.19 0.57 0.02 98.03 95.90 67.30 Other separation options are to be examined as both sphalerite and galena are conductive minerals, with distinct photometric properties with respect to the dolomite gangue, indicating good potential for pre-concentration by both optical and conductivity methods which will be confirmed during further testwork. 6.4 Cordilleran Copper Porphyry Ores Porphyry deposits are the world's most important source of Cu and Mo; they account for about 60 to 70% of world Cu production and more than 95% of world Mo production. Porphyry deposits are also major sources of Au, Ag, and Sn; significant by-product metals include Re, W, In, Pt, Pd, and Se (Sinclair, 2007). Copper porphyrys are a major economic deposit type in the Canada and South American Cordilleran region as well as Australia. The deposits are predominantly mined by open pit methods, however many of these mines are mature and production at many mines is scheduled to move to the block caving method; Rio Tinto predicts a predominance of Cu production by this method by 2014 (Cross, 2006), thus this is envisaged as a significant application for the integrated mining, processing and waste disposal approach. There are currently no pre-concentration plants operating on copper porphyrys, although there is some suggestion in the literature that potential for the pre-concentration of copper porphyry ores exists (Burns & Grimes, 1986). Copper porphyry ores are typically 1-2% copper and 98% gangue minerals, therefore the likelihood of a coarse, barren fraction seems high. Furthermore, while the mesotexture of a typical copper porphyry is highly disseminated by definition, the 143 typical microtexture of the ores indicates preferential deposition of the sulphides along interstitial cracks and fractures within the ore zone (Figure 6.4). Figure 6.4 – Massive to disseminated chalcopyrite mineralization along fractures and foliations in biotite porphyry, QC, Ca (after Sinclair, 2007). Further supergene enrichment may also occur in these ores, maximizing the potential for variable grade distributions which have been identified as characteristic of good pre- concentration potential. Possibly the best cited example in terms of this potential is from AMT Copper in Arizona (McCullough et al, 1999): bench and pilot scale metallurgical testwork was conducted at Mountain states R&D on a range of ores from the Copper Creek property in Arizona, demonstrating that, in the case of relatively coarse copper porphyry mineralization, pre-concentration by DMS was effective in improving the grade of the copper ore at a high metal recovery. Mineralization was principally porphyritic chalcopyrite, with some massive chalcopyrite, and minor mineralization of bornite and chalcopyrite in sericitic dolerite, with a nominal liberation size of 13mm. Metallurgical testwork indicated a good potential for DMS and the flowsheet was piloted. Rejects from the DMS pre-concentration step were to be crushed and prepared as backfill for the proposed mechanized cut-and-fill mining method. Further pre-concentration potential for copper porphyry ores has been demonstrated in ores from Rio Tinto’s Bougainville Mine in Papua New Guinea. It was observed during early operation of the mine that mineralization in the Panguna zone of the mine occurred mainly on fracture planes in the ore, whereas other porphyritic mineralization was highly disseminated. Sampling and testwork on 3 major ore types at the mine, Panguna Andesite, Kawerong Diorite and the biotite/diorite/granodiorite zone was undertaken in 1984 in order to assess this potential 144 (Burns & Grimes, 1986). Photographs of Panguna Andesite samples and biotite/granodiorite samples are shown in Figure 6.5 (after Laznicka, 1973, photos courtesy AMIRA Data Metallogenica). A significant increase in the grade of the fines fraction was noted in both the Panguna and Kawerong ores (Figure 6.5). The lower grade biotite and granodiorite zones, comprising a small portion of the mineral reserve, were significantly lower grade and did not indicate any increase in metal content in the fines fraction (Figure 6.6). Figure 6.5 –Panguna Andesite (L) and Biotite Zone (R) Ore Textures from Bougainville Mine (after Laznicka 1973) 145 Figure 6.6 - Size / assay relationationship in Bougainville copper porphyry ores (after Burns & Grimes, 1986) Historical cutoff grade at the mine was 0.3% copper. However, a significant upgrading effect was achieved for lower grade ore zones through crushing to nominally -150mm and screening of the ROM ore at -31.5mm (Table 6.2). An increase in metal recovery in flotation of 1% over that of the ROM ore was observed during testwork on pre-concentrated ore samples. Table 6.2 – Bougainville Ore Pre-concentration Results Feed -31.5mm Fraction Sample Cu%* Wt% Cu% Cu Wt% DC1 0.19 34.46 0.33 59.50 BS 0.29 32.46 0.58 64.49 1A 0.29 36.99 0.51 64.70 1B 0.53 50.09 0.80 76.32 2 0.52 45.63 0.85 75.75 3 0.70 68.91 0.86 84.88 4 0.31 47.01 0.43 65.36 5 0.82 63.10 1.19 91.08 PS 0.37 37.30 0.69 68.70 PN 0.39 47.40 0.57 69.50 PE 0.10 29.10 0.16 47.40 KD 0.13 36.30 0.21 56.70 Average 0.39 44.06 0.60 68.70 146 Based on the strength of these results, a crushing and screening plant was designed and built at Bougainville by Minenco Ltd to treat 35 Mt of ROM ore per annum at the mine. The impact on the mine was to decrease the cutoff grade to 0% Cu, and then reject uneconomic material through crushing and screening of the ROM ore at 38mm, thus increasing the extraction of the mineral reserve by 58% through the ability to mine and process previously uneconomic material (Burns & Grimes, 1986). 6.5 Footwall and Contact Type Ores of INCO’s Sudbury Operations In 2004, a preliminary study was conducted at INCO’s McCreedy East Mine, in the Sudbury Basin, Ontario to determine the potential for underground pre-concentration at the mine (Bamber, 2004). McCreedy East is a mature, medium depth base metal mine with increasing haul distances, poor ground control and resultant high unit mining costs. Sudbury ores fall into three principal categories: contact ores, footwall ores and offset-dyke deposits (Peredery & Morrison, 1984). Two of the principal ore types, Footwall and Contact ores were represented at the mine and were thus sampled and tested. Contact ores are generally hosted in the ultramafic zone between the igneous complex and the transitional Sudbury breccias, and are typically rich in Ni, but poor in other metals. Contact ores are pentlandite-rich, thick and shallow dipping, with a high Ni:Cu ratio and containing complex pentlandite / pyrrhotite with chalcopyrite occurring as massive sulphides (Coats & Snajdir, 1984). Footwall ores are narrow-vein, massive sulphide stringers, situated in the Sudbury breccias down-dip of the Contact ore. Footwall stringers are rich in Cu, and total precious metals (TPMs), but poor in Ni. Stringers vary in width from 0–6m, dip variably between 20-60˚, and are mined typically using mechanized drift- and-fill methods. The veins are mostly chalcopyrite, grading about 30% copper, with secondary veins of pentlandite, millerite and occasionally bornite (Coats & Snajdir, 1984). The highly variable thickness of the veins leads to mining dilutions of up to 70%, and thus the underground pre-concentration of this ore in particular is expected to generate significant benefits. Offset dyke deposits have not been included in this evaluation. Both ores are mined by cut and fill methods, and high levels of dilution of the narrow vein copper ore in particular was a problem at the mine. DMS testwork on the ore indicated that 55% by mass of waste could be rejected from the Footwall type ore at 97% Cu recovery. Optical sorting on core crushed to -19mm indicated that on average 44% of the ore could be rejected at 97% Cu recovery. Rejection by DMS from the Contact ore was 22% at 95% Ni recovery. Conductivity sorting was also shown to be effective on this ore (Buksa & Paventi, 2002). Waste 147 rejects were coarse, competent breccia aggregates sized between 9mm and 75mm and were potentially identified excellent material for inclusion in a suitably designed backfill mix. An economic evaluation was undertaken based on these results, indicating a potential 20% operating cost saving at the mine with the production of direct shipping ore (>20% Cu + Ni) from the NV copper through pre-concentration. The study also indicated that the Footwall ore concentrate could meet the feed grade of a typical copper smelter, and if accepted in the smelter, a higher overall PGM recovery of 95% vs. presently 80% in the mill would be achieved. A system cost estimate was developed, and based on the estimated cost of CDN$30m, the increase in NPV was CDN$134m over the life of mine. 6.6 Polymetallic Base Metal Sulphide Ores of Xstrata Nickel’s Ontario Operations Based on the successful conclusion of previous work on the pre-concentration of Footwall- and Contact-type Sudbury deposits at INCO, Falconbridge Limited was approached in August 2004 to see if there was potential for the application of the concept at Falconbridge’s Sudbury operations on similar ores. Previous discussions with Falconbridge had indicated possible applications at the Craig and Fraser Mines, however, two further sites, Thayer Lindsley Mine and the Fraser Copper Zone were identified where the technology might be applicable. The Fraser Mine is presently near the end of its ore reserves in the Fraser Copper Zone, which is a massive-stringer sulphide orebody, and thus there is little potential to implement underground pre-concentration here. However, the ore is considered representative of other ore types on the North Range, particularly the Footwall ore to be mined at Nickel Rim, and thus was to be evaluated for amenability to pre-concentration in this light. At Thayer Lindsley, the shaft was originally an exploration shaft and is currently the bottleneck to increased production. T-L mines a combination of classic contact and footwall ores as well as a low-grade area of banded pentlandites in Zone 1. Currently the mining of low grade ore is a bottleneck to revenue, and the pre-concentration of this ore in particular was identified as a means to hoist more metal to surface at Thayer Lindsley. Increased production could increase the earnings contribution of TL to Falconbridge operations and thus generate cash flow for other projects such as Nickel Rim, Fraser Morgan and Onaping Depth. Site visits were conducted over November 2004, with scoping tests for the mines undertaken on selected core samples. Results of preliminary metallurgical testwork were positive and it was decided to expand the scope of the study to include the 9 major ore types of Falconbridge’s Ontario Operations. 148 The expanded Phase II study comprised site visits, ore sampling, metallurgical and geotechnical testwork, as well as the evaluation of the pre-concentration rejects as a source of material for backfill. Further laboratory work involved evaluation of the impact of systemic pre- concentration on mining, ore transport and centralized milling activities at the Strathcona Mill. 6.6.1 Preliminary Core Evaluations on 2 Mines Core samples from two preliminary Xstrata mines were selected for evaluation. The Fraser Mine comprises two distinct ore bodies, Fraser Ni and Fraser Cu. Fraser Ni is a pentlandite rich ‘contact’ orezone typical of the Sudbury Complex, while the Fraser Copper Zone is a ‘footwall’ type ore comprising typically narrow-vein chalcopyrite in Sudbury breccia (Naldrett, 1984). The orebody grades at 0.53% Ni and 5.77% Cu with additional PGMs. Mineralization is typically narrow-vein stringers of chalcopyrite with secondary pentlandite, bornite and occasionally millerite (Coats & Snajidr, 1984). Veins vary greatly in orientation, and vary typically in width between 0-6m, thus dilution is high in panels with narrower veins (Figure 6.7). There is a very sharp contact between the orezone and the wall rock, and ore typically breaks off at the contact after fragmentation. Figure 6.7 – Fraser Copper Massive Vein Sulphides Thayer Lindsley is a copper-nickel mine situated on the Southern Rim of the Sudbury Igneous Complex, adjacent to the Murray Pluton. Present production at the mine is approximately 600,000 tpa, which is planned to be increased to 635,000 tpa through efficiency improvements 400mm 149 (Private communication with Falconbridge, 2005). It is planned to further increase the tonnage to a maximum of 660,000 tpa in 2007 through improved hoisting efficiencies and increased hoisting times at the shaft. It is presently felt that increasing production beyond 660 000 tpa is not possible due to constraints in terms of drilling, mucking and hauling equipment, as well as hoisting constraints. The orebody is a low-grade contact-type orebody, comprising principally banded and disseminated pentlandite in an ultramafic complex (Coats & Snajidr, 1984). Mineral reserves are estimated at 5 400 000t at an overall grade of 1.2% Cu and 1.1% Ni. Life of mine at the present mining rate of 635 000 tpa is thus currently planned to 2009. Ore zones vary in width from 4 to 30m, and thus mining presently occurs through a combination of blasthole stoping (±50%) and cut and fill methods (±50). Planned dilution is typically 24%, with unplanned sloughing sometimes contributing an additional 7% dilution in the longhole stopes. A photograph of a typical T-L cut-and-fill panel is shown in Figure 6.8. Note the orientation of the orezone and the angle of the contact in the cut and fill panel. Figure 6.8 – TL Cut and Fill Panel Showing Banded Pentlandite Ore Contact The respective core samples were photographed, weighed, evaluated visually and crushed to - 19mm to provide feed for the size-assay and heavy liquid separation testwork. SGs for the ore and waste fractions were taken and washability testwork was undertaken to determine a final cut SG. -3.4mm fines were removed for inclusion with the concentrate. Initial results for the TL ore were poor, and the core was re-crushed to nominally -6.7mm to evaluate the effect of increased fragmentation of the core. -1.7mm fines were taken to concentrate. In the testwork, 44.27% by 1000mm 150 mass of the Fraser Footwall ore was be rejected at a metallurgical recovery of 97.5% Cu and 89% Ni. Ni recoveries were poor due to some dissemination of the pentlandite into the breccias, an aspect which requires further mineralogical and metallurgical evaluation. The grade of the final concentrate (sinks plus -3.4mm fines) was 20.28% Cu and 6.5% Ni which, with suitable preparation, indicates potential for feed of this ore directly to a suitable smelter. For the banded pentlandite ore form T-L, mass rejection was 14% at a metallurgical recovery of 99% Ni and 97% Cu. The –1.7mm fines represented 36% by weight and were assayed at 0.81% Cu and 0.78% Ni. The final concentrate (sinks plus -1.7mm fines) grade was 0.72% Cu and 0.72% Ni, an increase of 16.5%. It must be noted that the grade of the core sample, at 0.63% Cu and 0.61% Ni, is below the planned cutoff grade for the mine, and cannot be considered representative of the optimum waste rejection potential. Pre-concentration of the Fraser ore was not practically considered due to the limited life of mine. However, for the TL ores, the results indicate that several strategies for increasing the metal producing capacity at the shaft are made possible. Based on the results, the mining rate could be increased to 770,000 tpa, from which 107,440 tpa of waste could be rejected through underground pre-concentration, delivering 660,000 tpa of high-grade ore to surface via the existing hoist. The fill factor would be reduced from 0.55 to 0.41 by the additional fill generated through underground processing, which would result in further unit cost savings versus the present scenario. Using such a system, the metal carrying capacity of the shaft would be increased from 7920t Cu and 7260t Ni to 8993t Cu and 8357t Ni per annum respectively. At the ruling 3-month average metal prices of US$1.47/lb Cu and US$7.30/lb Ni, and a US:CDN exchange rate of 1.2:1 this equates to an increase in revenue from Thayer Lindsley of approximately CDN$25m/annum (2006 terms). Results for both mines are encouraging. However, due to the limited life of mine at these operations, it is not feasible to undertake a project for the pre-concentration of these ores. However, the results were applicable to a number of Xstrata’s present and future operations, and additional ore types were recommended to be tested. Based on the results, two flowsheets were suggested. For the high grade narrow vein copper ores, a flowsheet based on size classification and optical sorting was suggested (Figure 6.9). For the contact type ores, a process of size classification followed by dense media separation of the middlings fraction was suggested (Figure 6.10). Further metallurgical work was indicated as required in order to determine the exact flowsheet and metallurgical balance for each envisaged application for the process at Xstrata Nickel. 151 11 -38 + 9mm Concentrate -38 Waste 8 -9mm Concentrate 1000t Ore feed Bin 7 -38 +19mm 1 Ore from Haul Trucks Primary Crusher Feed prep Screen Fine Sorter Hoist 2 Loading Pocket Surface Ore truck to Strathcona 15 Spray water Rockfill bin 13 Waste 3 4 14 Final Concentrate Rockfill LHD Secondary Crusher 5 6 Figure 6.9 – Size Classification and Optical Sorting for Narrow Vein Footwall Copper Ores 152 Concentrate Conveyor 22 S pray w ater 6 -9mm high grade fines Ore feed bin 4 Scalping Grizzly Primary Crusher Feed prep Screen DMS Drum 7 Sinks 8 Floats Rockfill silo CM Sump DM Sump Make up 17 D&R Screen 19 D&R Screen9 18 20 DMS water makeup Pipe densifier Medium Circuit 5 ROM ore 1 2 23 Process Water 21 FeSi make-up 3 16 Feed prep Screen Hoist Ore truck to Strathcona 14 Final Concentrate Surface Figure 6.10 – Size Classification and Dense Media Separation for Contact-type Nickel Ores 153 6.6.2 Phase II Study for 9 Ores of Xstrata Nickel’s Ontario Operations Based on the successful conclusion of previous work at the Fraser and T-L Mines, a comprehensive study of the benefits of pre-concentration, and potential strategies for the disposal of the pre-concentration rejects was initiated at Xstrata Nickel’s Ontario operations in 2005. Xstrata Nickel presently has three principal producing mines located around the Sudbury Basin in Ontario Canada: the Fraser, Craig and Thayer Lindsley Mines. A further mine, the Montcalm Mine was acquired from Outokumpu in 2003 and is located some 100km East of Xstrata’s Kidd Creek metallurgical complex in Timmins, Ontario. Raglan mine in Quebec was not considered in the study. Ores from the Fraser and Craig Mines are considered analogous to the ores of two significant future operations in the Group: Craig LGBX for Onaping Depth and Fraser copper and Fraser nickel for Nickel Rim footwall and contact sones respectively. The study comprised sampling, metallurgical and geotechnical testwork, as well as backfill mix design and evaluation for 9 principal ore types from the four operations. Results from Fraser and Craig were used to extrapolate results for the two future mines described. A systems engineering approach was used to model the impacts to mining, material transport, and milling activities at these operations. The decrease in overall energy usage arising from the decrease in mass, and increase in metal grade of the ores was thus evaluated for the combined present and future operations using energy-based methods. The results have significance in terms of improvements in the efficiency of existing operations as well as the operational management of future mines. A map showing the spatial distribution of Xstrata’s Ontario Operations is shown in Figure 6.11. Stope samples weighing about 300 kg each, with a topsize of approximately 300mm were collected from nine different ore zones at all four of the mines. Several of these ore types, including massive vein, massive- and banded sulphides have been examined in previous studies. However, the Craig LGBX and Montcalm type ores were new ore types to the programme and had not previously been tested. LGBX type ores are Lake-Granite Breccia ores with a matrix- type mesotexture (Figure 6.12). Montcalm ores are highly disseminated to matrix sulphides from the Timmins igneous belt East of Timmins (Figure 6.13). 154 Figure 6.11 – Xstrata Nickel Ontario Operations (operations shown in red) Figure 6.12– Sudbury breccia /matrix type ore texture 30mm 155 Figure 6.13 – Disseminated Ore Texture (from Montcalm, Timmins) 184.108.40.206 Metallurgical Results A significant density differential was observed between the mineralized sulphides and the siliceous gangue material in all of the ores tested. Liberation of the gangue material was also observed to be generally high. DMS testing was conducted on the samples at a separation S.G. of between 2.7 and 2.95 at SG intervals of 0.5. Final cut SG was selected depending on the results. -6.7mm fines were generally considered high-grade and were recombined with the DMS concentrate. The combined test products were dried, weighed and assayed to produce a metallurgical balance. The results of the DMS test program are presented in Table 6.3 (after Weatherwax, 2007). All nine ores exhibited high metal recoveries accompanied by significant mass rejection. The massive-vein Fraser Copper ore yielded the best results in the DMS study, with nickel and copper recoveries in excess of 96% and mass rejection in excess of 53%. Thayer Lindsay Footwall ore is of the massive-vein type and yielded excellent results, with nickel and copper recoveries greater than 97% and a mass rejection of 37%. Footwall ores in general present an excellent opportunity for pre-concentration as the pre-concentrate grades obtained are of the order of a typical smelter feed grade, which presents particular advantages in terms of the transport and processing of these ores. Metallurgical results for other ores are also generally acceptable; however, the Craig LGBX ore was more refractory and showed a significantly lower 20mm 156 copper recovery than nickel, indicating that the copper and nickel are not associated mineralogically. Further metallurgical work is required in this area. A preliminary conductivity test on ore and waste samples for each ore was conducted using the MineSense ‘B2’ MkII conductivity sensor. A clear discrepancy between the conductivity response of the selected ore and waste fractions was determined (see Appendix G). A sample of each ore type was then subjected to sorting using the INCO conductivity sorting test rig. Samples were typically crushed to -75mm, and the arising -10mm fines were removed. The -75 mm + 10mm fraction was sorted, and the sorter concentrate re-combined with the fines for a final concentrate. The results of the preliminary sorting tests are presented in Table 6.4. Waste rejections of between 19 – 64% were achieved, with metal recoveries varying between 50 – 96%. Montcalm Low ore showed excellent waste rejection, but poor metal recoveries. However, ores with moderate nickel contents such as Montcalm High, Craig LGBX, Craig 8112 and Fraser Ni all showed fair responses to the technology and further work is recommended on these ores. The TL670 and Fraser Cu showed the poorest results which was expected from the initial sensor responses of these ores. Optical Sorting Selected gangue and mineralized particles were selected for optical evaluation for sorting purposes. Samples were analyzed using the UBC NI Machine Vision Station. RGB data, as well as textural analyses were undertaken on the samples, with significant discrepancies between the valuable and non-valuable fractions in each case, particularly in terms of texture. Detailed results for the evaluation of Xstrata ores are presented in Appendix H. Results indicate a preliminary potential for the optical sorting of these ores, although preliminary sorting tests have not yet been completed. 157 Table 6.3. Summary of DMS testwork results on 9 Xstrata Nickel ores (after Weatherwax, 2007) Deposit Mineralogy Product Cut Grade (%) Distribution (%) SG Wt% Ni Cu Mg Ni Cu Mg Craig Massive/disseminated sulphide DMS Concentrate +2.95 85.19 1.22 0.57 5.80 97.34 97.34 86.12 8112 Final Concentrate 86.19 1.26 0.57 5.71 97.63 97.63 86.90 Craig Breccia/matrix sulphide DMS Concentrate +2.95 65.60 3.53 0.37 2.42 96.90 79.75 66.00 LGBX Final Concentrate 67.88 3.52 0.38 2.37 97.18 81.55 67.83 Fraser Massive/ disseminated sulphide DMS Concentrate +2.8 66.33 0.82 0.40 3.34 80.04 80.04 62.97 Ni Final Concentrate 75.45 0.82 0.48 4.00 91.35 91.50 42.70 Fraser Massive vein sulphide DMS Concentrate +2.9 35.79 0.70 23.62 0.54 92.74 97.03 9.60 Cu Final Concentrate 46.63 0.84 22.01 0.69 96.01 97.95 17.49 TL Massive vein sulphide DMS Concentrate +2.9 60.70 1.90 11.11 0.95 97.47 97.70 38.94 Footwall Final Concentrate 63.40 1.83 10.79 1.03 97.65 97.88 43.68 TL Massive sulphide DMS Concentrate +2.9 71.61 1.71 1.08 3.64 97.42 94.93 68.62 Zone 2 Final Concentrate 74.27 1.70 1.11 3.58 97.73 95.65 71.14 TL Disseminated/ banded sulphide DMS Concentrate +2.9 78.21 0.71 0.41 6.45 94.02 90.91 78.60 670 Final Concentrate 80.48 0.82 0.45 6.20 95.40 92.60 80.21 Montcalm Disseminated sulphide DMS Concentrate +2.95 72.67 2.14 0.84 4.38 97.36 92.62 65.25 East Final Concentrate 74.50 2.12 0.82 4.39 97.56 93.11 67.417 Montcalm Disseminated sulphide DMS Concentrate +2.8 57.78 0.56 0.24 6.09 84.64 82.49 70.97 West Final Concentrate 67.63 0.55 0.24 6.04 97.57 95.25 82.36 158 Table 6.4 – Summary of Conductivity Sorting Testwork on Xstrata Nickel Ores (after Weatherwax, 2007) Product Wt% Assay (% or g/t for PGMs)) Distribution (Wt%) Co Cu Mg Ni Au Ag Pd Pt Co Cu Mg Ni Au Ag Pd Pt Craig 8112 Reject 25.7 0.01 0.22 6.54 0.27 0.03 0.64 0.07 0.06 8.4 11.5 30.6 5.8 16.5 11.1 15.3 15.7 Final Conc 74.3 0.04 0.58 5.12 1.52 0.06 1.76 0.13 0.12 91.6 88.5 69.4 94.2 83.5 88.9 84.7 84.3 Total 100.0 0.04 0.48 5.48 1.20 0.05 1.47 0.11 0.10 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Craig LGBX Reject 15.8 0.02 0.28 3.47 0.51 0.03 0.61 0.05 0.04 4.6 12.2 21.7 3.7 20.9 10.6 7.1 7.0 Final conc 84.2 0.06 0.37 2.35 2.51 0.02 0.97 0.13 0.11 95.4 87.8 78.3 96.3 79.1 89.4 92.9 93.0 Total 100.0 0.06 0.36 2.53 2.19 0.02 0.91 0.11 0.10 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Fraser Ni Reject 19 0.18 6.08 0.29 0.03 0.66 0.02 0.04 0.19 9.9 9.4 27.1 6.6 17.5 9.4 5.7 6.8 Final conc 81 0.41 3.73 0.93 0.04 1.45 0.09 0.14 0.81 90.1 90.6 72.0 93.4 82.5 90.6 94.3 93.2 Total 100 0.37 4.17 0.81 0.04 1.30 0.08 0.12 1.00 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Fraser Cu Reject 50.2 0.01 4.85 2.59 0.27 0.08 23.22 0.70 0.77 28.3 19.4 77.5 15.3 26.0 22.5 19.1 17.0 Final Conc 49.8 0.01 20.39 0.76 1.48 0.24 80.94 2.98 3.78 71.7 80.6 22.5 84.7 74.0 77.5 80.9 83.0 Total 100.0 0.01 12.58 1.68 0.87 0.16 51.94 1.83 2.27 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 TL-15-2 Reject 32.0 0.01 3.25 3.51 0.20 0.38 6.93 0.41 0.32 7.2 11.5 59.5 5.1 32.2 11.3 5.8 10.1 Final conc 68.0 0.07 11.74 1.12 1.80 0.38 25.73 3.14 1.33 92.8 88.5 40.5 94.9 67.8 88.7 94.2 89.9 Total 100.0 0.05 9.03 1.89 1.29 0.38 19.72 2.27 1.00 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 TL670 Reject 50.9 0.02 0.39 6.33 0.44 0.17 1.22 0.09 0.12 31.7 43.9 55.1 29.6 59.1 43.0 33.5 50.8 Final Conc 49.1 0.04 0.52 5.36 1.09 0.12 1.68 0.18 0.12 68.3 56.1 44.9 70.4 40.9 57.0 66.5 49.2 Total 100.0 0.03 0.46 5.86 0.76 0.15 1.45 0.13 0.12 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 TL80 Reject 34.5 0.02 0.28 3.91 0.36 0.03 1.08 0.12 0.21 11.9% 13.7 37.8 8.7 5.6 10.8 12.6 13.3 Final Conc 65.5 0.06 0.93 3.38 1.99 0.26 4.69 0.43 0.72 88.1% 86.3 62.2 91.3 94.4 89.2 87.4 86.7 Total 100.0 0.05 0.71 3.56 1.43 0.18 3.45 0.32 0.55 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 MH Reject 23.0 0.02 0.33 5.94 0.43 0.05 1.13 n/a n/a 7.9 13.4 29.7 5.9 11.1 15.0 n/a n/a Final Conc 77.0 0.07 0.63 4.20 2.04 0.11 1.91 n/a n/a 92.1 86.6 70.3 94.1 88.9 85.0 n/a n/a Total 100.0 0.05 0.56 4.60 1.67 0.10 1.73 n/a n/a 100.0 100.0 100.0 100.0 100.0 100.0 n/a n/a ML Reject 64.24 0.01 0.09 5.94 0.19 0.01 0.37 n/a n/a 43.2 37.4 64.1 35.4 28.3 34.3 n/a n/a Final Conc 35.76 0.02 0.28 5.97 0.61 0.06 1.26 n/a n/a 56.8 62.6 35.9 64.6 71.7 65.7 n/a n/a Total 100.00 0.01 0.16 5.95 0.34 0.03 0.69 n/a n/a 100.0 100.0 100.0 100.0 100.0 100.0 n/a n/a 159 220.127.116.11 Investigation of Waste Disposal Aspects As an extension of the study into pre-concentration at Xstrata’s Ontario operations, it was decided to evaluate the capabilities of the various types of pre-concentration rejects as a material for geotechnical support in the underground context. Of the mines investigated, several mining methods were employed, ranging from drift and fill at Fraser Cu to open stoping methods at Montcalm. What is of benefit to the consideration of the concept is the fact while only the mines using drift and fill or cut and fill methods depend on fill for their success, it was found that the mining methods at all the operations were able to accommodate fill in the mining cycle, and thus the potential to dispose of the rejects underground in each case has been demonstrated. For the present and future mines that require fill for the success of the operation, it is expected that there would be additional benefits in terms of utilizing the material rejected through the pre-concentration of ore to create high strength backfills. A preliminary study into the impact of using pre-concentration rejects was conducted in 2005 and 2006 (Bamber et al, 2006). It was decided to use the rejects from some of the DMS testwork on a Ni-Cu sulphide or from the Sudbury basin as a starting point. Rejects from the concentration process were typically coarse, low-sulphide and considered suitable for use in an aggregate fill. As it is desired to maximize the degree of waste rejected from the ROM ore and disposed of as fill, classified tailings from the surface mill were used as a source of the fines component for the mix in order to indicate the overall potential for solids disposal underground. A typical composite fill would thus comprise up to 90% ROM material, consisting of the DMS rejects, classified tailings, plus cement and water for maximum disposal of the solid waste arising from the ROM ore underground. Results were positive and an extended programme was initiated for Xstrata Nickel’s Ontario division mines. Ores were typically pre-concentrated using dense media separation, although optical and conductivity methods were also considered, and the results presented in this Chapter are expected to be consistent for all pre-concentration methods considered in the thesis. A range of waste products were produced during the pre-concentration testing on the Xstrata ores, these products were physically and geochemically characterized in terms of the ASTM standards for use as aggregates; fill mixes were designed based on these results, and tested for uniaxial compressive strength (UCS) as well as tested to determine the rheology of the fresh fill mix for pumping purposes. 160 Results Approximately 50kg of the reject material was typically available from each sample (Figure 6.14). Figure 6.14 – Typical Dense Medium Separation Reject fraction (Fraser Ni) The size distribution of the waste rejects was typically -75+19mm due to the rejection of the - 19mm fines to concentrate in the process. -75 +19mm is generally considered too coarse for suitable aggregate so it was decided to produce two mix designs for testing, firstly with the rejects as-is, and a second mix with the rejects crushed to nominally 100% - 50mm in order to improve the rheology of the mix in the light of the desire to pump the fill. Results of the geotechnical characterization are shown in Table 6.5 (Weatherwax et al, 2008). 161 Table 6.5: Measured Geotechnical Properties of Pre-concentration Rejects Ore Body Hardness P80 P60 P10 Cu % Flat and Elongated Adsorption ABA SG Void Ratio moh's mm mm mm +9.5mm % Dry Wt. Craig LGBX 5-7 15 7 1.5 4.7 14.3 0.58 0.54 2.77 0.41 Craig 8112 >7 8.5 5.4 0.8 6.8 34.0 0.39 0.40 2.77 0.40 TL Zone 1 >7 6.5 5 1.3 3.8 82.8 0.82 0.47 2.98 0.42 TL Zone 2 >7 22 15 2.5 6.0 21.7 0.33 1.01 2.78 0.43 TL Footwall >7 23 17 2.5 6.8 25.4 0.29 0.63 2.86 0.39 Fraser Cu 5-7 22 17 3 5.7 24.0 0.29 0.85 2.86 0.51 Fraser Ni 5-7 20 15 2.5 6.0 27.3 0.19 0.27 2.94 0.44 Mont. East >7 22 17 3.5 4.9 18.3 0.24 0.99 2.95 0.51 Mont. West 5-7 22 18 5 3.6 10.3 0.66 1.19 2.93 0.49 Rejects arising from the pre-concentration of the Xstrata ores were found to be generally competent, with a high coefficient of uniformity, denser as well as generally more flat and Figure 6.15: Size Distribution of Pre-concentration Rejects elongated than standard recommended ASTM aggregates. The size distribution of the rejects arising from the pre-concentration testwork was analysed and compared to the recommended aggregate size distribution for fill aggregates (Figure 6.15). Size distribution of the rejects is ultimately determined by their feed preparation for concentration and thus this was not considered a variable. As noted in preliminary work for INCO at McCreedy East (Bamber, 2004), the ores demonstrated a characteristic high grade fines fraction, and removal of this fraction prior to concentration increases the relative size distribution of the rejects when compared to the ROM ore, increasing their compatibility in terms of aggregate uses. The 0 10 20 30 40 50 60 70 80 90 100 1.E-01 1.E+00 1.E+01 1.E+02 Size (mm) % P as si ng Craig LGBX Craig 8112 TL Zone 1 TL Zone 2 TL Footwall Fraser Cu Fraser Ni Mont. East Mont. West 162 rejects were measured generally coarser than a typical ASTM fill aggregate of similar topsize, but are nevertheless considered acceptable for use in the fill. Void ratios of the aggregates were also calculated from the size distribution data, indicating a high overall bulk density of the rejects. Results confirm the potential shown in preliminary mix design and testing (Bamber et al, 2006), and it can be concluded that the rejects of the pre-concentration process can typically be considered acceptable for use as aggregate in fill. Four standardized mixes were designed and tested for further evaluation of the concept (Table 6.6). Classified tailings obtained from Strathcona mill was typically 80% - 149um (Verdiel, 2006) and was used as the fine aggregate for the preparation of the paste fraction of the mix. Binder was 5% by mass Lafarge CSA A5 Type 10 NPC. Mixes were prepared and tested according to ASTM standards for backfills using an M-Test 841 testing machine. Table 6.6 –Mix Ratios for Xstrata Fill Testwork – Average Mix (dry weight basis) Sample Rejects Wt% Tailings Wt% Cement Wt% Water: Cement RockFill 95.0 5.0 1:1 CT Max 69.1 28.8 1.4 6.5:1 CT 1:3 36.5 59.3 3.0 6:1 CT 1:7 19.9 74.7 3.7 5.6:1 Full Max 69.7 29.1 1.5 6.5:1 Full 1:3 37.0 58.7 2.9 5.4:1 Full 1:7 20.1 74.5 3.7 5.2:1 Results for Rockfills Results of the UCS testing showed that the rejects could be used as aggregate to produce a cemented rockfill with a high fill strength / unit cement and high stiffness. 28-day strength of the samples was measured between 1.28 - 3.41 MPa with an average of 1.89MPa. Stress strain curves and observations during the test indicate that cement content is the primary mechanism in the strength of the fill, and that strength falls rapidly in these fills post-failure. Failure is generally stiff, yet plastic, with low residual strength. Figure 6.16 shows the UCS results for these fills. Some evidence of binder segregation by gravity was observed in the samples (Weatherwax, 2007). Failure in the reject particles was observed in only Thayer Lindsley Zone 2, where an individual reject particle showed signs of failure. 163 0 0.5 1 1.5 2 2.5 3 3.5 0 0.01 0.02 0.03 0.04 0.05 0.06 Axial Strain (mm/mm) U C S (M Pa ) CR 8112 CR LGBX F Cu F Ni MH ML TL Zone 1 TL Zone 2 TL Footwall Figure 6.16: UCS vs Axial Strain for Rockfills Results for Composite Fills A range of composite fills comprising various ratios of pre-concentration rejects, mill tailings and cement were designed, prepared and tested. Results are summarized in table 5. Strength generally increased with increasing reject content, with the ‘Maximum density’ samples showing the highest average strengths for the composite fills. The efficiency of the mixes was also analysed, and results showed that the efficiency of the fills in terms of UCS : Wt% cement increased significantly with increasing reject content, thus indicating the positive benefits of utilizing these rejects as fill material. Absolute compressive strengths achieved were generally low, albeit at a low cement content; ultimate cement content in the mixes as well as water cement ratio was highly variable and thus results varied substantially across the mixes (Figure 6.17). This can be attributed primarily to excessive water in several of the mixes which should be addressed in subsequent phases of testing (Talbot & Richart, 1923). 164 0 0.5 1 1.5 2 2.5 0 0.01 0.02 0.03 0.04 0.05 0.06 Axial Strain (mm) U C S (M Pa ) F Cu Rockfill F Cu Full Max F Cu Full 1:3 F Cu Full 1:7 Full Tailings Figure 6.17: UCS vs Axial Strain for Composite Fills Overall Results From the results it can be concluded that the addition of rejects has an increasing effect on the overall performance of the mixes in these samples, until maximum density is reached and point-to-point contact is made among the aggregates (Weatherwax, 2007). Absolute strength is a maximum for the rockfills. However, the maximum binder efficiency was achieved with the maximum density composite fills, with a resultant UCS:% cement ratio of 0.73 – 0.86, indicating that a fill strength of between 4 – 5 MPa would be possible at a binder content of 5% (Table 6.7). The confined strength of the fill, and strength post failure is of interest in high stress, plastic conditions in underground stopes, and the composite fills indicate good potential in this regard based on the high residual strength demonstrated in the testing. Table 6.7 – Overall UCS Results for Xstrata Fill Mixes Sample Rejects% Tailings% Cement% W:C UCS UCS: %Cement RockFill 95.0 5.0 1.0 1.89 0.38 CT Max 69.1 28.8 1.4 6.5 1.24 0.86 CT 1:3 36.5 59.3 3.0 5.9 0.8 0.27 CT 1:7 19.9 74.7 3.7 5.6 0.77 0.21 Full Max 69.7 29.1 1.5 6.5 1.07 0.74 Full 1:3 37.0 58.7 2.9 5.4 0.82 0.28 Full 1:7 20.1 74.5 3.7 5.2 0.7 0.19 165 18.104.22.168 Rheological Evaluation of Fill Mixes It is the intention to use the rejects as a material for composite fills in underground backfill applications, thus the transport of the fill becomes an issue and it was felt important to investigate the basic rheological properties of the fills for the purpose of determining an appropriate rheology for transportation by pumping. For the evaluation of this, the modified ‘slump test’ method was used to determine a dimensionless yield stress (τ') which allowed for a comparison of the different mixes (Hu et al, 1995). The results show that an increasing the amount of rejects increases the value of τ’. Results of the testwork are presented in Table 6.8 (Weatherwax et al, 2008). Table 6.8 – Average of Slump Test Results for Xstrata Fill Mixes Mix Rejects Tailings Cement Water:Cement τ (von Mises) RockFill 95 0 5 1.0 0.58 CT Max 69.13 28.77 1.4385 6.5 0.58 CT 1:3 36.52 59.28 2.964 5.9 0.15 CT 1:7 19.87 74.66 3.733 5.6 0.09 Full Max 69.74 29.05 1.4525 6.5 0.58 Full 1:3 37.04 58.66 2.933 5.4 0.13 Full 1:7 20.08 74.46 3.723 5.2 0.05 The shear stress in the mixes varies from 0.05 for the low density composite fills to 0.58 for the maximum density and rockfill mixes. For a pumpable fill mix, a shear stress of between 0.15 – 0.5 is suggested as reasonable, and further optimization to maximize strength while maintaining pumpability of the mixes is required. At this stage of testing, slump rate was not measured and a further phase of testing on fresh mixes is recommended to explore aspects of plastic viscosity as well as shear stress in order to fully characterize the fill mixes. 22.214.171.124 Grinding Index Testwork The majority of ores are presently delivered to the centralized Strathcona Mill near Levack village in the North East of the Basin for milling. Montcalm ores by exception are delivered to the Kidd Creek Metallurgical Complex in Timmins. The pre-concentration of ores is expected to result in power savings at the mills in three principal areas. Firstly, the rejection of waste results in a reduction in tonnage reporting to the mill. Secondly, the waste is primarily hard siliceous rock; removing this material leaves a product enriched in relatively soft, metal bearing sulphides which will have a lower grinding work index. Thirdly, the pre-concentration product will be crushed to a finer size than the present plant feed. This projected impact on 166 grinding power requirements was calculated from measurements of the Work Index using the Bond method. Feed, concentrate and reject samples from the 9 ores in the test programme were subjected to a series of tests in order to evaluate this impact. For the evaluation, the work index of the Fraser Nickel ore was determined through a complete Bond Work Index test and used as a reference. The Work Indices of the other ores were then determined through comparative work index tests on whole ore and the pre-concentration products. Results of the Work Index tests are presented in Table 6.9 (Altun, 2007). In the evaluation, it was also considered to evaluate the influence of pre-concentrate grade on the Work Index of the pre- concentrates. For the Xstrata ores, the lowest work indices were obtained for the highest grade Cu ores, Fraser Cu and TL 15. Also, the largest decrease in the work index occurred with the Fraser Cu ore, which had the highest waste rejection, and thus greatest increase in grade after pre-concentration. Overall Work Index of the ores was reduced by 8.8% through the rejection of between 30-54% waste from the ores. Table 6.9 - P80, F80 and Work Indices for the raw ores and their concentrates Sample F80 (microns) P80 (microns) Work Index (kWh/tonne) % Reduction in WI Fraser Ni Feed 2930.26 86.93 10.63 * 10.17Concentrate 2849.25 76.00 9.83 Fraser Cu Feed 2504.50 53.98 8.13 19.10Concentrate 2685.67 37.72 6.58 Montcalm West (ML) Feed 2784.50 87.48 10.73 6.99Concentrate 2739.12 77.34 9.98 Montcalm East (MH) Feed 2791.72 73.76 9.68 11.78Concentrate 2787.28 59.68 8.54 TL 15 Feed 2693.67 56.43 8.29 9.65Concentrate 2649.76 47.23 7.49 TL 80 Feed 2777.29 66.78 9.13 12.16Concentrate 2773.00 53.49 8.02 TL 670 Feed 2778.21 84.95 10.54 2.84Concentrate 2771.57 80.90 10.24 Craig 8112 Feed 2801.86 74.28 9.72 -Concentrate - - - Craig LGBX Feed 2791.57 70.02 9.38 -Concentrate - - - A direct relation between the grades and work indexes cannot be generally expected, as work index is affected by a number of additional mineralogical and petrographic characteristics of 167 the host rock and the valuable mineral associations. However, for this group of Xstrata ores, it was demonstrated that ore grade and grindability were somewhat related (Figure 6.18). The work index vs Cu grade correlation and in particular the relationship between Work Index and total Cu+Ni grades, thus total sulphides, shows an extremely high correlation (R2 = 0.85) and thus a relationship between the raw metal content and the Work Index of these ores was demonstrated. R2 = 0.7813 R2 = 0.8526 0 5 10 15 20 25 30 0 2 4 6 8 10 12 Work Index (kWh/tonne) G ra de (% ) Ni Cu Ni+Cu Power (Cu) Power (Ni+Cu) Figure 6.18 - Work Index-Grade Correlation 6.6.3 Evaluation of Impacts on Energy Usage at Xstrata’s Ontario Operations All of Xstrata’s ores in the Sudbury region are transported by truck to the Strathcona Mill located at the North West edge of the Basin. Nickel concentrates from Strathcona are transported back to the Ni smelter near Falconbridge Town; copper concentrates are shipped 300km north to Xstrata Copper’s Kidd Creek metallurgical complex in Timmins. Ore from Montcalm is hauled 100km on surface to the metallurgical complex at Kidd Creek. Montcalm and Thayer Lindsley are shallow mines, while Craig and Fraser are medium depth. However, these mines are mature and have limited mine life, and the future operations, Onaping Depth and Nickel Rim South, are significantly deeper, and more distant from the existing metallurgical complex. Ore from Nickel Rim is planned to be trucked 75km through the town of Sudbury for processing at Strathcona. Furthermore, continued processing of these ores at Strathcona places enormous pressure on existing tailings disposal facilities at this location. 168 Any savings in terms of hoisting, ore transport and tailings disposal requirements for these future ores will be significant. Xstrata is presently evaluating various options in terms of the strategic processing of these ores as well as additional synergies arising from the fact that CVRD-INCO mines and processes similar ores at different locations (Romaniuk, 2006). Significant benefits arising from these synergies have been identified, which are expected to be further enhanced through the selected application of an appropriate pre-concentration strategy at each organization. Figure 6.17 shows the spatial relationship of the operations from both Xstrata Nickel as well as CVRD INCO who also operates significant mines in the Basin. Operational parameters for each mine were investigated, discussed and agreed with Xstrata personnel prior to the evaluation (Proudfoot, 2007). Pre-concentration is principally anticipated to impact material handling activities such as haulage and hoisting, with additional impacts expected in the mill through the reduction in tonnage, particle size distribution and increase in metal grade of the pre-concentrates. Material transport energy utilization relates primarily to production rate, and deposit depth for hoisting to surface, and distance from the Strathcona Mill for trucking. At existing operations, deposit depths range from 150 m at Montcalm to 1400 m at Craig. Future operations Onaping Depth and Nickel Rim are deeper at 2500 and 1500m respectively. While the present Fraser and Craig Mines are close to the mill, the haulage distance for the other mines are significant. Montcalm transports ore 100 km to the Kidd Creek Mill while trucking distances for the Thayer Lindsley Mine and Nickel Rim Mine are 54.4 km and 75 km, respectively. Evaluation model parameters are summarized in Figure 6.19. 169 1300m 10 00 m 10 00 m H yd ra ul ic fi ll H yd ra ul ic fi ll 2500m Slag fill Slag fill Figure 6.19 – Xstrata System Evaluation parameters An evaluation model was developed to estimate energy usage and savings for each mine operating with- and without pre-concentration (Pitt & Wadsworth, 1980). The key areas of energy usage relate to hoisting, transportation by road to the mill, and beneficiation of the ore. The additional energy requirement of the pre-concentration plant is also included in the model. An energy cost of $0.057/kWh plus the expected maximum demand charge of $5000/MW was used to calculate the power cost and savings at each stage (Delphi Group, 2004). Additional credits due to potential overall savings in greenhouse gas emissions (GHG) of $15/t CO2 and 1720 kWh per tonne CO2 equivalent and are also accounted for at the end of the evaluation1. Energy usage for processing derives primarily from the feed size distribution, the grinding work index of the ore and the tonnage processed. The pre-concentration test results indicate the amount of waste that can be rejected and therefore s not processed at the mill. Further energy benefits result from the lower grinding work index of the relatively soft sulphide rich pre-concentration product as compared to 1 http://www.nrcan.gc.ca/es/etb/ctfca/PDFs/electrical-markets/en/3-1.html 170 the whole ore, and the reduction in feed particle size to the mill. In the case of the footwall ores, additional potential to bypass the mill and deliver a smelter feed directly is also considered. 126.96.36.199 Impact of the Pre-concentration Step Table 6.10 summarizes the projected energy usage and power costs for the pre-concentration plants. The energy value was obtained from the estimated power requirements for crushing, screening and dense media separation of the tonnage indicated based on flowsheets and equipment lists developed in previous metallurgical testwork (Bamber et al, 2005). Table 6.10. Estimated Annual Energy Usage and Costs for the Pre-concentration Plant Operation Montcalm Thayer Lindsley Fraser Copper Fraser Nickel Craig Onaping Depth Ni Rim S Contact Ni Rim S F/W Mining rate tpa 800000 400000 250000 500000 750000 750000 650000 650000 Power (kW) 846 532 388 618 810 810 736 736 Energy kWh/a 2354701 1479945 1080153 1718602 2255052 2255052 2048884 2048884 188.8.131.52 Impacts on Hoisting Energy The estimated energy usage for hoisting was obtained from information on the hoisting depth, cycle time, production rate, number of trips per hour, plus the calculated payload-, skip- and rope mass. Input parameters were a hoisting acceleration of 2 m/s2, 15 m/s max hoist velocity, a wait time of 20 s and hoist efficiency 90% (Figure 6.20). Skip payloads and skip mass are derived from Edwards (1990). Savings are enjoyed in terms of reduced payloads, reduced skip and hoisting rope masses and savings in overall rotational inertia in the hoisting system. Table 6.11 presents the estimated savings arising from pre-concentration in hoisting energy requirements and costs. Energy savings vary between $84 200/annum for T-L (shallow and low tonnage), to $640 000/annum for Nickel Rim (deep and high tonnage). 171 H o is t P a y lo a d S k ip m a s s R op e m as s D ep th a v Ac ce l C on st an t v el oc ity D ec el Figure 6.20 – Hoisting Impact Evaluation Model Table 6.11 Summary of Annual Hoisting Energy and Cost Savings. Operation Thayer Lindsley Fraser Copper Fraser Nickel Craig Onaping Depth Ni Rim S Ni Rim S F/W Depth 900 1400 1000 1000 2000 1100 1300 Energy (kWh/a) 6514 16552 8769 12469 42315 13025 38574 Cost saving ($)/annum $84,209 $277,816 $106,316 $144,077 $503,526 $154,400 $641,879 184.108.40.206 Impacts on Surface Haulage Of the current operating mines, only Montcalm and Thayer Lindsley have significant haul distances to the mill of 100 km and 54 km, respectively. Fraser and Craig Mines are proximal to Strathcona mill and thus haul savings are considered minimal. However, ore from the future Nickel Rim operation will be hauled some 75 km from the mine on the east side of the Basin, through the town of Sudbury to Strathcona mill in the northeast. Energy savings were calculated for transport of the pre-concentrated ore (based on the degree of waste rejection as shown in Figure 6.8) versus the transport of the whole ore to the mill (Maxim, 2001; Kodjak, 172 2004). Additional energy savings are enjoyed at Thayer Lindsley due to a reduction in the transportation of fill from Falconbridge to the mine. For the evaluation, a calorific value of 50337 kJ/l and cost of $1.10/l for the fuel was used. Figure 6.21 shows the energy impact in each case and the total savings resulting simply from transporting less ore. Cost savings vary from about $300,000 per annum for the Thayer-Lindsley Mine to almost $900,000 per annum for the Nickel Rim S Footwall. 0 100000 200000 300000 400000 500000 600000 700000 800000 900000 1000000 Montcalm Thayer Lindsley Ni Rim S Ni Rim S F/W $/ an nu m 0 5000 10000 15000 20000 25000 30000 35000 40000 (T ho us an ds ) kW h/ an nu m Total saving $/a Haul energy base kWh/a Fill haul energy kWh/a Haul energy precon kWh/a Figure 6.21 - Haul Energy Cost Savings from pre-concentration of the ore 220.127.116.11 Impacts on Grinding and Overall Beneficiation of the Ore For both the Strathcona and Kidd Creek mills, pre-concentration results in power savings in three principal areas. Firstly, the rejection of waste results in a reduction in tonnage reporting to the mill. Secondly, the waste is primarily hard siliceous rock; removing this material leaves a product enriched in relatively soft metal bearing sulphides which will have a lower grinding work index. Thirdly, the pre-concentration product will be crushed to a finer size than typical plant feed. The testwork indicates that the ore feed size is reduced from typically 300mm topsize to nominally -75mm, and reductions in Bond Work Index through the rejection of between 20 and 54% coarse, siliceous waste can vary from 6 -13% (Altun, 2007). Results are presented in Figure 6.22. 173 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Montcalm Thayer Lindsley Fraser Copper Fraser Nickel Craig Onaping Depth Ni Rim S Ni Rim S F/W (T ho us an ds ) kW h/ a 0 10 20 30 40 50 60 tp h Grinding energy kWh/a Grinding energy pre- con kWh/a Waste rejection Figure 6.22 - Grinding energy requirements with- and without pre-concentration Impacts on the overall energy requirements for the beneficiation of the ore were also analyzed. Savings in crushing and screening at the existing mill were accounted for at 50% of present requirements, and grinding energy savings are as calculated above. These savings were offset against the additional energy required for the pre-concentration step shown in section 2.1. The overall estimated impact of pre-concentration on.. milling energy requirements for these ores is shown in Figure 6.23. A significant further opportunity to use the pre-concentration step to produce ore feeds that can be customized for processing at a particular mill has been identified. The results indicate that it would be possible to produce low nickel, low-copper, high copper and high nickel feeds from various mines, as well as barren waste streams with potential for use as fill. Custom feeds thus produced could be directed to either the Clarabelle or Strathcona Mill in a possible integration scenario between Xstrata and CVRD-INCO. Pre-concentrates from Fraser Copper, the T-L Footwall ore as well as Nickel Rim South are indicated as potentially meeting a smelter feed grade (Bamber et al, 2006; Weatherwax, 2007), thus creating the opportunity to deliver ore directly to either the Falconbridge smelter, Kidd Creek or potentially INCO’s smelter at Copper Cliff. Additional research is required to evaluate this opportunity; however, major positive benefits would accrue particularly at T-L and Ni Rim should some potential for synergies with INCO be established. 174 0.00 500.00 1000.00 1500.00 2000.00 2500.00 P ow er (k W ) Montcalm Thayer Lindsley Fraser Copper Fraser Nickel Craig Onaping Depth Ni Rim SContact Ni Rim S F/W Milling Base Mill w/ pre-con Precon Precon + mill Figure 6.23 – Impacts on Power Requirements for Full Ore Beneficiation 18.104.22.168 Impacts on Overall Energy Usage The individual energy impacts evaluated in the previous sections were combined in order to evaluate the potential overall energy savings achievable at each operation through the application of pre-concentration. The degree of savings varies principally with the overall throughput, depth, haul distance and ultimately the degree of waste rejection. Individual energy savings have been converted into cost savings at the applicable energy cost of $0.057/kWh and $5000/MW demand, and offset against the additional energy cost for the pre- concentration step in order to present a balanced picture. The additional GHG credit for the overall energy saving has also been calculated at $0.01/kWh; the combined result is presented in Table 6.12. Table 6.12 – Overall Projected Impact on Energy Costs at Xstrata Nickel’s Ontario Operations Operation Montcalm Thayer Lindsley Fraser Copper Fraser Nickel Craig Onaping Depth Ni Rim S Hoisting $84,209 $277,816 $106,316 $144,077 $503,526 $398,140 Haul $786,583 $302,422 $884,600 Pre-con -$282,564 -$177,593 -$129,618 -$206,232 -$270,606 -$270,606 -$245,866 Grinding $367,053 $156,510 $160,848 $204,137 $261,724 $268,455 $258,883 Milling $1,030,760 $542,396 $275,969 $669,496 $1,048,726 $1,041,994 $876,840 GHG Credit $138,054 $65,907 $42,466 $56,164 $85,940 $112,033 $157,708 Total $2,039,885 $973,852 $627,480 $829,880 $1,269,860 $1,655,402 $2,330,305 175 Direct savings vary from $585 000 per annum at Fraser Copper to $2 173 000 per annum at Nickel Rim. Savings at Nickel Rim are maximized due to the depth of the deposit, a high degree of waste rejection obtained and the extreme haul distance planned for the operation. Accounting for the potential additional greenhouse gas credits, total savings range from $627 480 at Fraser Copper to $2,330,305 at Nickel Rim. 6.6.4 Conclusions Nine principal ore types of Xstrata Nickel’s Ontario operations have been tested to evaluate the potential for the application of the integrated mining, pre-cocnentration and waste disposal strategy for these deposits. The ores represent the majority of the present and future production Xstrata Nickel. Metallurgical results are good overall, with further work being required to maximize waste rejection and metal recovery for LGBX-type ores. Several of the Footwall ores show potential for the production of a direct shipping ore through pre-concentration. Due to geospatial constraints and the short mine life, pre-concentration is not expected to be utilized to benefit present operations. However, the two future operations, Onaping Depth and Nickel Rim, by nature of their increasing contribution to metal production, as well as their increased depth, and distance from Strathcona Mill present ideal candidates for the application of the proposed approach as a means of reducing waste and improving the efficiency, economics as well as environmental performance and sustainability of these operations. Geotechnical evaluation, backfill mix design and testing has shown that the rejects would be suitable for use in as conventional rockfills or as composite fill, with major benefits to be accrued in terms of fill strength and fill economics. The results of the testwork and subsequent evaluations have been used as inputs into a model to evaluate the overall impact on energy utilization for the mining and processing of these ores of Xstrata Nickel’s Ontario operations. The savings in terms of energy are significant at each operation that was evaluated. It has been shown that hoisting, haulage and milling energy can be reduced by 9% – 15% overall, resulting in energy cost savings (including additional GHG credits) of between $0.6 - $2.3m per annum. Savings are maximized for Montcalm and Thayer Lindsley due to the significant saving in ore- and fill- haulage, and at Nickel Rim due to the high degree of waste rejection achieved, the depth of the operation, and the distance of the planned ore haulage on surface. Additional savings opportunities exist through the potential production of direct shipping ores, as well as the opportunity, with some modifications to the existing flowsheets, to produce 176 custom feeds for delivery to the Strathcona Mill, Clarabelle Mill, or in the case of direct shipping concentrates, the Falconbridge or Copper Cliff smelters. These results must be further evaluated in terms of the synergies that have identified between Xstrata’s Nickel and CVRD INCO. The strategic application of pre-concentration to Xstrata Nickel’s Ontario operations has thus been shown to have great potential in terms of reducing the energy requirements, and thus the Greenhouse Gas emissions, of a group of major Canadian mining operations.
UBC Theses and Dissertations
Integrated mining, pre-concentration and waste disposal systems for the increased sustainability of hard… Bamber, Andrew Sherliker 2008
Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.
- 24-ubc_2008_spring_bamber_andrew.pdf [ 10.54MB ]
- JSON: 24-1.0066375.json
- JSON-LD: 24-1.0066375-ld.json
- RDF/XML (Pretty): 24-1.0066375-rdf.xml
- RDF/JSON: 24-1.0066375-rdf.json
- Turtle: 24-1.0066375-turtle.txt
- N-Triples: 24-1.0066375-rdf-ntriples.txt
- Original Record: 24-1.0066375-source.json
- Full Text