CAVE-TO-MILL: MINE AND MILL INTEGRATION FOR BLOCK CAVE MINES by B.Eng., The University of Sydney, 2005 M.A.Sc., The University of British Columbia, 2012 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Mining Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) November 2018 © Stefan Nadolski, 2018 Stefan Nadolski ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: CAVE-TO-MILL: MINE AND MILL INTEGRATION FOR BLOCK CAVE MINES submitted by Stefan Nadolski in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Mining Engineering Examining Committee: Professor Bern Klein Supervisor Dr. Davide Elmo Co-supervisor Dr. Craig Hart Supervisory Committee Member Professor Scott Dunbar University Examiner Professor David Dreisinger University Examiner Additional Supervisory Committee Members: Allan Moss Supervisory Committee Member iii Abstract Population growth and economic development are expected to increase future global copper demand. The depletion of significant near-surface deposits and advances in detecting deeply buried ore has led to the mining industry progressively exploring further below the surface to discover new copper deposits. Accordingly, block and panel cave mining methods are being increasingly proposed as they allow massive, deeply situated ore-bodies to be mined economically. To improve the productivity of a mining method that will be used to excavate a growing proportion of global copper supply, an integrated mine and mill approach for planning and operating block cave mines, termed Cave-to-Mill, was developed. Key distinguishing features of cave mining, in comparison to other mining methods, are the uncertainty in the size of rock being fed to the mill and the lack of selectivity. As part of the Cave-to-Mill framework, fragmentation and sensor-based sorting studies were carried out at the New Afton block cave mine to investigate opportunities to improve overall productivity. Cave fragmentation is a key cave-to-mill parameter as it has implications on the productivity of both mining and milling processes. Fragmentation measurements of drawpoint muck, comminution tests and calibrated mill models were used to assess the impact of variations in feed size and hardness on New Afton mill performance. Analysis of historical mine and mill data showed that mill feed size and subsequently mill throughput are sensitive to the areas being mucked within the cave. A sensor-based ore sorting study, incorporating bulk and particle sorting systems, showed that rock from the New Afton copper-gold porphyry deposit is amenable to prompt gamma neutron activation analysis, and to X-Ray fluorescence sensors. A conceptual flowsheet, where both technologies are used as separate unit operations, was evaluated. It was found that the sorting concept demonstrated an improvement in the net smelter return of excavated material. Results from the study were used to develop a method to design and evaluate a block cave for the case where sensor-based sorting systems are included in the flowsheet. iv Lay Summary An integrated mine and mill approach, termed Cave-to-Mill, was developed to improve the productivity of cave mines, which will be important contributors to future copper supply. In comparison to other mining methods, key distinguishing features of cave mining are the uncertainty in the size of rock being fed to crushing processes and the lack of selectivity between ore and waste within cave excavations. Research with the New Afton block cave mine showed that implementation of grade sensors provides significant opportunity to address the lack of selectivity and improve project value. Analysis of caved rock size and mill throughput indicated that cave mine productivity can be improved by including rock size and hardness data in the decision making that goes into planning underground production schedules. Overall, results showed that the proposed cave-to-mill approach stands to improve the productivity of a mining method that will be increasingly applied. v Preface This dissertation is ultimately based on outcomes of the Cave-to-Mill research project with New Gold Inc. A version of Section 2.9 has been published in Mining Technology [Nadolski, S., Klein, B., Elmo, D., Scoble, M. Cave-to-Mill: a Mine-to-Mill approach for block cave mines, Transactions of the Institutions of Mining & Metallurgy, Section A, 2015. Vol 124, Issue 1, Pages 47-55, DOI: 10.1179/1743286315Y.0000000001]. I was the lead researcher, responsible for concept development, literature review, data collection, analysis and manuscript composition. Elmo D. and Klein B. were involved in development of the Cave-to-Mill flowchart, Figure 2.9. All authors contributed with manuscript edits. A version of Chapter 4 has been published in Mining Technology [Nadolski, S., Munkhchuluun, M., Klein, B., Elmo, D., Hart, C. Cave fragmentation in a Cave-to-Mill context at the New Afton Mine part I: fragmentation and hang-up frequency prediction, Transactions of the Institutions of Mining & Metallurgy, 2017. Vol 127, Issue 2, Pages 75-83, DOI: 10.1080/14749009.2017.1351115]. I was the lead researcher, responsible for the literature review, sample collection, test work, data collection, analysis and manuscript composition. Munkhchuluun, M. assisted with image-based size analysis of drawpoint photos and review of data. Hart C. assisted with the geological description of the New Afton deposit. All authors contributed with manuscript edits. A version of Chapter 5 has been published in Mining Technology [Nadolski, S., O’Hara, C., Klein, B., Elmo, D., Hart, C. Cave fragmentation in a Cave-to-Mill context at the New Afton Mine part II: implications to mill performance, Transactions of the Institutions of Mining & Metallurgy, 2018. Vol 127, Issue 3, Pages 155-166, DOI: 10.1080/25726668.2018.1437334]. I was the lead researcher, responsible for the literature review, sample collection, test work, data collection, analysis and manuscript composition. O’Hara, C. assisted with data collection and review. All authors contributed with manuscript edits. vi A version of Section 6.1 has been published in the proceedings of MassMin2016 [Nadolski, S., Liu, Y., Klein, B., Elmo, D., Scoble, M. and Scholar, J. Investigation into the implementation of sensor-based ore sorting systems at a block caving operation, Seventh International Conference and Exhibition on Mass Mining, May 2016, Pages 393 - 399]. I was the lead researcher, responsible for the literature review, sample collection, test work, data collection, analysis and manuscript composition. Scholar, J. assisted with test work and review. All authors contributed with manuscript edits. A version of Sections 6.2 to 6.3 has been published in the proceedings of the Canadian Mineral Processors Conference [Nadolski, S., Klein, B., Elmo, D., Samuels, M., Hart. C. Evaluation of cave-to-mill opportunities at the New Afton Mine. The 50th Annual Canadian Mineral Processors Conference, Ottawa, Canada. January 2018]. I was the lead researcher, responsible for the literature review, sample collection, test work, data collection, analysis and manuscript composition. All authors contributed with manuscript edits. A version of Chapter 7 has been published in Minerals Engineering [Nadolski, S., Samuels, M. Klein, B., Elmo, D., Hart, C. J. Evaluation of bulk and particle sensor-based sorting systems for the New Afton block caving operation. Minerals Engineering, Volume 121, 2018, Pages 169-179, ISSN 0892-6875, DOI: 10.1016/j.mineng.2018.02.004]. I was the lead researcher, responsible for the literature review, sample collection, test work, data collection, analysis and manuscript composition. All authors contributed with manuscript edits. A version of Chapter 8 has been accepted for publication in the proceedings of Caving2018 [Nadolski, S., Klein, B., Hart, C. J., Moss, A., Elmo, D. An approach to evaluating block and panel cave projects for sensor-based sorting applications. Fourth International Symposium on Block and Sublevel Caving, Australian Centre for Geomechanics, October 2018]. I was the lead researcher, responsible for the literature review, modelling, production simulations, analysis and manuscript composition. All authors contributed with manuscript edits. vii Table of Contents Abstract ......................................................................................................................................... iii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ........................................................................................................................ vii List of Tables ................................................................................................................................ xi List of Figures ............................................................................................................................. xiv List of Symbols ......................................................................................................................... xviii List of Abbreviations ................................................................................................................. xxi Acknowledgements .................................................................................................................. xxiii Dedication ................................................................................................................................. xxix Chapter 1: Introduction ................................................................................................................1 1.1 Problem Statement .......................................................................................................... 1 1.2 Research Objectives ........................................................................................................ 2 1.3 Cave-to-Mill Project Overview ....................................................................................... 3 1.4 Thesis Outline ................................................................................................................. 4 Chapter 2: Literature Review .......................................................................................................6 2.1 Block and Panel Caving Methods ................................................................................... 6 2.2 Mine-to-Mill ................................................................................................................... 9 2.3 Cave Evaluation and Production scheduling ................................................................ 12 2.4 Rock Mass Characterization ......................................................................................... 14 2.5 Cave Fragmentation ...................................................................................................... 16 2.6 Gravity Flow of Caved Material ................................................................................... 21 2.7 Sensor-based Sorting .................................................................................................... 24 2.8 Cave-to-Mill Concept ................................................................................................... 28 2.9 Summary of Literature Review ..................................................................................... 31 Chapter 3: Site Description and Methodology ..........................................................................33 3.1 Introduction ................................................................................................................... 33 3.2 Site Description: New Afton Mine ............................................................................... 33 3.3 Fragmentation Measurements ....................................................................................... 35 viii 3.4 Drawpoint Sampling and Test Regimen ....................................................................... 37 3.5 Mill Sampling and Surveying ....................................................................................... 40 3.6 Primary Crusher and Mill Modelling ............................................................................ 42 3.7 Rock Type Specification and Characterization ............................................................. 44 3.8 Sensor-based Sorting .................................................................................................... 46 3.9 Cave and Mill Comminution Testing ........................................................................... 49 3.10 Block Cave Planning and Scheduling ........................................................................... 52 Chapter 4: Predicting Cave Fragmentation and Hang-Up Frequency...................................54 4.1 Introduction ................................................................................................................... 54 4.2 Fragmentation Measurements ....................................................................................... 55 4.3 Comparisons with Block Cave Fragmentation (BCF) Predictions ............................... 57 4.4 Back-calculation of Gyratory Crusher Feed ................................................................. 59 4.5 Fragmentation Size and Rock Type .............................................................................. 61 4.6 Comparison of Hang-Up Frequency to Fragmentation Results .................................... 63 4.7 Relating Fracture Intensity from DFN Models to Fragmentation Size ......................... 65 4.8 Influence of Veining on Fragmentation Size at New Afton ......................................... 69 4.9 Discussion and Conclusions ......................................................................................... 71 Chapter 5: Relating Fragmentation Size to Mill Performance ...............................................74 5.1 Introduction ................................................................................................................... 74 5.2 Analysis of Mine and Mill Data.................................................................................... 74 5.3 Cave-to-Mill Modelling and Simulation Methodology ................................................ 78 5.4 Size Analysis of Drawpoint Samples ............................................................................ 79 5.5 Material Hardness Characterisation .............................................................................. 81 5.6 Mill Surveys, Model Calibration and Validation .......................................................... 82 5.7 Mill Model Validation .................................................................................................. 84 5.8 Influence of Feed Size on Simulated Mill Performance ............................................... 88 5.9 Relationships Between Point Load Index, Impact Breakage and Abrasion ................. 89 5.10 Discussion ..................................................................................................................... 92 5.11 Conclusions ................................................................................................................... 93 Chapter 6: Heterogeneity Analysis for Assessment of Sorting Potential................................94 6.1 Introduction ................................................................................................................... 94 ix 6.2 Evaluation of Distributional Heterogeneity and Implications on Bulk Sorting ............ 96 6.3 Analysis of Heterogeneity at the Extraction Level ....................................................... 97 6.4 Variation of Grade with Size ........................................................................................ 99 6.5 Discussion and Conclusions ....................................................................................... 101 Chapter 7: Bulk and Particle Sorting System Evaluation .....................................................102 7.1 Introduction ................................................................................................................. 102 7.2 PGNAA Bulk Sensor Evaluation ................................................................................ 103 7.3 XRF Particle Sensor Evaluation ................................................................................. 106 7.4 Sorting System Evaluation .......................................................................................... 110 7.5 XRF Particle-Sorting Models ..................................................................................... 111 7.6 Process and Cost Assumptions ................................................................................... 114 7.7 Application of Sorter Models to Grade Block Models ............................................... 118 7.8 Discussion and Conclusions ....................................................................................... 120 Chapter 8: A Methodology for Cave Evaluation with Sensor-Based Sorting ......................123 8.1 Introduction ................................................................................................................. 123 8.2 Description of the Cave Evaluation Method ............................................................... 123 8.3 Relating Sorting Performance to Head Grade ............................................................ 124 8.4 Nominating Footprint Geometry, Orientation and Elevation ..................................... 125 8.5 Applying Models to Mixed Material in Draw Cones ................................................. 127 8.6 Production Scheduling and Evaluation ....................................................................... 128 8.7 Discussion and Conclusions ....................................................................................... 129 Chapter 9: Conclusions and Recommendations .....................................................................131 9.1 Main Outcomes ........................................................................................................... 131 9.2 Recommendations for Future Work ............................................................................ 134 Chapter 10: Claims of Original Contributions .......................................................................136 References ...................................................................................................................................138 Appendices ..................................................................................................................................153 Vertimill Operational Data ................................................................................. 153 An Abbreviated DropWeight Test Method for Lump Rock .............................. 155 B.1 Introduction ............................................................................................................. 155 B.2 Background ............................................................................................................. 155 x B.3 Methodology ........................................................................................................... 156 B.4 Comparison of Testing Methods ............................................................................. 158 B.5 Conclusions and Recommendations ....................................................................... 162 PGNAA and XRF Test Results .......................................................................... 164 Point Load Test Results ...................................................................................... 168 DropWeight Test Results .................................................................................... 206 Bond Ball Mill Index Results ............................................................................. 222 Drawpoint Logs .................................................................................................. 235 Image-based Size Distributions .......................................................................... 238 Mill Survey Results for Survey Carried Out March 3rd, 2015 ............................. 243 Mill Survey Results for Survey Carried Out March 16th, 2016 ........................... 247 Sample Preparation Flowsheet for PGNAA Testing .......................................... 257 xi List of Tables Table 3.1 Guideline for macro-description of rocks ..................................................................... 45 Table 3.2 Specific energy and size summary for DropWeight testing ......................................... 51 Table 4.1 Summary of draw point muck images .......................................................................... 56 Table 5.1 New Afton mill feed and geological unit characteristics .............................................. 82 Table 5.2 Summary of New Afton mill surveys ........................................................................... 83 Table 5.3 Comparison of measured and simulated mill performance .......................................... 84 Table 5.4 Equipment and process constraints for simulations ...................................................... 85 Table 6.1 Chronological mixing / comminution events and their impact on heterogeneity ......... 95 Table 7.1 Main lithology and alteration types and implications on the New Afton operation ... 103 Table 7.2 Measurement repeatability in relation to the material handling system ..................... 105 Table 7.3 Cost and process assumptions .................................................................................... 115 Table 8.1 Cave design and costing parameters for the copper-gold case study ......................... 126 Table 8.2 Mass Balance for a conceptual cave ........................................................................... 129 Table B.1 Abbreviated DropWeight Test Regimen .................................................................... 156 Table B.2 Comparison of Axb results ......................................................................................... 162 Table C.1 Bulk samples used for PGNAA testing. ..................................................................... 165 Table D.1 Point load test results for mill feed sample taken in March 2016 .............................. 169 Table D.2 Point load test results for drawpoint sample B13S taken Nov 2015.......................... 173 Table D.3 Point load test results for drawpoint sample D7S taken Nov 2015 ........................... 176 Table D.4 Point load test results for drawpoint sample D11N taken Nov 2015 ......................... 179 Table D.5 Point load test results for drawpoint sample D11S taken Nov 2015 ......................... 182 Table D.6 Point load test results for drawpoint sample D38N taken Nov 2015 ......................... 185 Table D.7 Point load test results for drawpoint sample E13S taken Nov 2015 .......................... 188 Table D.8 Point load test results for drawpoint sample E15N taken Nov 2015 ......................... 191 Table D.9 Point load test results for drawpoint sample E23N taken Nov 2015 ......................... 194 Table D.10 Point load test results for drawpoint sample E30N taken Nov 2015 ....................... 197 Table D.11 Point load test results for drawpoint sample F33S taken Nov 2015 ........................ 200 Table D.12 Point load test results for drawpoint sample G40N taken Nov 2015 ....................... 203 xii Table E.1 DropWeight test results for mill feed sample collected March, 2016 ........................ 206 Table E.2 DropWeight test results for 2015 mill feed sample, SMC method ............................ 210 Table E.3 DropWeight test results for drawpoint sample B13S (collected Nov, 2015) ............. 211 Table E.4 DropWeight test results for drawpoint sample D7S (collected Nov, 2015) ............... 212 Table E.5 DropWeight test results for drawpoint sample D11N (collected Nov, 2015) ............ 213 Table E.6 DropWeight test results for drawpoint sample D11S (collected Nov, 2015) ............. 214 Table E.7 DropWeight test results for drawpoint sample D38N (collected Nov, 2015) ............ 215 Table E.8 DropWeight test results for drawpoint sample E13S (collected Nov, 2015) ............. 216 Table E.9 DropWeight test results for drawpoint sample E15N (collected Nov, 2015) ............ 217 Table E.10 DropWeight test results for drawpoint sample E23N (collected Nov, 2015) .......... 218 Table E.11 DropWeight test results for drawpoint sample E30N (collected Nov, 2015) .......... 219 Table E.12 DropWeight test results for drawpoint sample F33S (collected Nov, 2015) ........... 220 Table E.13 DropWeight test results for drawpoint sample G40N (collected Nov, 2015) .......... 221 Table F.1 Bond ball mill work index test results for mill feed sample (2016) ........................... 222 Table F.2 Bond ball mill work index test results for mill feed sample (2015) ........................... 223 Table F.3 Bond ball mill work index test results for drawpoint sample B13S .......................... 224 Table F.4 Bond ball mill work index test results for drawpoint sample D7S ............................. 225 Table F.5 Bond ball mill work index test results for drawpoint sample D11N .......................... 226 Table F.6 Bond ball mill work index test results for drawpoint sample D11S .......................... 227 Table F.7 Bond ball mill work index test results for drawpoint sample D38N ......................... 228 Table F.8 Bond ball mill work index test results for drawpoint sample E13S .......................... 229 Table F.9 Bond ball mill work index test results for drawpoint sample E15N ......................... 230 Table F.10 Bond ball mill work index test results for drawpoint sample E23N ........................ 231 Table F.11 Bond ball mill work index test results for drawpoint sample E30N ....................... 232 Table F.12 Bond ball mill work index test results for drawpoint sample F33S ........................ 233 Table F.13 Bond ball mill work index test results for drawpoint sample G40N ....................... 234 Table G.1 Logs of rock types reporting to drawpoints ............................................................... 235 Table G.2 Logs of rock types reporting to drawpoints ............................................................... 236 Table G.3 Rock types identified in mill feed sample (collected March, 2016) .......................... 237 xiii Table H.1 Drawpoint size distributions from image-based measurements and sieving ............. 238 Table H.2 Drawpoint size distributions from image-based measurements ................................ 239 Table H.3 Drawpoint size distributions from image-based measurements ................................ 239 Table H.4 Drawpoint size distributions from image-based measurements ................................ 240 Table H.5 Drawpoint size distributions from image-based measurements ................................ 240 Table H.6 Drawpoint size distributions from image-based measurements ................................ 241 Table H.7 Drawpoint size distributions from image-based measurements ................................ 241 Table H.8 Drawpoint size distributions from image-based measurements ................................ 242 Table I.1 Summary of samples taken March 3rd, 2015 .............................................................. 243 Table I.2 Summary of samples taken March 3rd, 2015 ............................................................... 244 Table I.3 Size distributions for samples taken March 3rd, 2015 ................................................. 245 Table J.1 Summary of samples taken March 16th, 2016 ............................................................ 247 Table J.2 Size distributions of samples taken March 16th, 2016 ................................................ 248 Table J.3 Size distributions of samples taken March 16th, 2016 ................................................ 249 Table J.4 Size distributions of samples taken March 16th, 2016 ................................................ 250 Table J.5 Outline of measured and calculated streams ............................................................... 253 xiv List of Figures Figure 1.1 Cave-to-Mill project team and scope ............................................................................ 3 Figure 2.1 Cross-sectional draw zone for different cut-off grades at the Northparkes Mine ......... 8 Figure 2.2 Examples of caving project schedules ......................................................................... 12 Figure 2.3 Schematic of block model to slice file conversion in PC-BC ..................................... 14 Figure 2.4 Relation of tonnes drawn per secondary breakage event and percentage extracted .... 18 Figure 2.5 Sequence to implement intensive pre-conditioning at Cadia East .............................. 20 Figure 2.6 Conceptual flow of material in a cross-section of the cave ......................................... 22 Figure 2.7 Cave Tracker detectors and beacons within fragmented muck pile ............................ 24 Figure 2.8 Sortability of ores ........................................................................................................ 27 Figure 2.9 Cave-to-Mill Concept .................................................................................................. 30 Figure 3.1 Geological and fault structures at the extraction level ................................................ 34 Figure 3.2 Simplified New Afton flowsheet ................................................................................. 35 Figure 3.3 Fragmentation analysis of drawpoint F10N ................................................................ 36 Figure 3.4 PortaMetricsTM Tablet ............................................................................................... 37 Figure 3.5 Test regimen for the core of the test program ............................................................. 39 Figure 3.6 Simplified New Afton flowsheet and sampling points ................................................ 40 Figure 3.7 Diverted mill feed sample (2.2 tonnes) from the underground to surface conveyor. .. 41 Figure 3.8 Collected +50 mm samples from drawpoint D11N ..................................................... 45 Figure 3.9 Sample being loaded into Geoscan-M with conveyor belt cut-off .............................. 48 Figure 3.10 JK DropWeight test unit at the NBK Institute of Mining Engineering ..................... 50 Figure 4.1 Size distribution curves for secondary fragmentation at New Afton .......................... 57 Figure 4.2 Size distribution curves for secondary fragmentation at New Afton, B2 cave ........... 59 Figure 4.3 Modelled crusher product and feed sizes .................................................................... 61 Figure 4.4 Secondary fragmentation size in relation to carbonate and fault rock content ............ 62 Figure 4.5 Hang-Up Frequency for East and West Caves ............................................................ 63 Figure 4.6 100% passing sizes of drawpoint muck....................................................................... 64 Figure 4.7 80% passing sizes of drawpoint muck......................................................................... 64 Figure 4.8 Hang-up frequency for HOD of 60 to 100 metres ....................................................... 65 Figure 4.9 In situ fragmentation based on DFN analysis .............................................................. 66 xv Figure 4.10 Geostatistical P32 block model ................................................................................... 67 Figure 4.11 Hang-up frequency and fracture intensity P32 ........................................................... 68 Figure 4.12 Hang-up frequency and fracture intensity P32 for three HOD ranges ...................... 69 Figure 4.13 Drawpoint sample E13S (West Cave) taken November, 2015 (+50 mm) ................ 70 Figure 5.1 New Afton drawpoint layout ....................................................................................... 75 Figure 5.2 Mill feed size and throughput ...................................................................................... 76 Figure 5.3 Mill feed size and proportion of material from drawpoints within B1 Cave ............. 77 Figure 5.4 Mill throughput and the proportion of material from B1 Cave drawpoints ................ 78 Figure 5.5 Example of estimated gyratory crusher feed size for sample D7S .............................. 81 Figure 5.6 Simulated mill throughput rates and mill feed size .................................................... 86 Figure 5.7 Simulated mill throughput rates and DropWeight Index parameters .......................... 87 Figure 5.8 DropWeight Index and measured F80 size of sample drawpoint muckpiles .............. 87 Figure 5.9 Pearson correlation coefficient for material sizes and simulated mill throughput. ..... 89 Figure 5.10 Point Load Index IS(50) and DropWeight Index results ........................................... 90 Figure 5.11 Point Load Index IS(50) and Abrasion, ta, results for twelve samples ..................... 90 Figure 5.12 Abrasion Index, ta, and measured F80 size of drawpoint muck ............................... 91 Figure 5.13 Point Load Index IS(50) and measured F80 size of drawpoint muck ....................... 91 Figure 6.1 Distribution heterogeneity of caved muck and bulk sorter performance .................... 97 Figure 6.2 Distribution heterogeneity of caved muck .................................................................. 98 Figure 6.3 Distribution heterogeneity and average draw point grade ........................................... 99 Figure 6.4 Copper grade of different size fractions in hypogene drawpoint samples ................ 100 Figure 6.5 Copper grade of different size fractions in drawpoint samples ................................. 100 Figure 7.1 Predicted copper grades of bulk samples using PGNAA sensors ............................. 104 Figure 7.2 Predicted gold grades of bulk samples using PGNAA sensors ................................. 106 Figure 7.3 Predicted copper grades from surface readings ......................................................... 107 Figure 7.4 Particle sorting performance based on pXRF testing of 291 particles ...................... 108 Figure 7.5 Predicted copper grades from surface readings taken under dynamic conditions ..... 109 Figure 7.6 Particle sorting performance from testing under dynamic conditions ...................... 110 Figure 7.7 Simplified flowsheet of the mine and sorting systems .............................................. 111 Figure 7.8 Constitution heterogeneity based on surface readings of rocks ................................ 113 Figure 7.9 Constitution heterogeneity and Cu grade of underground samples........................... 113 xvi Figure 7.10 NSR and copper recovery in relation to mass pull .................................................. 117 Figure 7.11 Economic mass pull and corresponding copper recovery ...................................... 118 Figure 7.12 NSR in relation to head grade and sorting option ................................................... 120 Figure 8.1 Summary of approach to cave-sorting evaluation ..................................................... 124 Figure 8.2 Conceptual cave footprint for a proposed extraction level elevation ........................ 127 Figure 8.3 Example production schedule runs for a conceptual cave ......................................... 128 Figure A.1 Vertimill operating performance and ball mill main motor requirements ................ 154 Figure B.1 Distribution of particle mass for sample #1 .............................................................. 158 Figure B.2 Distribution of specific energies for 11 samples ...................................................... 159 Figure B.3 Distribution of specific energies for 11 samples, ..................................................... 160 Figure B.4 Mean squared error using a typical DropWeight approach ...................................... 161 Figure B.5 Mean squared error using the proposed DropWeight method .................................. 162 Figure C.1 Rock type and pXRF measurement for Ni................................................................ 164 Figure C.2 Rock type pXRF surface measurement for Cr. ......................................................... 164 Figure C.3 PGNAA and ICP measurements for Ni .................................................................... 165 Figure C.4 PGNAA and ICP measurements for Cr .................................................................... 166 Figure C.5 Ratio of copper to sulfur from PGNAA measurements ............................................ 166 Figure C.6 Copper grade of drawpoint sampleand copper grade of fines (-2 mm) .................... 167 Figure C.7 Fe content in fines (-2 mm), and the proportion of carbonate rock .......................... 167 Figure D.1 Point load test results for Mill Feed (sample collected March 2016) ....................... 172 Figure D.2 Point load test results for drawpoint sample B13S taken Nov 2015 ........................ 175 Figure D.3 Point load test results for drawpoint sample D7S taken Nov 2015 .......................... 178 Figure D.4 Point load test results for drawpoint sample D11N taken Nov 2015 ....................... 181 Figure D.5 Point load test results for drawpoint sample D11S taken Nov 2015 ........................ 184 Figure D.6 Point load test results for drawpoint sample D38N taken Nov 2015 ....................... 187 Figure D.7 Point load test results for drawpoint sample E13S taken Nov 2015 ........................ 190 Figure D.8 Point load test results for drawpoint sample E15N taken Nov 2015 ........................ 193 Figure D.9 Point load test results for drawpoint sample E23N taken Nov 2015 ........................ 196 Figure D.10 Point load test results for drawpoint sample E30N taken Nov 2015 ...................... 199 Figure D.11 Point load test results for drawpoint sample F33S taken Nov 2015 ....................... 202 Figure D.12 Point load test results for drawpoint sample G40N taken Nov 2015 ..................... 205 xvii Figure E.1 Breakage index t10 and input energy with the standard A & b model ..................... 208 Figure E.2 Breakage index t10 and input energy and fitted M and fmat model ......................... 208 Figure E.3 Breakage index t10 and tn for mill feed .................................................................... 209 Figure E.4 Breakage index t10 and energy for drawpoint sample B13S .................................... 211 Figure E.5 Breakage index t10 and energy for drawpoint sample D7S ...................................... 212 Figure E.6 Breakage index t10 and energy for drawpoint sample D11N ................................... 213 Figure E.7 Breakage index t10 and energy for drawpoint sample D11S .................................... 214 Figure E.8 Breakage index t10 and energy for drawpoint sample D38N ................................... 215 Figure E.9 Breakage index t10 and energy for drawpoint sample E13S .................................... 216 Figure E.10 Breakage index t10 and energy for drawpoint sample E15N ................................. 217 Figure E.11 Breakage index t10 and energy for drawpoint sample E23N ................................. 218 Figure E.12 Breakage index t10 and energy for drawpoint sample E30N ................................. 219 Figure E.13 Breakage index t10 and energy for drawpoint sample F33S .................................. 220 Figure E.14 Breakage index t10 and energy for drawpoint sample G40N ................................. 221 Figure G.1 Mill feed sample (~2 tonnes) and example of rock type groupings (+50 mm) ........ 237 Figure I.1 Size distributions for samples taken March 3rd, 2015 ................................................ 246 Figure J.1 Size distributions for the SAG mill circuit (March, 2016) ........................................ 251 Figure J.2 Size distributions for the SAG mill circuit (March, 2016) ........................................ 251 Figure J.3 Mass balances relating to SAG mill screen undersize calculation ............................ 252 Figure J.4 Mass balance calculation #1 for SAG mill screen undersize ..................................... 254 Figure J.5 Mass balance calculation #2 for SAG mill screen undersize ..................................... 255 Figure J.6 Size distributions SAG mill screen undersize ............................................................ 256 Figure K.1 Sample preparation procedure for sulfide samples ................................................... 257 Figure K.2 Sample preparation procedure for oxide sample ...................................................... 258 xviii List of Symbols ϒ sorter Proportion of sorter feed mass, or mass yield, that reports to sorter concentrate A Parameter in energy-breakage equation describing used in application to DropWeight testing. Represents the maximum breakage for infinite energy input ai Grade of an individual particle within a lot in Constitutional Heterogeneity aL Weighted-average grade of an entire lot of material in Constitutional or Distributional Heterogeneity equations an Grade of an individual group in a Distribution Heterogeneity equation b Parameter in energy-breakage equation describing used in application to DropWeight testing. Represents the maximum breakage for infinite energy input 𝐶?̂? Predicted copper content of a fragment based on a sensor measurement Cuavg Average XRF copper measurement for a fragment where multiple faces are measured Cumax Maximum XRF copper measurement for a fragment where multiple faces are measured effsort Efficiency of particle sorting systems downstream of grade sensors f Head grade of material F80 80 % passing size of feed material F50 50 % passing size of feed material fines dilution Grade of fines (-12.5 mm) as a percentage of head grade fmill feed Copper feed grade in % xix IS(50) Size-corrected point load strength index. The original point load strength index value multiplied by a factor to normalize the value that would have been obtained with a diametral test of diameter 50 mm (ASTM, 2013) K Size modulus in the Gaudin-Schuhmann equation L A Lot of particles or groups of material in Constitutional and Distributional Heterogeneity equations m Slope parameter in the Gaudin-Schuhmann equation Mi Mass of an individual particle within a lot in Constitutional Heterogeneity ML Weighted-average mass of an entire lot of material in Distribution Heterogeneity equations Mn Mass of an individual group in Distributional Heterogeneity equations M(x) Weight of undersize in the Gaudin-Schuhmann equation n Number of groups in Distributional Heterogeneity equations Ng Number of groups in a lot in Distributional Heterogeneity equations NF Number of fragments (particles) in a lot in Constitution Heterogeneity equations P12.5mm Mass proportion of material that passes a 12.5 mm aperture screen P32 Fracture area divided by unit volume (m2/m3). A 3D measure of fracture intensity P10 Number of fractures per unit length of borehole (1/m). A 1D measure of fracture intensity R Metal recovery Rflotation Metal recovery by flotation xx RR Reduction Ratio Rsort Proportion of metal in sorter feed that is recovered by a sorter t10 Abrasion parameter determined from abrasion testing Wio Operating work index. xxi List of Abbreviations BBWI Bond Ball Mill Work Index BCF Block Cave Fragmentation (software) BPC Blast Pre-Conditioning CH Constitutional Heterogeneity DE Discrete Element DFN Discrete Fracture Network DH Distributional Heterogeneity DWI DropWeight Index DWT DropWeight FE Finite Element HF Hydraulic Fracturing HFEMS High Frequency Electro Magnetic Susceptibility HOD Height of Draw HPGR High Pressure Grinding Rolls HR Hydraulic Radius ICP Inductively Coupled Plasma LHD Load Haul Dump MRMR Mine Rock Mass Rating xxii NMR Nuclear Magnetic Resonance NPV Net Present Value NSR Net Smelter Return PCBC Personal Computer Block Cave (software) PGNAA Prompt Gamma Neutron Activation Analysis PLI Point Load Index pXRF Portable X-Ray Fluorescence QEMSCAN Quantitative Evaluation of Minerals by SCANning RMR Rock Mass Rating ROM Run-Of-Mine RQD Rock Quality Designation SAG Semi-Autogenous Grinding SRM Synthetic Rock Mass SWIR Short Wave InfraRed XRF X-Ray Fluorescence UCS Unconfined Compressive Strength xxiii Glossary This glossary defines the main technical terms as they are used in this thesis. Where practical, the definitions follow “Caving Terminology & Glossary” by Van As (2010). Alteration: The change in the mineralogic composition of a rock due to various natural processes, especially by the action of hydrothermal solutions Apex Level: The level in a cave mine which provides access to the apex of a crinkle-cut (chevron shaped) undercut. Usually developed to ensure the continuity of undercut blasts Aspect Ratio (caveability): The ratio of cave footprint length to width when considered in plan view Aspect Ratio (fragment): A parameter used in Block Cave Fragmentation software that relates cave fragment volume, surface area and length Best Height of Draw: The column height above a drawpoint that provides the highest dollar value for a given set of mining and processing costs, and factors for product revenue and metal recovery Block Model: A set of specifically sized blocks used to describe the characteristics of a rock mass. Although the size of the blocks is usually the same, each block has unique characteristics such as grade, density, rock type and any other features that are known and relevant to mining and/or processing Caveability: The ability of a rock mass to cave for a designated layout Cave Back: The under-surface of the in situ, but possibly disturbed, rock above a cave or muckpile of caved ore (located above the extraction level) Caving: The gravitational collapse of in situ rock from the cave back Caving Rate: The average rate at which the cave naturally propagates upwards Dilution Entry: The percentage of the ore column that has been drawn before dilution appears at the drawpoint (Laubscher, 2000) xxiv Drawbell: An excavated structure, ideally having the shape of an inverted bell, which channels caved or broken rock to a drawpoint Draw Control: The process of controlling the amounts of rock drawn from individual drawpoints in order to achieve a number of mining objectives Drawpoint: The excavated structure on the extraction or production level through which caved or broken rock is loaded and moved to underground material handling systems Drawpoint Spacing: The spacing in plan view between like points in adjacent drawpoints. Usually defined by two distances: the distance between drawpoints that are separated by the major apex and the distance between drawpoints that are separated by the minor apex Drift: A horizontal excavation in or near an orebody driven from one working place to another. It is usually of a relatively small cross section compared with larger sections which are usually referred to as tunnels Extraction Level: The level in a caving mine through which caved or broken rock is extracted and transported away from the cave Fault: A fracture or a fracture zone in crustal rocks along which there has been displacement of the two sides relative to one another parallel to the fracture. The displacement may be a few millimetres or many kilometres long Footprint Finder: An application within Gems Geovia PCBC software which works off geological block models and determines a block cave layout’s elevation, footprint shape and size for a given set of inputs Froth Flotation: The separation of ore and waste minerals made possible by the selective adhesion of air bubbles to mineral or gangue surfaces in a slurry Fragmentation: The process, or the result, of blasting, caving and draw of initially in situ rock Geometallurgy: The combination of geological, metallurgical and other information relating to mine and mill production for use in spatial or geology-based models Hang-up: The wedging of one or more large blocks of rock in or above the drawpoint or drawbell such that they will not move further by gravity xxv Haulage Level: The level in an underground mine through which caved or broken rock is transported away from the production area to crushers or material handling systems (for transport to surface) Height of Draw (HOD): The vertical distance travelled by material reporting to a drawpoint from its original location in the in situ rock mass Hydraulic Radius: In the context of underground mining, the ratio of the surface area to the perimeter of an excavated surface, such as the back of an undercut Hypogene (Rock): Rock formed beneath the surface of the Earth by ascending solutions In Situ: In the natural or original position. Applied to a rock, soil, or stress etc., occurring in the original, unaltered state prior to mining In Situ Fragmentation: The size of naturally occurring blocks within the rock mass prior to being disturbed by caving, pre-conditioning or undercutting Isolated Draw Zone (IDZ): A draw zone isolated from other draw zones as a consequence of being spaced too far away for any interaction to occur with its neighboring drawpoints LHD (Load-Haul-Dump): A mechanical shovel or other machine for loading ore or rock Lithology: A description of the physical characteristics of lump rock or core samples, such as colour, texture, grain size, or composition. Referred to as a summary of the gross physical character of a rock Major Apex: The shaped structure or pillar above the extraction level formed between two adjacent drawpoints that are separated by an extraction (production) drift Minor Apex: The shaped structure or pillar formed between two adjacent drawbells that are located on the same side of an extraction (production) drift. Generally oriented normal to the major apex Monzonite (Rock): Coarse grained plutonic rock containing approximately equal amounts of orthoclase and plagioclase (Mindat, 2018) Orepass: A vertical or inclined passage for the downward transfer of ore. An orepass is driven in ore or country rock and connects a level with a lower level xxvi Picrite (Rock): A dark basaltic rock which is rich in olivine Pre-conditioning: A process, or well-defined group of processes, implemented before the initiation of caving to alter the inherent characteristics of the target rock mass to enhance its caving and fragmentation characteristics Primary Fragmentation: The size reduction that occurs as blocks separate from the cave back at the onset of caving Production Schedule (Draw Schedule): The planned allocation of draw tonnes for each drawpoint over the life of the mine Rilling: The downward movement of material along the surface slope of caved muck located above the extraction level Rock Mass: The rock as it occurs in situ, including all its discontinuities Secondary Fragmentation: The breakage of caved rocks as they fall onto caved (or blasted muck) and move through a draw column to the exit point of drawpoints Supergene (Rock): Rock formed or altered by processes, such as the circulation of meteoric waters, occurring relatively near the surface of the Earth Undercut: A horizontal slot mined to initiate caving of a block or panel xxvii Acknowledgements I would like to express my deepest appreciation to my supervisor Bern Klein for providing a platform to investigate new concepts with mining operations at the beginning of the PhD program. In addition to the support received while carrying out the Cave-to-Mill research, I’d also like to express my gratitude to Bern for providing me with a wide range of opportunities to develop as an engineer since the beginning of my studies with UBC Mining. I am very grateful to my supervising committee, Davide Elmo, Craig Hart and Allan Moss, for providing valuable insight during the research project. Paper writing was thoroughly enjoyable thanks to the timely support received. Davide Elmo’s direction for the geotechnical scope was critical to linking the doctoral work to that of other students within the Cave-to-Mill project group. The research benefited considerably from Craig Hart’s guidance on defining and relating geology to research outcomes. Allan Moss’ input kept the work grounded and was key to assessing the applicability of the research to practical implementation. I would like to extend my sincere thanks to personnel at New Afton and New Gold for supporting the Cave-to-Mill project and for providing access to samples and data from the New Afton block cave operation. Special thanks to Mike Samuels, who had belief in the project and provided guidance throughout. Thanks also to Marty Henning who provided considerable assistance with sampling, data collection and facilitated most of the site work. I gratefully acknowledge the assistance from industry sponsors who provided in-kind support. ALS Metallurgy in Kamloops, BC carried out Bond ball mill work indices, froth flotation tests and assays. MineSense Technologies provided access to an Olympus pXRF and a ConductOre test unit. MotionMetrics provided access to a PortaMetricsTM tablet for on-site work. I would also like to thank the Natural Sciences and Engineering Research Council of Canada for the financial support provided to this research through a Collaborative Research Development grant (Grant No. 11R74149; Mine-to-Mill Integration for Block Cave Mines). xxviii Many thanks to fellow research students of the Cave-to-Mill research project. Munkhtsolmon Munkhchuluun’s and Yubo Liu’s assistance with on-site work and fragmentation analysis was invaluable to the project. Their efforts for ensuring that Discrete Fracture Network research outcomes dovetailed with the scope of the Cave-to-Mill research project are greatly appreciated. I would also like to acknowledge the sensor test work and data analyses carried out by Elberel Erdenebat. I very much appreciate the assistance with sampling, mill surveys and laboratory testing provided by Tugba Cebeci, Daozhen Gong and Ding Zhang. Additionally, I’d like to thank the following students for helping with the geotechnical, metallurgical and sensor-based sorting test programs: Emilio Adriano, Monong Huang, Huaizhe (Harvey) Li, Joshua Scholar, Hiten Sulakh, and Nawoong Yoon. Santiago Seiler and Amit Kumar provided valuable input during the research. I am very grateful to my wife, Yuri Nadolski, for her patience over the course of the research, dealing with moving timelines, and for being an amazing mother to our two daughters, Anika and Mia. I would like to thank my mother, Eva Nadolski, for carefully proofreading manuscripts and providing valuable editorial feedback. xxix Dedication To Anika and Mia1 Chapter 1: Introduction The problem statement and research objectives are presented in this section. An overview of the Cave-to-Mill research team and an outline of the thesis is included at the end of this chapter. 1.1 Problem Statement Population growth and economic development are expected to increase copper demand by 213 to 341% by the year 2050 (Elshkaki et al., 2016). Meeting future copper demand will require a combination of primary raw materials from mining operations and recycled materials (ICA, 2017). Due to the depletion of significant near-surface deposits and advances in detecting deeply buried ore, the mining industry is progressively exploring further below the surface to discover new copper deposits. Approximately 69% of known copper is contained in copper-porphyry deposits (Singer, 2018) and exploration trends indicate that they will continue to account for the majority of copper reserves as discoveries of deeply-situated copper deposits continue. Block and panel caves, which have the lowest mining costs of all underground mining methods, can achieve ore excavation rates which are comparable to that of open pit operations. The solution to supplying future copper demands lies in the discovery of deeply-situated deposits and their excavation through block and panel caving. The productivity of operating and future cave mines, defined as the real dollar output from pay metals in concentrate per unit of capital and operational dollar input, will play a large role in addressing copper demand. Furthermore, the mining method will be pushed to its extremes as deeper and larger deposits will be discovered with advances in exploration methods. Caving of more heterogeneous and competent rock masses will bring about risks associated with, among others, caveability, dilution, ore recovery, ground support and variability in mill feed. The variable and relatively uncontrollable nature of cave fragmentation and lack of selectivity are key distinguishing features of block cave mining when compared to other mining methods. Innovative approaches to planning and operating cave mines will be required to mitigate risks and drive productivity. Operational strategies that have an end-to-end view and avoid silos by taking an integrated approach have been recognized as being conducive to productivity within 2 the mining industry (EY, 2017). The development of an integrated operational approach for block cave mines stands to play an important role in the mining industry’s efforts to meet future copper production needs. 1.2 Research Objectives The goal of this doctoral thesis is to develop an integrated mine and mill approach for planning and operating block cave mines, termed Cave-to-Mill, to improve productivity. Copper-porphyry applications are of primary focus, as they account for the majority of copper reserves and the deposit types where block caving is applied. To develop Cave-to-Mill, the following research objectives were defined for completion with involvement of an operating block cave mine: • Identify key linkages between areas of cave planning and operation, and define opportunities for their coordination to improve cave productivity (Section 2.9) • Relate geology, geotechnical properties, footprint location and operational parameters to cave fragmentation size, which affects productivity at the cave mine and mill (Chapter 4) • Determine the influence of underground production scheduling on mill productivity (Section 5.2) • Identify the size fractions within caved muck that are critical to mill performance (Section 5.8) • Identify relationships between geology, geotechnical data, fragmentation and geometallurgical data for more efficient rock characterization and improved valuation of blocks contained within block models (Section 5.9 and Table 7.1) • Addressing the non-selectivity of the cave mining method, evaluate grade heterogeneity on a bulk scale to identify suitable locations for separate handling of ore and waste underground (Section 6.3) • Evaluate bulk and particle sensor-based sorting systems as a means to add selectivity to the mining method and improve productivity (Chapter 7) 3 • Develop a method to incorporate sensor-based sorting parameters in cave planning and production scheduling software to assess the change in reserve size and cave value (Chapter 8) 1.3 Cave-to-Mill Project Overview At the onset of the Cave-to-Mill project it was recognized that a research team would be required to properly address the broad scope of the project. Three master’s level students were recruited to focus on objectives relating to fragmentation and grade heterogeneity, as described in Figure 1.1. A multi-disciplinary committee from the University of British Columbia covering the fields of geology, geotechnical engineering, mining and mineral processing oversaw the research group. Similarly, a multi-disciplinary group from the sponsoring block cave mine, New Afton, provided input and direction over the course of the project. Figure 1.1 Cave-to-Mill project team and scope Each of the three master’s theses generated valuable outcomes as individual research projects and as contributions to development of the Cave-to-Mill approach. Key contributions from each thesis are shown below. 4 Linking the fracture intensity of an in situ rock mass to block cave mine fragmentation Munkhtsolmon Munkhchuluun • Developed a Discrete Fracture Network (DFN) model of the New Afton rock mass and related fracture intensity to cave fragmentation size (Section 4.7) Characterisation of block cave mining secondary fragmentation Yubo Liu • Compared the outputs of PortaMetricsTM and WipFragTM methods for size measurement of drawpoint muck (Section 3.3) Study of New Afton ore heterogeneity and its amenability to sensor-based ore sorting Elberel Erdenebat • Related bulk scale grade heterogeneity (distribution heterogeneity) to the average grade of material reporting to individual drawpoints (Figure 6.3) 1.4 Thesis Outline Chapter 2 includes a review of literature pertaining to the development of a Cave-to-Mill approach for improved productivity. The Cave-to-Mill concept was developed as an outcome of the literature review and is presented in the same chapter. Chapter 3 describes the New Afton block cave mine, which served as the basis of the research. The methodology used to address focus areas of Cave-to-Mill, such as assessing the impact of cave fragmentation on mill performance and evaluating sensor-based sorting systems, is included in this chapter. 5 Chapter 4 presents findings from a fragmentation study, which relates cave fragmentation and hang-up frequency to geology, geotechnical properties, footprint location and the vertical distance travelled by ore blocks. Chapter 5 identifies drawpoints at the New Afton mine that provide finer mill feed and are associated with increased mill throughput. Results from mill simulations show how mill throughput varies due to changes in mill feed size and hardness. Chapter 6 describes New Afton grade heterogeneity on a bulk scale as an initial step towards evaluation of sensor-based sorting opportunities. Grade data for different size fractions within drawpoint muck is also presented. Chapter 7 describes the results of a bulk and particle sensor-based sorting evaluation. A flowsheet which combines both bulk and particle sorting technologies is proposed. The increase in the net smelter return (NSR) as a result of implementing the flowsheet is presented. The chapter includes a summary of key rock types and their implications on mine and mill productivity. Chapter 8 presents a method to include sensor-based sorting parameters in the outputs of cave planning and production scheduling software. Results from the sorting study, which was carried out at the New Afton block cave mine, are used to demonstrate the method for a conceptual cave in a copper-porphyry deposit. Chapter 9 describes the conclusions drawn from the research. Recommendations for further development of Cave-to-Mill are presented in the same chapter. Chapter 10 outlines the original claims that resulted from this Cave-to-Mill doctoral thesis. 6 Chapter 2: Literature Review This chapter presents a review of literature from seven broad areas of research that are relevant to the development of an integrated mine and mill approach for cave mines. An outcome of the literature was the development of the cave-to-mill concept, which integrates different areas of research with a systems engineering approach. A review of the block and panel caving mining methods is presented in Section 2.1. Mine and mill strategies, based on fragmentation size and geometallurgical information, are presented in Section 2.2. Cave evaluation and production scheduling methods are reviewed in Section 2.3. A review of relevant rock mass characterisation methods is presented in Section 2.4. Models and measurement methods for cave fragmentation size are discussed in Section 2.5. Models for predicting gravity flow and the use of markers for tracking caved ore are reviewed in Section 2.6. An overview of sensor-based sorting, for bulk and particle applications, is shown in Section 2.7. From review of literature, a cave-to-mill concept is put forward in Section 2.8. Reviewed literature is summarized in Section 2.9. 2.1 Block and Panel Caving Methods Block and panel caving are low-cost underground mass mining methods that are suitable for excavating weak, massive and steeply-dipping orebodies. Accordingly, cave mining methods have predominantly been used to mine copper porphyry and diamond bearing deposits. Due to the depletion of near-surface orebodies, it is expected that block and panel caving methods will be increasingly used in the copper-gold industry. Typical production rates for caving operations in massive deposits range from 17,500 tpd at the New Afton mine to 160,000 tpd at the Grasberg complex. Examples of cave mines proposed for British Columbia include the Kwanika Mine, Red Chris underground, Kemess Underground project, and the Iron Cap and Mitchell cave mines of the Kerr-Sulphurets and Mitchell (KSM) project (Moose Mountain Technical Services, 2017; Imperial Metals, 2017; Golder Associates, 2017; Tetra Tech, 2016). Significant future caves proposed worldwide include the Grasberg Block Cave, Oyu Tolgoi’s Hugo North Cave, Resolution Mine and Chuquicamata. 7 In both panel and block cave mines, an excavation is blasted below the base of the ore body, referred to as an undercut, inducing gravity-assisted fragmentation known as caving. Subsequently, caved material flows through a number of funnel like drawbells and is subjected to a series of hauling, crushing, grinding and beneficiation processes. As ore is removed from drawbells by load haul dumps (LHD), the cave back (roof) progresses vertically towards the surface. Although block and panel caving are recognized as being the most cost-efficient underground mining methods, approximately 70 percent of capital costs are consumed before any revenue is generated (Oancea, 2013). In panel caving operations, the orebody is not undercut completely, rather, a number of panels or strips of the orebody are undercut and caved individually until the planned footprint is excavated (Brown, 2003). Generally, variations of panel and block caving methods are implemented at caving operations. For clarity, further use of the term ‘block caving’ will generally refer to both panel and block caving methods. Both block and panel caving methods are associated with low selectivity in terms of the capability of selecting grades within a spatial distribution. The limitation of the method is illustrated in Figure 2.1, which shows the cross-sectional draw zone for different cut-off grades at the E48 Northparkes Mine. The illustration shows that the selectivity of the mining method for material located above the extraction level is practically limited to providing control over the height of each draw column. Underground mining methods that provide greater selectivity, while being applied at lower production rates, include sub-level caving, long-hole stoping, cut-and-fill stoping, and vein mining. 8 Figure 2.1 Cross-sectional draw zone for different cut-off grades at the E48 Northparkes Mine, looking North (Sagawa & Yamatomi, 2009) Significant developments in the caving method have occurred since its first reported implementation in 1895 at iron mines in Michigan (Woo et al., 2013). Over the last 20 years, the following major technological changes, relating to the proposed Cave-to-Mill approach, have been observed from review of literature: • Automation/Semi-automation of LHDs by implementing tele-remote systems (Sandvik, 2012) • Installation of electronic markers to monitor the vertical and horizontal flow of caved ore in real-time (Whiteman et al., 2016) • Development of single-shot drawbell blasting solutions (Lovitt & Degay, 2004, and Dunstan & Popa, 2012) 9 • Development of independent block caves, called macro blocks, to excavate larger footprints and reduce interference between development and production (de Wolfe & Ross, 2016, Fuentes, 2014 and Manca, 2013) • Improvements in conveying systems leading to decline access being a favorable option over shaft access for extraction levels located at greater depths (de Wolfe & Ross, 2016) • Installation of sizers as a more compact alternative to jaw crushers or gyratory crushers (Paredes et al., 2016 and Orozco, 2010) As a result of operational block caving experience and development of technologies, the method is being pushed to new extremes in terms of its application. In particular, increasing draw heights from 550 to over 1000 meters, caving of more heterogeneous ore bodies, and caving of deeper and more competent rock masses (Manca, 2013, Orozco, 2010, Chitombo, 2010 and Flores 2014). From review of trends in the block caving industry, it is envisaged that Cave-to-Mill will be increasingly relevant. Developments in mine automation are conducive to implementation of broader operational strategies, as the installation of extensive monitoring and control equipment allows operational silos to be broken down and systematic approaches to be adopted. Furthermore, caving of more heterogeneous ore bodies will result in a greater variation of ore types, in terms of milling properties, being fed to process plants. 2.2 Mine-to-Mill Mine-to-Mill spans (at a minimum) the domains of geology, geotechnical/rock mechanics, mining and processing (Mckee, 2013). The benefits of Mine-to-Mill approaches in open-pit mines have been well documented over the last 20 years by industry. Historically, the term has been used to define two operational approaches: 1) Optimization of blasting and crushing/grinding processes with the aim of increasing overall productivity, also referred to as a drill-to-mill strategy; 2) Characterisation of the ore body in terms of process performance and using this information within the mine plan and milling approach to improve productivity, 10 widely referred to as geometallurgy. Both operational approaches are relevant to the proposed Cave-to-Mill approach. Mine-to-Mill, Fragmentation and Comminution The productivity of concentrators, that use semi-autogenous (SAG) mills, is known to be sensitive to the competency and size of mill feed; lower throughput rates are associated with coarser and more competent mill feed (Morrell & Valery, 2001). For open-pit operations, studies have shown how improvements in productivity were achieved by adjusting blast designs to reduce the size of mill feed. At the Paddington Gold Operations, implementation of a higher intensity blast design resulted in a 36% improvement in mill throughput and a slight coarsening of the final product size of the ball mill grinding circuit (Kanchibotla et al., 2015). Improvements in throughput were also attributed to changes in ore hardness and process equipment. At the Cerro Corona copper-gold mine, which is an open-pit mine and uses a SAG mill based comminution circuit, a 19.4% increase in mill throughput was achieved primarily by modifying the blast pattern to increase the powder factor (La Rosa et al., 2015). In block and panel cave mines, fluctuations in caved ore hardness and fragment size will also impact the performance of downstream crushing and grinding circuits. Significant challenges in mill operation were experienced as the Palabora mill transitioned from treating material from the open-pit to that of the block cave mine (Condori et al., 2012). A greater variation in material hardness, due to non-selectivity of the mining method, and variation in the size distribution of feed to the autogenous mills prevented Palabora mill operators from achieving design throughput rates. Dance (2016) modelled SAG and ball mill throughput performance for a range of cave fragmentation sizes which were predicted based on fragmentation modelling. Mill throughput performance was estimated to increase by up to 44% as drawpoints within the planned cave (of an anonymous operation) matured. The variable and relatively uncontrollable nature of cave fragmentation necessitates a different approach for mine and mill integration to those used for open-pit operations. Addressing this need, the Cave-to-Mill concept is proposed to give existing and future block cave operations an opportunity to improve production performance. 11 Mine-to-Mill, Geometallurgy The field of geometallurgy includes aspects of geotechnical engineering, process mineralogy, geology, metallurgy, resource modelling, geostatistics and mine planning (Bye, 2010). Typically, geometallurgical characterizations are carried out using core samples and the results are used to quantify the value associated with including an individual block in the mining footprint. The majority of reviewed geometallurgical studies involved the process of defining geometallurgical domains or units. Lotter (2011), describes a geometallurgical unit as an individual or group of ore types that possess a unique set of textural or compositional properties from which it can be predicted that a similar metallurgical performance will result. Bye (2010) mentions that in order to define the inherent variability of an orebody, coordinating smaller volume tests, rather than compositing of core samples, is a dominant trend in the field of geometallurgy. A challenge associated with handling geometallurgical data is the fact that not all geometallurgical characteristics are additive. For example, Bond ball mill work indices are not additive and need to be converted into alternative units through simulation if they are to be used with kriging or inverse distance methods to populate a block model (Barratt et al., 2008). Alternatively, blend-response models, as proposed by Bye (2011), have been identified as a solution to modelling mill performance for a variety of ore types. Many geometallurgical studies have shown strong correlations between geotechnical test data and crushing/grinding indices. Burger et al. (2006) found good correlations between the Point Load Index Is(50) and DropWeight Indices for certain ore types. Advantages of the Point Load Index (PLI) test are the relatively low test costs and short turn-around time in comparison to DropWeight type breakage tests. Proper geotechnical characterization of ore bodies is critical for caving projects, resulting in an abundance of geotechnical data that is available for consideration in a Cave-to-Mill approach. 12 2.3 Cave Evaluation and Production Scheduling Cave evaluation involves the use of planning and scheduling software to determine the associated value of a caving lift for an ore body. Geological block models, which include revenue factors such as mining cost and mill recovery, can be input by users along with the planned extents of the extraction level to determine the best Height of Draw (HOD) for each draw-point and the associated project value (Diering et al., 2010). Examples of cave production schedules are shown in Figure 2.2 for a number of potential caving projects. Evaluation of an ore body requires consideration of a multitude of production scenarios which account for, among others, variations in footprint geometry, orientation, sequence, elevation and height of draw. Gantumur et al. (2016) assessed nearly 54,000 production scenarios when evaluating the Resolution Copper Project; initial evaluation was carried out using the Footprint Finder tool in PCBC production scheduling software (Dassault Systèmes, 2018) to identify high value scenarios, these were evaluated further with detailed production schedule runs in PCBC. Figure 2.2 Examples of caving project schedules – construction, ramp-up, steady stage and ramp-down (Stewart & Butcher, 2016) 13 To date, a number of software applications have been developed to address the need for scheduling underground processes at caving operations. Examples include Linear Programming (LP), Mixed-Integer Linear Programming (MILP), Mixed-Integer Programming (MIP) and Quadratic Programming (QP) based methods (Khodayari et al., 2014). Scheduling programs generally consider ore grade, mine and mill production capacity, and the sequence of draw-point development as primary inputs. In addition to their respective mathematical methods, programs also differ according to the degree to which they consider mining/processing costs, penalty elements, horizontal and vertical mixing, geotechnical constraints and interference between construction and production activities. In direct contrast to the many scheduling programs that are based on directly optimizing cave mine net present value (NPV), Diering (2012) describes a method, available in the PCBC scheduling program, where the cave geometry is set as the objective function while ore grades are set as constraints. Figure 2.3 shows a schematic of the general steps involved in setting up a PC-BC project. Rahal (2003) proposes an MILP-based method which sets the panel profile as the goal of the objective function. The suitability of cave geometry has implications on geotechnical factors, such as subsidence profiles and air gaps, which can lead to critical mining disruptions if they are not appropriately considered within the production schedule. 14 Figure 2.3 Schematic of block model to slice file conversion in PC-BC (Diering, 2010) Production scheduling software offer significant potential for incorporating geometallurgical and sensor-based sorting data for use in the proposed Cave-to-Mill approach. Although schedules are constrained by cave stresses, flow geometry and cave geometry, they do provide the possibility to control properties of the plant feed that affect mill performance when considered as a means to blend ore types in a fashion that increases mill productivity. An example showing that sufficient flexibility exists within production schedules is the practice of blending drawpoint muck according to moisture at the Grasberg block caving operation. The fines content observed at different areas of the extraction level is one of the inputs that is considered when setting the production schedule. The fines are blended such that the moisture in the material handling system is below levels that cause problems at the haulage level (Moss, 2017). 2.4 Rock Mass Characterization Rock mass characterization methods are essential to block cave design. Empirical approaches that are typically referenced for block cave applications include the Mining Rock Mass Rating (MRMR), Barton’s Tunneling Quality Index (Q) and the Mathews stability number, N. 15 The MRMR classification method was introduced for cave mining applications and has been related to estimations for caveability, subsidence angles, failure zone, fragmentation, undercutting sequence and support design (Laubscher, 2011). The method relies on a variation of the Rock Mass Rating (RMR) value, as defined by Bieniawski, and adjusts it to account for in situ and induced stresses, join orientations, stress changes and the effects of blasting, water and weathering (Hoek, 2007). A limitation of empirical rock mass characterization methods, such as MRMR, is the sensitivity of their application to the judgement of the practitioner. Additionally, the relevance of MRMR for cave design is limited to the range of applications that are included in the empirical dataset. Future caving projects are pushing the mining method to new extremes in terms of its application, thereby prompting industry to supplement empirical methods with modelling tools. Discrete Fracture Network (DFN) methods are one example of the modelling tools that are used for rock mass characterization. DFN uses statistical distributions to characterize each discontinuity set within a structural domain; variables included in the characterization include orientation, persistence and spatial location of discontinuities (Elmo et al., 2010). A major outcome of DFN modeling is the spatial distribution of fracture intensity, referred to as P32 and expressed in units of m2/m3 (fracture area/unit volume). DFN-based methods are considered to be particularly advantageous as they rely on quantifiable fracture data that is collected from field analysis of the rock mass. Since these fracture properties are preserved during the modeling process, the heterogeneity of the fracture system is better defined, resulting in an appropriate method to describing local scale problems (Rogers et al., 2010). The three main parameters that are used to validate DFN models are: 1) fracture frequency along a borehole or scanline, commonly referred to as P10; 2) orientation of fractures intersected by a borehole; and 3) comparison of trace length distributions in the case where considerable mapping data is available (Elmo et al., 2014). In the context of the proposed Cave-to-Mill research, DFN models also provide considerable value as a means to predict cave fragmentation; a critical input parameter for predicting flow geometry, draw-point availability and crushing/grinding requirements. Rogers et al. (2010) 16 developed a method to predict all three stages of cave fragmentation (in situ, primary and secondary) using DFN and 2D Finite Element/Discrete Element (FE/DE) methods coupled with field data and cave stress predictions. Synthetic Rock Mass (SRM) modeling is a method for characterising the strength of rock masses and, integrated within geomechanical models, can be used to predict cave fragmentation and caving rates. Key inputs include the previously described DFN model, as well as stress conditions, and the strength of joints and individual rock blocks (Chitombo, 2010 and Pierce, 2010). An advantage of Synthetic Rock Mass (SRM) methods is that the response of rock masses to caving conditions can be evaluated numerically at a scale that is significantly larger than the limit of geotechnical test equipment. 2.5 Cave Fragmentation In the context of the proposed Cave-to-Mill approach, fragmentation size is a critical input parameter for predicting flow geometry, draw-point availability, sorting potential and crushing/grinding requirements. The variable and relatively uncontrollable nature of cave fragmentation is a key distinguishing feature of block cave mines when compared to other mining methods. Three different stages of fragmentation are generally defined in the field of the study: 1) In situ fragmentation, which is represented by naturally occurring blocks within the rock mass 2) Primary fragmentation, representing the size reduction that occurs as blocks separate from the cave back at the onset of caving 3) Secondary fragmentation, which refers to the breakage of blocks as they fall onto caved muck and move through a draw column to drawpoints (Brown, 2003). Several commercial software packages have been developed or adapted to address the different stages of cave fragmentation, including: Discrete Fracture Network (DFN) based methods (Rogers et al., 2010) and BCF, a rules-based software program (Laubscher, 2000). The BCF program is based on analytical and empirical rules describing the processes and factors that play a role in cave fragmentation. Input data to the software includes jointing, fracture frequency, stresses in the cave face, and the strength of the rock mass, and intact rocks (Esterhuizen, 2005). 17 To quantify the fragmentation occurring in a draw column (secondary fragmentation), Gómez et al. (2017) conducted experiments on gravity flow under confinement and adopted modelling approaches used to describe comminution occurring in crushers and grinding mills. Their experiments, using a 700 mm diameter piston press with an outlet representing a drawbell, showed that the extraction of material affected the size reduction of finer material through abrasion. Dorador (2016) carried out 1-D compression tests to evaluate secondary fragmentation on broken ore within the plug flow zone in a draw column. Findings showed that hang-up potential was more sensitive to block strength than to block shape and block size distributions (primary fragmentation). Comparisons of measured and predicted fragmentation results at Cadia East Panel Cave 1 (CEPC1), located at the Cadia Valley Operations, showed that DFN-based predictions of secondary fragmentation were significantly coarser than that measured using image-based methods (Brunton et al., 2016). The same study also found that BCF predictions of secondary fragmentation corresponded relatively well to coarser fractions of measured size distributions, however, for finer fractions it was found that the content of fines was considerably underestimated. Similarly, the content of fines was underpredicted by Core2Frag software for Forsterite and Forsterite-Magnetite skarn rock at the Grasberg Deep Ore Zone (DOZ) operation (Srikant et al., 2004). Although the importance of predicting cave fragmentation is well addressed by mining literature, there is a general lack of published results from cave fragmentation studies. Image-based size analysis of caved material at drawpoints has been reported by Brown (2003) to be feasible, should there be sufficient lighting and minimum airborne dust. The quantification of fines is a known limitation of the method (Narendrula, 2004). Many operations, including the Grasberg mine in Indonesia, assess fragmentation based on human observations of certain size fractions (Srikant, 2006). Campbell and Thurley (2017) demonstrated a 3D laser scanning technique for size measurement of drawpoint muck at the Ernest Henry sub-level caving operation. Operational factors that serve as an indication of fragmentation include hang-up frequency, drawpoint availability and secondary blasting efforts (Tollenaar, 2008). Viera et al. (2014) propose a metric for secondary fragmentation, which is defined by the tonnage drawn 18 divided by the number of recorded secondary breakage events. Figure 2.4 shows a comparison of the tonnes drawn per secondary breakage event and the proportion of ore extracted from the draw column, directly related to height of draw (HOD), at the Esmerelda Sur Mine. From the graph it can be seen that less secondary breakage is required as draw columns are depleted due to the greater degree of secondary fragmentation taking place. Figure 2.4 Relation of tonnes drawn per secondary breakage event and percentage extracted (Viera et al., 2014) Comparing the outlined methods for assessing cave fragmentation, 3D laser scanning and digital image-based fragmentation programs are a solution to producing a size distribution curve that is relatively free from operator bias. The size distribution of blasted material is typically referred to as a critical parameter of Mine-to-Mill studies for open-pit operations (Burger et al., 2006 and Esen et al., 2007). Similarly, quantifying the size distribution of caved material is a key element of the proposed research, as it serves to verify the output of fragmentation models and is an important input to comminution models for downstream crushing and grinding equipment. Pre-conditioning Methods The ability of an ore body to cave is related to characteristics of the rock mass such as the nature of inherent discontinuities. Fracturing of the rock mass, using pre-conditioning methods (PC) is Tonnes per secondary breakage event Average percentage of primary column extraction, [%] 19 typically carried out before undercutting competent rock masses to ensure that caving will initiate and continue at an acceptable rate. Since the production rate for a certain footprint is directly linked to the caving rate, the success of pre-conditioning methods has a significant impact on project economics. Early records of attempts at pre-conditioning date back to 1968 where hydraulic fracturing was carried out at the Shabanie Mine (Laubscher, 2000). Chitombo (2010) states that pre-conditioning has proven to be a cost-effective means for cave inducement, improving caving rates and fragmentation, and will become an important part of cave mining in the future. The industry trend of applying intensive pre-conditioning is a marked changed from original applications of the block caving method where naturally fragmented ore was caved. The Cadia East Mine is the first major panel caving operation to use both hydraulic fracturing (HF) and blast pre-conditioning (BPC), referred to as intensive pre-conditioning (Catalan et al., 2017). Figure 2.5 shows the sequence used for implementing pre-conditioning at Cadia East. The rock mass has a Mine Rock Mass Rating (MRMR) of 57 to 62, which is relatively high when plotted on Laubscher’s (2000) graph of MRMR for historical cave mines. The blast design for ore located up to 130 meters above the undercut equates to a powder factor of 0.04 kg per tonne, which is approximately 1/10th of the powder factor that is usually used for production blasts in open-pit mines. Similar to Mine-to-Mill practices at open-pit mines, intensive pre-conditioning was recognized by Newcrest Mining Ltd, the owner of the Cadia East Mine, as having the potential to improve mill performance (Newcrest Panel Session, 2012). 20 Figure 2.5 Sequence to implement intensive pre-conditioning at Cadia East (Catalan et al., 2017) Catalan et al. (2017) found that hydraulic fracturing does not affect intact rock strength, while noticeable reductions in intact rock strength were attributed to blast pre-conditioning. Results of the study indicate that implementation of blast pre-conditioning has the potential to reduce crushing and grinding requirements. Hydraulic fracturing is considered by Catalan et al. (2017) to have a greater impact on enlarging existing discontinuities and creating macro scale discontinuities, thereby its influence is more noticeable on a rock mass scale. Pre-conditioning is considerably relevant to Cave-to-Mill, as it is an opportunity to coordinate confined blasting and/or hydraulic fracturing methods with cave productivity and mill operation. 21 2.6 Gravity Flow of Caved Material As a rock mass fragments due to stresses in the cave back, individual rock fragments fall onto a pile of caved material and follow a flow path that is influenced by inter alia draw strategy, size distribution of the caved muckpile and undercutting rate and direction. Understanding the flow of broken ore is critical to be able to link the grade and properties of the in situ rock mass, which are characterized using samples from exploration drilling, to the quality of ore reporting to drawpoints during production. A simplified representation of the flow regions of interest in a cross-section of the cave is presented in Figure 2.6. Interactive flow, as shown in the figure, is achieved when drawpoints are spaced close enough such that volumes of mobilized material overlap and an appropriate draw schedule is implemented. Isolated draw zones occur when drawpoint spacing is excessive or a drawpoint is drawn exclusively within a cluster of drawpoints. 22 Figure 2.6 Conceptual flow of material in a cross-section of the cave (Dorador, 2016) When drawpoints are drawn in isolation, rat holes occur with rapid introduction of dilution from waste overlying the ore body (Laubscher, 2000). Dilution entry is also known to occur due to horizontal movement of material. Paredes et al. (2014) state that lateral dilution entry occurs due to: 1) horizontal migration of waste from a laterally located waste boundary by rilling over the pile of caved ore, 2) lateral movement of waste due to inclination of the Isolated Movement Zone (IMZ) towards the waste boundary. Various models for explaining the gravity flow of fragmented ore have been developed for application to block cave mines. Pierce (2010) developed a model for flow of fragmented rock for use in REBOP (rapid emulator based on particle flow code), a code originally developed by Cundall et al. (2000) that emulates particle flow interactions using observations from other models (physical and numerical). The Cellular Automata model, a stochastic model, uses 23 probability distributions to estimate the movement of blocks within the fragmented ore column (Alfaro & Saavedra, 2004). A 3D version of the model has been adopted for use in the PCBC scheduling package (Villa & Farías, 2016). Recent developments in marker technology have significantly improved the quality of information available to track the flow of broken ore. Early efforts involved the use of tire markers such as old loader tyres and various steel tube designs (Talu et al., 2010). Markers were assigned unique IDs, placed in exploration shafts or left on undercut or extraction levels (in the case where a lift is planned below an existing cave). The use of Smart Markers, fitted with Radio-frequency identification (RFID) tags, was trialled at the Northparkes block cave mine and the Telfer sub-level cave mine. Scanners located in ore cross cuts, perimeter drives or orepasses, were used to link original Smart Marker locations to the drawpoints from which they were extracted (Whiteman 2010). Real-time monitoring of cave flow was presented by Whiteman et al. (2016), where Cave Tracker markers and detectors were placed in the in situ rock mass, as shown in Figure 2.7. The system was used to monitor the movement of markers in 3D at a time resolution of two days (Whiteman et al., 2016). Real-time monitoring of cave material allows production schedules to be refined daily so that favourable draw conditions are achieved. The technology has direct implications to cave management and the proposed Cave-to-Mill research. 24 Figure 2.7 Cave Tracker detectors and beacons within fragmented muck pile (Whiteman et al., 2016) 2.7 Sensor-based Sorting A growing number of mining operations are implementing sensor-based sorting technologies to pre-concentrate excavated material. Sensor technologies can be mounted on shovels, hoppers and conveyors, and sorted as bulk lots of material or on a particle-by-particle basis. Wyman (1966) presented a list of the necessary elements for successful mechanical sorting, they include: 1) a means of presenting the mineral pieces for examination; a sensing device for selecting the pieces to be removed; 3) an electronic system to act upon the information provided by the sensing device; and 4) a means of removing the selected pieces. Sensor types for mineral applications include prompt gamma neutron activation analysis (PGNAA), optical, X-Ray transmission (XRT), X-Ray Luminescence (XRL), near-infra red, X-25 ray fluorescence (XRF), electro-magnetic and radiometric sensors (Klein & Bamber, 2018 and Kobzev, 2014). At Newcrest’s Ridgeway Deeps mine, trials of a nuclear magnetic resonance (NMR) sensor for rapid (20 second) bulk sensing of chalcopyrite, the main copper-bearing mineral, showed that the technology has potential for use in a bulk sorting system through integration with a diverter (Coghill et al., 2018). Teck trialled shovel-mounted XRF sensors at their Highland Valley Copper operations to measure grade information in real time and instruct operators whether loads are to be sent to ore or waste stockpiles (Teck, 2018). Sensor technologies can be mounted on shovels, hoppers, conveyors and slurry systems (Van Haarlem, 2017), and sorted as bulk lots of material or on a particle-by-particle basis. At the Priargunsky Mine in Russia, both bulk and particle sorting systems are used for beneficiation of uranium ore. On a bulk scale, the grade of material inside rail cars or dump trucks is measured using XRF sensors; material that is above cut-off grade is fed to XRF-based particle sorters for further beneficiation while low grade material is directed to a heap leaching site (Kobzev, 2014). Ore sorting at the Khumani Iron ore operation is another example where both bulk and particle sorting technologies are combined. At Khumani, the grade of run-of-mine (ROM) material is measured with a PGNAA online analyser. Low grade ROM material is sent to waste stockpiles, intermediate grade material is sent to washing, screening and jig beneficiation circuits, and product grade material bypasses treatment and is sent directly to product stockpiles (Matthews & du Toit, 2011). An advantage of this arrangement is the reduction in capacity requirements of the beneficiation plant. Furthermore, the bulk sorting system reduces fluctuations in the grade of material being sent to the beneficiation plant, allowing the jig beneficiation circuit to be operated more effectively. PGNAA grade sensors are also used at the Sepon copper-gold operations in Laos. Kurth (2017) describes the technology used at both Sepon and Khumani as having a source of neutrons, Californium-252 located under the conveyor belt, and the emitted neutrons are absorbed by 26 elemental nuclei in the material being transported on the conveyor belt. Each excited nucleus generates a gamma ray having an energy level related to the element from which it has been emitted; detector arrays positioned above the conveyor belt measure the energy of received gamma rays (Kurth, 2017). The measured elemental content of transported material is output to the plant control system that operates diverter gates within the material handling system. The effectiveness of particle sorting systems relies on the use of grade sensing technologies that are suitable for the feed material (Rule et al., 2015 and Tong et al., 2015). In particle sorting applications, XRF sensors use the interaction of x-rays with rock surface material to determine its elemental composition. Sensor data is used in algorithms to infer the grade of the rock. A sorting decision is made based on the relation between the inferred grade and the setpoint cut-off grade. Sorting is typically carried out through use of blasts of compressed air, or mechanical paddles. XRF-based sorters have been used in precious metal, base metal, ferrous metal, industrial and rare earth ore applications (Tong et al., 2015). The capacity constraints of XRF-based particle sorters support the use of bulk sorting to reduce the quantity of material requiring beneficiation and thereby decrease the number of sorter modules operating in parallel. Bamber et al. (2008) describe four key components of evaluating the feasibility of sensor-based sorting: ore heterogeneity, sensor response evaluation, sorting analysis and feasibility, as presented in Figure 2.8. 27 Figure 2.8 Sortability of ores (Klein & Bamber, 2018) The natural grade heterogeneity of the material being assessed for sorting controls the limit of sorting potential (Duncan, 2016, Kobzev 2018 and Mazhary & Klein, 2015). Gy (2004) defines heterogeneity as occurring in two main forms: Constitutional Heterogeneity (CH): The variation in the content of a certain component, for example copper or elements related to gangue minerals within individual rock fragments which make up an ore domain. Blending does not affect CH; however, CH increases with comminution. Distributional Heterogeneity (DH): The variation in the content of a certain component, such as copper, within individual groups of ore (which report to draw-points or surface in a caving application). The sum of all of the groups makes up a lot. The DH parameter is considered to represent the potential to sort ore types at a bulk scale. DH can be reduced by blending ores but is not affected by comminution stages. CH has been used as an indicator of particle sortability by Robben (2014), and Mazhary and Klein (2015), while DH shows potential for assessment of bulk sorting systems. 28 As a cave matures, it is expected that the large number of mixing events that have occurred within caved ore results in a lower variation in grade on a bulk scale, represented by a lower DH value, thereby reducing the potential for bulk sorting. The nature of cave operations makes them ideal for incorporation of sensor-based systems to reject rock and allow the operation to become more dynamic. An integrated mine and mill approach will be necessary to coordinate mine schedules and beneficiation processes so that a satisfactory level of throughput and recovery can be achieved. 2.8 Cave-to-Mill Concept Addressing the objective of improving block cave productivity, reviewed literature formed the basis of a proposed systems engineering approach for cave mine operations, termed Cave-to-Mill. Both mine and mill models are set up at the beginning of a project and continuously refined as the project progresses. Figure 2.9 shows how key aspects of a caving operation are linked so that a coordinated effort can be made to maximize Net Present Value (NPV). The presented Cave-to-Mill strategy focuses on its application to copper porphyry deposits, as the majority of present and future caving operations are based on excavating copper ores. Focus areas and enabling technologies, highlighted in blue and green respectively, represent significant opportunities to improve the productivity of block caving operations by implementing these technologies in Cave-to-Mill. The major objective of adjusting design and operational parameters within Cave-to-Mill is to maximize project value. A central component is the ore block model, which is continuously refined during project development and operation through input of geotechnical, geological and metallurgical information. At the exploration stage, initial access to an orebody is generally provided through exploratory drilling, allowing block models and design parameters to be established through analysis of core samples. As projects progress towards development, excavations and boreholes provide access to additional sample and data, which can be used to repopulate block models. Following plant commissioning, logs of mine and mill performance, 29 such as fragment size measurements and specific energy consumption of mill processes, can serve to calibrate predictive models and to refine block models. The cave-to-mill approach is proposed for trial at a caving operation to quantify the benefits associated with its implementation. 30 Figure 2.9 Cave-to-Mill Concept 31 2.9 Summary of Literature Review Review of literature from seven areas of research was carried out with respect to the objective of developing Cave-to-Mill. A summary of key findings is included in this section. Block and panel caving methods will be increasingly used in the copper-gold industry. Future caving projects will push the limits of applicability of the mining method in terms of increasing cave heights, caving of more heterogeneous ore bodies, and caving of deeper and more competent rock masses. Significant benefits in the productivity of open-pit mines have been attributed to implementation of mine-to-mill strategies, where blast fragmentation and/or geometallurgical information were used for coordination of mining and processing activities. Many geometallurgical studies have shown strong correlations between geotechnical test data, which are in abundance in cave mines, and crushing/grinding indices. Cave planning and scheduling software can be used together with geological block models, which include revenue factors such as mining cost and mill recovery to determine the associated value of a caving lift for an ore body. Conceptually sorting and geometallurgical models can be included in the block model as a Cave-to-Mill approach. Existing caving operations show that flexibility in production schedules exist, providing the possibility to control properties of the plant feed that affect mill performance. Rock mass characterization methods are critical to block cave design and planning. Empirical and numerical approaches are required to confirm the suitability of cave mining for a proposed footprint and estimate caveability, subsidence angles, failure zone, fragmentation, undercutting sequences and support design. Cave fragmentation size is a critical parameter that affects flow geometry, draw-point availability, sorting potential and crushing/grinding requirements. Application of caving methods 32 to competent rock masses necessitates the use of pre-conditioning methods, which present an opportunity to influence the size of feed to sorters and milling equipment. Systems for tracking of fragmented ore in real-time have recently become available. In addition to reducing uncertainty in muckpile shape (above the extraction level), this development provides opportunity to track grade and geometallurgical properties of material through the cave, material handling system and up to the mill. For caving project evaluation, cave planning is still reliant on gravity flow models. Grade heterogeneity is a key material characteristic that governs the opportunity to sort material. The capacity constraints of particle sorters supports the use of bulk sorting to reduce the quantity of material requiring beneficiation and thereby decrease the number of sorter modules operating in parallel. Literature was reviewed with a focus on finding a means to improve block cave productivity. An outcome of the review was the development of a Cave-to-Mill concept that was put forward for trial at a block caving operation. 33 Chapter 3: Site Description and Methodology 3.1 Introduction This chapter describes the methodology used to develop a Cave-to-Mill approach which targets improvements in cave mine productivity. Access to samples and data from an operating block caving operation was identified as a key requirement at the onset of the research. The New Afton block cave mine, located in British Columbia, Canada, was engaged for the research project and served as the main source of operational data and sample. Site visits were carried out by the research group to measure drawpoint muckpiles with image-based systems, collect samples and carry out mill surveys. Further detail of the New Afton mine and methodology used is provided in the following sections of this chapter. 3.2 Site Description: New Afton Mine The New Afton Mine is a 17,500 tonne per day copper-gold operation. The mine officially commenced production in July, 2012. The B1 and B2 caves, which were in production at the time of this study, are located 615 metres below surface and their combined mining footprint is approximately 800 metres in length and 150 metres in width. The majority of the deposit occurs within crystalline and polylithic fragmental volcanics belonging to the Triassic Nicola Group and lesser monolithic intrusive breccias consolidated into a single-host package, which is informally referred to as BXF (Davies, 2015). A plan view of geological and structural features at the extraction level is shown in Figure 3.1. The footwall and hanging-wall faults bound the deposit to the North and South respectively. Numerous faults, which in many cases are carbonate-healed, transect the deposit. The concentration of faults is significantly greater in the in situ rock mass of the B1 cave. Hypogene zones, where copper primarily exists in the form of copper sulphide minerals, and supergene zones, where copper primarily exists in the form of native copper, both occur within the BXF unit. A large picrite unit lies at the southwestern edge of the deposit, adjacent to the hanging-wall. The picrite is distinctly less competent than other rock types within the cave footprint. A 34 monzonite stock intrudes the south-central part of the deposit and represents the most competent rock type. In the eastern half of the deposit, the BXF is intruded by the diorite phase of the iron mask complex. The supergene is also hosted within the diorite intrusive. The rock mass rating (RMR) of the economically viable portion of the deposit is in the order of 35 – 55 RMR(1976). Due to the presence of several fault structures, a lower mining rock mass rating (MRMR) of 40 is associated with the B1 cave while an MRMR of 45 has been reported for the B2 cave (Bergen et al., 2015). Figure 3.1 Geological and fault structures at the extraction level A simplified flowsheet for the New Afton mine is shown in Figure 3.2. The mill feed consists of material that has been caved and mucked from drawpoints, dropped into orepasses, which are 25 m in length and 2.4 m in diameter, and fed to a gyratory crusher located at the haulage level. To generate finer mill feed, the gyratory crusher is operated at its smallest close-side setting of 90 mm. Crusher product is conveyed to a surface stockpile and fed by two interdependently controlled reclaim feeders to a SAG mill, which operates in closed circuit with a screen and a pebble crusher for treating screen oversize. Material segregation within the bed of material on the underground-to-surface conveyor results in coarser material being thrown further past the head pulley towards the northern reclaim feeder, while finer material discharges closer to the southern reclaim feeder. An expert control system is used to adjust the throughput rate of each of the two feeders, effectively blending coarser and finer material within the stockpile, to maximize 35 mill throughput performance. The feeder that handles coarser material is slowed down when the SAG mill circuit is throughput constrained and it is sped up when SAG mill throughput capacity is available; the feeder handling finer material is operated in the opposite manner. SAG mill screen undersize is fed to a ball mill circuit. In 2015, a VTM3000 VertiMill was installed in a tertiary grinding role as part of a mill expansion. The target P80 size of the Vertimill cyclone overflow, which is the feed to rougher flotation cells, is approximately 145 microns. Figure 3.2 Simplified New Afton flowsheet 3.3 Fragmentation Measurements Cave fragmentation is a key parameter within cave-to-mill that affects the extraction level layout (Laubscher, 2000), mixing within the cave, hang-up frequency (Tollenaar, 2008), mill performance (Condori et al., 2012) and the quantity of material available for particle sorting. To measure the size distribution of drawpoint muck at the New Afton B1 and B2 caves, image- 36 based methods, as used by Brunton et al. (2016) at the Cadia Mine, were used. Prior to taking photographs, scale reference markers were placed on muckpiles and portable lighting equipment was arranged to minimize shadow effects. The majority of images taken were processed with WipFragTM software. Edge delineation was carried out manually within the software for particles that were coarser than approximately 80 mm, amounting to approximately one hour of manual correction per image. Analysis of finer sizer fractions significantly extended the amount of time required for image delineation. An example of a muckpile photo and edge-delineation is shown in Figure 3.3. Figure 3.3 Fragmentation analysis of drawpoint F10N (taken on November 4th, 2013) Overall, 79 photos were taken such that a range of lithologies, alterations and height of draws (HOD) were captured. The quantity of photos that could be taken and analyzed was limited by the amount of time required for manual correction of particle delineations generated by WipFragTM. Comparisons of fragmentation size and hang-up frequency, outlined in Section 4.6, showed similar trends with height of draw, supporting the approach used for image-based size analysis. PortaMetricsTM, a portable fragmentation analysis tablet, which includes three high-resolution cameras, was used to analyse ten muckpiles. Figure 3.4 shows the model that was used for the study. In a previous study, good agreement between the outputs of PortaMetricsTM and WipFragTM was found when manual delineation was used with the software package (Liu et al., 2015). Particle delineation was found to be improved with the PortaMetricsTM tablet resulting in less time being spent on manual correction. Additionally, automatic scaling functions within the tablet meant that scale reference markers did not need to be placed on muckpiles. 37 Figure 3.4 PortaMetricsTM Tablet The quantification of fines is a known limitation of image-based size measurement (Narendrula, 2004). To characterise the fines content of drawpoint muck, the -25 mm fractions from each drawpoint were sieved using a sieve shaker to a final screen size of 0.5 mm. Sieving methods were also used for three primary crusher product samples. 3.4 Drawpoint Sampling and Test Regimen Over the course of the research, twenty-five drawpoints were sampled after photos had been taken for size distribution analysis using WipFrag software. Samples weighing approximately 50 kg for each drawpoint were shovelled from along the width of the base and mid-section of each drawpoint muckpile. Care was taken to collect samples from the full width of the muckpile. The sample quantities collected from each drawpoint and fragment size were limited by the practical constraints associated with shovelling samples into pails and transporting them with personnel/service vehicles. The top size of collected samples was approximately 150 mm. Ross (2012) mentions that one of the sources of error associated with drawpoint sampling is the tendency for fines to flow around coarser fragments that are hang-up within a drawbell. Effectively, the coarse fragments behave like a sieve, resulting in size classification within the draw column. When sampling at New Afton, drawpoints that contained noticeably hung fragments were avoided to reduce the effect of size segregation on sampling. 38 Material from eleven drawpoints was used for a comprehensive test program, shown in Figure 3.5. Drawpoints were chosen such that each of the critical lithology and alteration types, in terms of impact on mine and mill operations, were collected. Surface samples were also taken during mill surveys so that mill models could be calibrated based on feed size and hardness, and mill performance. The results of two mill surveys were used to validate the mill models. The program included geotechnical, metallurgical and sensor-based sorting tests to characterize the ores for cave-to-mill modelling.39 Figure 3.5 Test regimen for the core of the test program 40 3.5 Mill Sampling and Surveying To estimate the change in mill performance due to variations in cave fragmentation size and material hardness, mill models were calibrated with New Afton mill survey data. The outputs of calibrated mill models could then be validated and used to characterize each drawpoint sample in terms of mill performance for use in the Cave-to-Mill approach. Sieving of collected mill feed samples also provided an indication of the quantity of fines in caved material, which could be linked to the cave fragmentation component of the study. Two mill surveys were carried out in 2015 and 2016 for the cave-to-mill research program. Results from a later survey, carried out by Cebeci et al. (2017), were also made available for the study. Figure 3.6 shows the sampling points for the mill surveys that were used for this study. Figure 3.6 Simplified New Afton flowsheet and sampling points For the 2015 survey, 0.7 tonnes of mill feed was sampled from the SAG mill feed conveyor, which is located below the mill feed stockpile. To avoid the effect of size segregation occurring 201 201 201 (Cebeci et al, 201 )Sampling year Application to models SymbolModel calibrationModel validationModel validation41 in the mill feed stockpile, the second mill survey (2016) involved sampling of 2.2 tonnes of rock that was diverted from the underground-to-surface conveyor, shown in Figure 3.7. Cebeci et al. (2017) collected 1.1 tonnes of mill feed from the mill feed conveyor and sieved the sample as part of a New Afton mill survey. Guidelines for surveying and sampling grinding circuits recommend that mill feed sample quantities be in the range of 500 to 1,500 kg (Global Mining Standards and Guidelines Group, 2016), which was met or exceeded by all three surveys. Data from the mill survey of 2016 was nominated for mill calibration, due to the large quantity of mill feed collected and number of sampling points sampled from within the grinding circuit. Figure 3.7 Diverted mill feed sample (2.2 tonnes) from the underground to surface conveyor. Sample collection for 2016 mill survey For the 2016 survey, which was used for model calibration, the stockpile live capacity was estimated as being two-hours at the time of sampling. Approximately two hours after mill feed collection, samples from within the grinding circuit were taken at 15-minute intervals over the course of one hour; an exception was the pebble crusher feed, for which approximately 200 kg was taken just after the survey. Procedures for carrying out the survey were in line with guidelines for surveying and sampling grinding circuits (Global Mining Standards and Collected sample (2.2 tonnes) 42 Guidelines Group, 2016). Collected samples were weighed, dried/filtered, sieved and used for the test program shown in Figure 3.5. 3.6 Primary Crusher and Mill Modelling Calibrated crusher and mill models were used to estimate mill productivity for samples collected from drawpoints. The boundary limits for comminution modelling were defined as starting at the feed to the primary crusher and ending at the cyclone overflow of the Vertimill operating in a tertiary grinding role. A significant number of mine-to-mill studies have successfully made use of JKSimMet, a software package for simulation of comminution circuits, including that of Burger et al. (2006) and Kanchibotla et al. (2015). For the presented cave-to-mill research, JKSimMet was used to model the primary and pebble crushers, and the SAG mill within the New Afton circuit. The Anderson-Arachie-Whiten crusher model and the variable rates SAG mill model, included in JKSimMet, are based on population balance principles and separate equations are used for predicting power requirements (Foggiatto, 2017). Crusher modelling The primary inputs for crusher modelling included measured parameters: • Feed size from image analyses and sieving of samples • Appearance functions from DropWeight testing of samples, which define the product size distribution of a particle following crushing and parameters that were fitted based on results from mill surveys, and crusher closed and open-side settings: • Breakage parameter t10, which represents the proportion of product passing 1/10th of the original feed size • Classification parameters K1-3, which define the size of material that is selected for breakage 43 SAG mill modelling To apply the JKSimMet Variable Rates model to the New Afton fully-ported SAG mill, the guidelines of Bailey et al. (2009) were followed. Stark et al. (2008) found that experimental error in the DropWeight test method resulted in a coefficient of variation, which is the standard deviation divided by the mean, of 3.8% for JKSimMet throughput predictions in a SAG mill circuit. Two validation surveys were included in the test program to verify the accuracy of the calibrated model. Similar to the approach used for crusher modelling, measured ore parameters from sieving, DropWeight and abrasion testing were used along with SAG mill operational parameters as inputs to the variable rates SAG mill model. Ball mill modelling For the ball mill circuit, a Bond efficiency method, as published by the Global Mining Standards and Guidelines Group (2015) was used to calculate the Bond Standard Circuit Energy Factor, which is the ratio of the operating work index to the Bond ball mill work index of sampled mill feed. The efficiency factor could then be used for different material types. 𝐵𝑎𝑙𝑙 𝑀𝑖𝑙𝑙 𝑃𝑜𝑤𝑒𝑟 (𝑘𝑊)= 𝐵𝑜𝑛𝑑 𝐸𝑓𝑓 𝐹𝑎𝑐𝑡𝑜𝑟 𝑥 10 𝑥 𝑡𝑝ℎ 𝑥 𝐵𝐵𝑊𝐼 (𝑘𝑊ℎ𝑡) 𝑥 (1√𝑃80−1√𝐹80) (3.1) Where Ball mill power includes drivetrain inefficiencies, F80 and P80 are the 80% passing sizes of SAG mill screen undersize ball mill cyclone overflow respectively in microns. BBWI is the Bond ball mill work index determined for sampled material and tph is the rate of fresh feed to the circuit. Vertimill modelling To model the Vertimill, operating in a tertiary role in the New Afton grinding circuit, operational data and Bond ball mill work indices for sampled ore were reviewed. Results showed that the Vertimill main motor power draw is approximately 62% of the power draw of a ball mill that is carrying out the same comminution duty, shown in Figure A1 of the appendix. Efficiency factors from Rowland (1982) were applied to account for the low reduction ratio (< 2) associated with the tertiary grinding duty and the use of modern larger diameter mills (> 3.81 m diameter). Ball 44 mill drivetrain efficiencies of 97% for the gear reducer and 96% for a wound rotor induction motor were assumed based on Doll and Barratt (2010). The following is the resulting equation for relating Vertimill power draw, as measured at the DCS, feed F80 and product P80 sizes: 𝑉𝑒𝑟𝑡𝑖𝑚𝑖𝑙𝑙 𝑝𝑜𝑤𝑒𝑟 (𝑘𝑊) = 0.62 x Ball mill power (kW) (3.2) = 0.62 𝑥 10 𝑥 𝑡𝑝ℎ 𝑥 𝐸𝑓𝑓3 𝑥 𝐸𝑓𝑓7 𝑥 𝐵𝐵𝑊𝐼 (𝑘𝑊ℎ𝑡) 𝑥 (1√𝑃80−1√𝐹80) 𝑥 10.96 𝑥 0.97 (3.3) Where F80 and P80 are the 80% passing sizes of ball mill cyclone overflow and Vertimill cyclone overflow respectively. Eff3, equal to 0.914, corrects calculated power draw to account for improved energy performance of modern larger diameter mills ( > 3.81 m). Eff. 7 is applied to correct power draw for low-reduction ratios ( < 6 ), determined by: 𝐸𝑓𝑓7 =𝟐 (𝑹𝑹−𝟏.𝟑𝟓)+𝟎.𝟐𝟔𝟐(𝑹𝒓−𝟏.𝟑𝟓) (Rowland, 1982) (3.4) Where Rr is the reduction ratio, F80 divided by P80. For each feed size and material type, the calibrated comminution models were able to be used to determine associated mill throughput and energy performance. Thereby, changes in the quality of caved ore could be related to fluctuations in mill productivity, an aim of cave-to-mill. 3.7 Rock Type Specification and Characterization At the beginning of the test program, collected drawpoint samples were sieved into three size fractions (+50, – 25, and -50, +25 mm). Rocks coarser than 25 mm were washed, individually marked with a unique ID number and subjected to the test program with the objective of compiling a large database of information that could be referenced for the cave-to-mill program. An example of rock classification is presented in Figure 3.8. Following washing and inspection of the collected samples, a rock classification system was developed to group rocks according to their lithology or type of alteration. For rocks between 12.5 and 25 mm in size, the mass 45 proportions of each rock type were measured. Table 3.1 shows the criteria established to define each rock type. Table 3.1 Guideline for macro-description of rocks Rock Group # Lithology / Alteration Code Characteristics Ore / Dilution 1 Hypogene Biotite dominant. K-Spar*< 20% Ore (high grade) 2 Picrite Talc (soapy texture). Serpentinised. Dilution 3 Hypogene 0 > K-Spar > 20% Ore (medium grade) KK K-Spar > 0% Ore / Waste (monzonite) Fault Rock Round. Very soft. Alteration obscured Dilution Carbonates White or clear carbonate minerals Dilution 8 Supergene Visible iron oxides (red) Ore *K-Spar: Observable potassium feldspar Figure 3.8 Collected +50 mm samples from drawpoint D11N 46 The specific gravity of each (+25 mm) rock was recorded using the buoyancy method and a rare earth magnet was used to identify magnetic rocks. 3.8 Sensor-based Sorting The lack of selectivity and potential for dilution entry associated with the block cave mining methods results in both ore and waste being caved and transported through material handling systems to the surface for processing. Sensor-based sorting systems provide an opportunity to automate the discrimination between ore grades and rock types, providing an enhanced level of selectivity for ore control and thereby improving mine productivity, an objective of cave-to-mill. A combination of bulk and particle sorting systems was considered for trial at New Afton based on a similar approach used at the Priargunsky Mine in Russia. For the bulk sensing application, PGNAA technology was trialled as a solution to measuring the grade of material being conveyed from underground to surface and diverting bulk lots according to copper grade. An advantage of PGNAA technology over other sensor technologies is its ability to measure the grade of material within the bed of rock on a conveyor, while a disadvantage is the long sensing time (typically 0.5 to 2 minutes) required to record a suitably accurate grade measurement. For the particle sorting application, the amenability of XRF technology to New Afton ore was tested. The XRF test method relies on surface grade measurements for estimations of whole rock grade. Potential alternatives to XRF technology includes XRT sensors, however Mazhary (2017) found that the presence of iron minerals decreases the grade-sensing efficiency of XRT sorters when the metal of interest is of a similar atomic density. At New Afton, iron oxides in the form of hematite and magnetite occur in abundance and do not have a reliable association with either copper grade or waste. An XRF sorting study by Tong et al. (2015) on copper porphyry material from the Spence Mine in Chile indicated that XRF surface measurements could be used for effective pre-concentration. 47 Bulk Sorting As a separate test component, PGNAA bulk sensing test work was carried out using sample from drawpoints, mill feed and waste stockpiles. Each sample weighed approximately 220 kg. Overall, a wide range of copper grades and rock types were prepared. The grade of each bulk sample was determined by crushing and screening to 100% passing 9.5 mm, homogenizing, sub-sampling, and assaying sub-samples with inductively coupled plasma (ICP) spectrometry and fire assays. The nine samples were sent to Scantech testing facilities in South Australia for testing with a Geoscan-M unit, shown in Figure 3.9, which was fitted with four detectors. Assay results were provided for six of the nine samples, while the remaining three were treated as blind samples. During testing, samples were placed on a section of New Afton conveyor belt provided by the mine to account for any influence of belt material on sensor response. Since the accuracy of the PGNAA sensor system is not affected by particle size (Kurth, 2017), results for the -9.5 mm test samples could be used to evaluate application of the technology with New Afton ROM material. A sub-sampling and crushing procedure was established referencing François-Bongarçon and Gy’s (2002) sampling method. The maximum error associated with sub-sampling was recorded for both copper and gold using the following formula from François-Bongarçon and Gy (2002): 𝑆𝐹𝑆𝐸2 = (1𝑀𝑠−1𝑀𝐿) 𝑓𝑔𝑐𝑙𝑑3 (3.5) where SFSE2 is the sampling relative variance, Ms and Ml are masses of the sample and lot respectively. Constants f and g, which are usually set to 0.5 and 0.25 respectively, are ‘convenience’ constants, c is the mineralogical factor, d is the nominal size of the rock fragments (95% passing size), and l is the liberation factor. From grind-recovery performance in the regrind circuit of the process plant, the liberation size, used to determine the liberation factor in equation 3.5, was assumed to be 35 microns for copper-sulphides and gold. Quantitative Evaluation of Minerals by SCANning (QEMSCAN) analyses indicated a copper-sulphide liberation size of approximately 20 microns. A gold deportment 48 study indicated that the majority of liberated gold is in the size range of 10 to 75 microns (AMTEL, 2006). An analysis of native copper deportment showed that approximately 50% of native copper was in the sub-10 microns size fraction while 40% was coarser than 200 microns (AMTEL, 2006). A conservative liberation factor of 500 microns was assumed for supergene ore. The sub-sampling procedure provided relative standard deviations of less than 1% for copper and less than 25% for gold. Figure 3.9 Sample being loaded into Geoscan-M with conveyor belt cut-off (Balzan, 2016) Each sample was measured for a period of at least three hours. During the three-hour measurement time, individual instantaneous analysis periods were varied throughout the test work to achieve a number of results for different measurement periods. The expected accuracy of a Geoscan-M unit when installed on a conveyor could then be estimated for different measuring times (which are directly related to the quantity of conveyed material). Sensor-based Particle Sorting An Olympus model 500316 XRF unit, mounted in a tripod stand, was used to measure the elemental content of surface rocks. Samples were separate from those used for PGNAA testing and consisted of material taken from drawpoints and mill feed stockpiles. 49 XRF measurements were carried out on the surfaces of 1,226 rocks in the (-150, +25 mm) size range using a measuring time of 10 seconds. For each individual reading, the pXRF unit measured the elemental content of material located on the surface of the rock within a circle having a diameter of 1 cm. Four readings were taken per rock. From the master set of 1,226 rocks, 291 rocks were selected based on rock type and subjected to crushing and pulverizing. A representative sub-sample of the finely pulverized material was measured with the same pXRF unit to determine the grade of each rock. To calibrate the pXRF unit, ICP measurements and pXRF measurements were carried out on 20 pulverized New Afton rocks. A coefficient of determination, R-squared, of 0.999 resulted from comparison of ICP and pXRF results for copper. 3.9 Cave and Mill Comminution Testing Comminution processes occurring within the cave and the mill were relevant to the research program, and thus accounted for in the test regimen. Eleven drawpoint samples and one mill feed sample were characterized with the whole comminution program which included Point load, DropWeight (JK Tech, 2016), low-energy abrasion tests (JK Tech, 2016) and Bond ball mill work index testing. Due to the importance of proper geotechnical characterization of rock masses being considered for caving, an abundance of point load strength data is typically available for caving projects. In the case that point load data correlates to milling parameters, such as DropWeight Indices, a database of point load variation within the orebody is very relevant to the proposed Cave-to-Mill approach. Published work by Burger et al. (2006) showed that useful correlations between the Point Load Index Is(50) and DropWeight Indices can be found for some deposits. Advantages of the Point Load Index (PLI) test are the relatively low test costs and short turn-around time in comparison to DropWeight type breakage tests. The test program provided opportunity to investigate for correlations between point load strength, muckpile size distributions and metallurgical performance. 50 Point Load Tests Following sensor-based testing, point load tests were carried out according to ASTM standards for lump rock (ASTM, 2013). Tested rock samples were generally in the size range of -100, +50 mm. Overall, samples from twenty-four drawponts were tested (of which eleven were used for the full metallurgical program). Approximately 30 rocks were tested per drawpoint sample. DropWeight Testing Following point load testing, sub-samples were sieved at 63 mm, which is the coarsest size of material used in a standard JK Drop Weight test (JKTech Pty Ltd, 2016). Results were used for crusher and SAG mill modelling within JKSimMet. For drawpoint samples, an abbreviated form of the DropWeight test was used to conserve sample for the sensor-based sorting scope of work. The abbreviated dropweight test was based on the SMC method, where five energy levels are tested using one feed particle size range (SMC Testing, 2018). Testing with mill feed sample was used to determine the relationship between feed size, energy and product size, t10. Figure 3.10 JK DropWeight test unit at the NBK Institute of Mining Engineering 51 For each drawpoint sample, 150 particles were selected in the size range of -31.5, +26.5 mm and their individual weights were recorded. Based on the weight distribution of particles, rocks were grouped into heavy, medium and light groupings. The standard JKDropWeight test regimen involves random selection of a group of similarly sized rocks for breakage at an equivalent impact energy. To generate product for one energy-size combination, the height of the drop weight is selected so that a target specific energy input kWh/t is achieved based on the mean particle mass (typically 10 or more particles are used per energy-size combination). Due to variations in particle mass, this method results in particles experiencing specific energy inputs that are above and below the setpoint. To reduce the variation in specific energy input to test sample, particles were grouped into high, medium and light masses such that the overall error in specific energy input would be minimized. An example is shown in Table 3.2. A detailed description of the abbreviated DropWeight test method is described in Appendix B. Table 3.2 Specific energy and size summary for DropWeight testing Target specific energy level, kWh/t Size range, mm Particle weight grouping 2.5 -31.5, + 26.5 Heavy Medium Light 1.8 -31.5, + 26.5 Heavy Medium Light 1 -31.5, + 26.5 Heavy Medium Light 0.5 -31.5, + 26.5 Heavy Medium Light 0.25 -31.5, + 26.5 Heavy Medium Light 52 Abrasion Tests Standard ta abrasion tests were carried out using a JK abrasion mill having approximate dimensions of 1 ft diameter and width. Tests were carried out according to the test procedure (JKTech Pty Ltd, 2016), which was as follows: Three kilograms of -53, + 37.5 mm sample was processed in the mill for ten minutes at a speed which is equivalent to 70% of critical speed for the internal mill dimensions. Products were sieved to find the percentage passing 1/10th of the feed size. The abrasion parameter ta, was reported as the product t10 divided by 10. An energy meter connected to the unit indicated an approximate specific energy input of 1 kWh/t per test run. Results from abrasion testing were used within JKSimMet for SAG mill modelling. Bond Ball Mill Work Index Tests Ball mill and Vertimill models relied on outcomes from the Bond ball mill test method. Drawpoint and mill feed sub-samples were crushed to -3.35 mm and subjected to the standard Bond ball mill test method (Bond, 1962). Care was taken to ensure that mass-weighted proportions from each size fraction were taken for Bond ball mill testing. ALS Metallurgy in Kamloops, BC carried out twelve Bond ball mill work index tests as an in-kind contribution to the study. Approximately 7 kg of sample was required per test. Separate from the presented cave-to-mill research, New Afton mine had regularly conducted Bond ball mill tests using a closing screen size of 106 microns. The same closing screen size was nominated for the test program to be able to compare results to historical grindability data. 3.10 Block Cave Planning and Scheduling The PCBCTM module within GEMS GEOVIA software, version 6.8.1 package (Dassault Systèmes, 2018) was used for cave planning and production scheduling. Block models provided by New Afton were used in combination with outcomes from fragmentation and sensor-based sorting studies. For each block of ore, Net Smelter Return (NSR) calculations were carried out 53 using the block manipulation editor which is available within GEMS GEOVIA. Scripts were written in Cypress Enable Language (CEL), as accepted by the software. For the case where sorting systems were investigated, block attributes were created to represent: • Proportion of each block accepted to high grade, intermediate grade or waste stockpiles • Recovery of contained copper and gold within each block to its designated stockpile • Resulting grade of material sourced from a block following sorting • Flotation recovery of copper and gold as a function of block grade following sorting Addition of the above attributes allowed sensor-based sorting and flotation recovery parameters to be included in the cave planning and production scheduling solution. A mass balance was carried out at the end of each production scheduling run. Further information on the steps taken is included in Chapter 8. 54 Chapter 4: Predicting Cave Fragmentation and Hang-Up Frequency 4.1 Introduction The size of caved rock fragments has significant implications to the profitability of caving operations. Predictions of secondary cave fragmentation, representing the size of material reporting to drawpoints, are a key input to extraction level design and cave evaluation at the project development stage. Furthermore, knowledge of the size of mill feed is critical for accurate equipment sizing. Even though a significant amount of effort may be directed towards estimating cave fragmentation at the beginning of a block cave project, these models tend not to be reliably calibrated and referenced for ongoing mine planning and scheduling activities. Advances in the prediction of cave fragmentation stand to significantly reduce the risk associated with caving projects. With the objectives of Cave-to-Mill in mind, drawpoint samples, measurements and operational data from two caves at the New Afton operation were analysed to gain a better understanding of the fragmentation associated with the footprint. Results of the analysis could then be applied at the project design and evaluation stages for future caves being considered within the New Afton deposit. The differences in fragmentation size at the New Afton Mine can be attributed to variation in the geological composition of caved rock. A method to identify rock types relevant to fragmentation size, using data available in the ore block model, was established. Consistent with other reported studies on the topic, finer drawpoint muckpile size distributions were measured for material that had travelled a greater height through the ore column, referred to as Height of Draw (HOD). Block Cave Fragmentation (BCF) software (Esterhuizen, 2005), version 3.05, predictions were found to be in-line with measurements for aperture sizes coarser than 0.5 m. However, significant deviations were apparent for the cumulative percentage passing of sizes finer than 0.3 m. 55 A comprehensive fragmentation study involving two image-based measurement methods, WipFragTM (WipWare Inc., 2017) and PortaMetricsTM (Motion Metrics International Corp., 2017), as well as sieving of muck was carried out as a component of the cave-to-mill research program being conducted at New Afton. A cave-to-mill project team member, Mukhtsolmon Munkhchuluun, developed a Discrete Fracture Network (DFN) model for the in situ rock mass of the B2 (West) Cave. Correlations between fracture intensity and hang-up frequency were apparent. A distinguishing feature of the fragmentation study is the abundance of sample characterization data available for drawpoint muck samples that had been measured using image-based techniques, inter alia, X-ray Fluorescence (XRF) measurement for sensor-based sorting evaluation, comminution test work for mill performance prediction and point load testing. 4.2 Fragmentation Measurements Measurements of the size of caved rock fragments located at drawpoints of the B1 and B2 caves were taken using an image-based approach. Prior to taking photographs, scale reference markers were placed on muckpiles and portable lighting equipment was arranged to minimize shadow effects. Overall, 79 photos were taken such that a range of lithologies, alterations and height of draws (HOD) were captured. Table 4.1 shows the number of images that were analysed for different ranges of HOD. 56 Table 4.1 Summary of draw point muck images B2 Cave (West Cave) B1 Cave (East Cave) Height of Draw Range Average Height of Draw Number of Images Height of Draw Range Average Height of Draw Number of Images [m] [m] [#] [m] [m] [#] 9 - 14 12 8 23 - 35 29 6 14 - 40 24 8 35 - 62 52 7 40 - 62 52 9 62 - 90 72 6 62 - 98 74 8 90 - 140 112 7 98 - 120 107 5 140 - 180 164 4 120 - 190 151 12 190 - 289 239 9 In order to determine the content of fines (-10 mm material) within caved muck, two bulk samples of gyratory crusher product were collected and sieved. In 2015, 0.7 tonnes were sampled from a mill feed conveyor, which is located below the mill feed stockpile. To avoid the effect of size segregation occurring in the mill feed stockpile, in 2016 a second sample of 2.2 tonnes of rock was diverted from the underground-to-surface conveyor, collected and sieved. Huang et al. (2017) collected 1.1 tonnes of mill feed from the mill feed conveyor and sieved the sample as part of a New Afton mill survey. Figure 4.1 shows size distributions for measured secondary fragmentation and sieving results for gyratory crusher products. Secondary fragmentation for picrite, reporting to drawpoints located near the southern boundary of the cave, was generally found to be finer, while coarser fragmentation was associated with monzonite rock. The remaining rock types, hypogene and phyllic-altered rock, were grouped based on similarities in fragmentation size. 57 Figure 4.1 Size distribution curves for secondary fragmentation at New Afton 4.3 Comparisons with Block Cave Fragmentation (BCF) Predictions Comparisons of predictions from BCF software and measurements are shown in Figure 4.2. The program does not account for the influence of undercut blasting on fragment size. For this reason, outputs of the program for material that was taken below the back of the undercut were not compared to measurements of drawpoint muck. The presented data is relevant to the central area of the B2 cave where hypogene material is predominantly collected. Size distributions for measurements taken underground were grouped with respect to their corresponding HOD, which was estimated using the PCBCTM production scheduling module in the GEMS GEOVIA, version 6.7.4, software package (Dassault Systèmes, 01020304050607080901000.01 0.1 1 10 100 1000 10000Percentage Passing, [%]Particle Size, [mm]Crusher Product 2015Crusher Product 2016Crusher Product 2017 (Huang et al, 2017)Hypogene & Phyllic (B1 & B2), mean of 53 measurementsMonzonite (B2), mean of 5 measurementsPicrite (B1), mean of 3 measurements58 2017). Within PCBCTM, HOD is calculated based on drawpoint spacing, tonnage drawn from a drawpoint, bulking factors and an assumed geometry of the drawcone. A constant bulking factor and drawcone shape was used within the software for all drawpoints. The results for approximately eight images were averaged for each HOD value shown in Figure 4.2. BCF predictions were converted from a block volume size, in units of m3, as output by the software, to a screen aperture size that could be compared to the output of WipFragTM software. Size conversion was carried out under the assumption that blocks are in the shape of a square prism, where the screen aperture size is equal to the side-length of the square. The assumption that blocks output by the BCF program be treated as square prisms was recommended by the developer of BCF, GS Esterhuizen (2017 email from GS Esterhuizen to the author, unreferenced). Block aspect ratios were provided by the BCF software for each block and used in the calculation process. BCF-based predictions were generally in line with measurements for aperture sizes coarser than 0.5 m. However, the software was found to under predict the content of material in the finer size fractions. 59 Figure 4.2 Size distribution curves for secondary fragmentation at New Afton, B2 cave 4.4 Back-calculation of Gyratory Crusher Feed Feed to the gyratory crusher, located on the haulage level, consists of drawpoint muck that has passed through orepasses and been hauled to the crusher surge pocket. In order to estimate the size of crusher feed, sieving results for the 2.2 tonne sample of crusher product were used in combination with Anderson-Arachie-Whiten crusher models (JKTech Pty Ltd, 2015) that are available in the JKSimMet software package, version 6.1 (JKTech Pty Ltd, 2015). At the time of sampling, the crusher was operating with a closed-side setting of 80 mm. Following sieving, standard JK Drop Weight tests (JKTech Pty Ltd, 2016) were carried out on samples of crusher product to determine their hardness and appearance functions. Analysis of 0102030405060708090100100 1,00050m HOD (BCF) 100m HOD (BCF) 200m HOD (BCF)49m HOD (WipFrag) 107m HOD (WipFrag) 238m HOD (WipFrag)PercentagePassing, [%]Particle Size, [mm]60 individual rocks and assays of representative sub-samples showed that the bulk sample was composed of high grade hypogene ore. To model the crusher feed, WipFragTM measurements for drawpoints that were similar to the collected crusher product sample in terms of rock lithology and alteration were averaged. The average size distribution was treated as initial feed to the crusher model within the JKSimMet package. Following comparison of modelled product sizes and sieving results, the percentage content of each size fraction in the feed to the crusher model was manually adjusted and the crusher model was run again. The approach was repeated until reasonable alignment between modelled and measured product was achieved, shown in Figure 4.3. Included in the graph is the average size distribution of secondary fragmentation, 61 image-based measurements in total, taken up to the time of mill feed sampling. A similar approach was used by Esen et al. (2007) to estimate the size of blasted material being fed to a primary crusher. The modelling exercise matches the results of image analysis for sizes coarser than 125 mm. For size fractions below 3 mm, the feed and product curves overlap. An explanation for this is that the gyratory crusher does not influence the content within these size fractions. Alternative methods are required to estimate the percentage content of the finer fractions. 61 Figure 4.3 Modelled crusher product and feed sizes 4.5 Fragmentation Size and Rock Type For the eleven drawpoints that had been sampled, size distributions from image-based measurements were compared to the content of each rock type at the time of sampling. It was found that finer muckpiles, as measured with WipFragTM, contained larger proportions, on a mass basis, of both fault and carbonate rock. Since carbonate rocks at New Afton originate from fault-related alteration, observations indicated that the secondary fragmentation size at New Afton is strongly related to the degree of faulting. Figure 4.4 shows a comparison of fragmentation size and the quantity of carbonate and fault rock. The HOD for all samples was at least 28 metres higher than the undercut, meaning that fragment sizes of these rocks would be unaffected by blasting. 01020304050607080901000.01 0.1 1 10 100 1000 10000Percentage Passing, [%]Particle Size, [mm]Crusher Product 2016 Ave. WipFrag Size (B1 & B2)Estimated Crusher Feed Modelled Crusher Product62 Figure 4.4 Secondary fragmentation size in relation to the content of carbonate and fault rock Through analysis of test data, it was found that fault and carbonate rocks can be distinguished from other rocks within the New Afton deposit by referencing their respective XRF, magnetic susceptibility and point load index measurements. In comparison to other rock types reporting to drawpoints, carbonates were found to have high specific gravity, high concentrations of calcium and manganese, and low magnetic susceptibility. Fault rocks were distinguishable due to their low point load strength and high sulphur content. Magnetic susceptibility, point load strength and assay data are included in the New Afton block model. Therefore, fragmentation predictions can be carried out for future lifts by estimating the content of carbonate and fault rocks in future cave footprints and combining the results with historical fragmentation logs for existing caves, as shown in Figure 4.4. 0501001502002503003504004500 5 10 15 20Muckpile F80 Size at Drawpoint, [mm]Content of Both Carbonates & Fault Rock in Collected Sample, [%]Y = 341.0.exp(-0.115*X) + 56.32 R2 = 0.76 63 4.6 Comparison of Hang-Up Frequency to Fragmentation Results Hang-up and production logs from the period of September, 2011 to November, 2016 were used to determine the hang-up frequency for each month, shown in Figure 4.5. The figure represents 8,430 datapoints for both B1 and B2 caves. Figure 4.6 and Figure 4.7 show how the top-size and 80% passing sizes changed with increasing HOD. Intervals for 90% confidence, assuming data follows a normal distribution, have been included in all three figures. In the case of the B2 cave, for which 59 size measurements were available, a distinct peak in both hang-up frequency and fragmentation size is apparent for an HOD of approximately 30 metres. This trend is expected because material originating from an HOD that is less than this height has been fragmented by drawbell and undercut blasting. Therefore, material above the undercut level has broken from the cave back, representing primary fragmentation, and experienced minimal breakage through secondary fragmentation. As more material is drawn, it can be noticed that both hang-up frequency and fragment size reduce. For the B1 cave, a total of 30 fragmentation measurements were available. Hang-up logs and fragmentation measurements indicated that fragmentation at the B1 cave is significantly finer. Figure 4.5 Hang-Up Frequency for East and West Caves (including 90% confidence intervals) 0.00.10.20.30.40.50.60.70.80.91.00 50 100 150 200 250Hang-Ups per kilotonne MuckedHeight of Draw, [m]B1 (East Cave) B2 (West Cave)64 Figure 4.6 100% passing sizes of drawpoint muck (including 90% confidence intervals) Figure 4.7 80% passing sizes of drawpoint muck (including 90% confidence intervals) 0200400600800100012000 50 100 150 200 250Top Size, [mm]Height of Draw, [m]B1 (East Cave) B2 (West Cave)01002003004005006007000 50 100 150 200 25080% Passing Size, [mm]Height of Draw, [m]B1 (East Cave) B2 (West Cave)65 Higher hang-up frequencies are observed at drawpoints that are located at the perimeters of the B1 and B2 caves, as shown in Figure 4.8. This is particularly apparent at the western edge of the B2 cave, where coarser fragmentation was also measured. Figure 4.8 Hang-up frequency for HOD of 60 to 100 metres 4.7 Relating Fracture Intensity from DFN Models to Fragmentation Size A DFN model was developed for the New Afton B2 Cave footprint by Munkhtsolmon Munkhchuluun, as part of his Master’s research in the Cave-to-Mill project. The following is a summary of the methodology used to build a DFN model and relate fracture intensity to cave fragmentation. More detail can be found in Munkhchuluun (2017). P10 values, which represent the number of fractures per unit length, were referenced from provided geotechnical drill core data. Information from drift mapping at the apex level provided data for fracture orientations and length. Through iterative DFN modelling, this information could be used to relate P10 for a drill core interval to the equivalent P32 value, which represents fracture area for a unit of volume. The relationship between the volumetric size distribution of the in situ rock mass and fracture intensity is presented in Figure 4.9. 66 Figure 4.9 In situ fragmentation based on DFN analysis for varying P32 intensities, and weighted averages for B2 (West) and B1 (East) Caves (Munkhchuluun, 2017) Ordinary kriging was used to build a P32 block model with 10 m cubic blocks for the B2 Cave. Major structures were deterministically located in the block model and assigned a P32 value of 20 (high fracture intensity), which accounted for 8% of block model volume. Figure 4.10 shows the resulting P32 block model. 67 Figure 4.10 Geostatistical P32 block model (Munkhchuluun, 2017) The P32 block model was used with New Afton Mine’s PCBC parameters, including the profile of draw columns and historic production schedules, to determine the P32 values reporting to individual drawpoints for each month of production. The corresponding HOD of drawpoint muck could be estimated from knowledge of the cumulative tonnage drawn from each drawpoint. Essentially, fracture intensity P32 was treated as a grade parameter within PCBC. In order to relate in situ fracture intensity to hang-up frequency, hang-up frequency data was grouped into 10 m ranges of height of draw to reduce the influence of secondary fragmentation. An example is presented in Figure 4.11, where hang-up frequency and fracture intensity data for a height of draw of 80 to 90 metres within the B2 Cave is included. The graph shows that hang-ups were less frequent when material originating from more highly fractured rock mass (higher P32) was drawn. As a comparison, Brzovic et al. (2016) reported P32 values ranging from 2 to 15 m2/m3 for the El Teniente Mine. 68 Figure 4.11 Hang-up frequency and fracture intensity P32 for material with a HOD of 80-90m with 90% confidence intervals. From data generated by Munkhchuluun, (2017) A relationship between hang-up frequency and fracture intensity was also apparent for other HOD ranges, as shown in Figure 4.12. Based on the results, Munkhchuluun (2017) concluded that the fragmentation of material at low HOD values (affected by undercut blasting) is strongly influenced by fracture intensity while the fragmentation of more mature draw columns (high HOD) is more reliant on breakage through secondary fragmentation. Therefore, the proposed method is particularly relevant for predicting fragmentation size and by extension, mine productivity, for earlier stages of drawpoint production. y = -0.03x + 0.46R² = 0.830.000.050.100.150.200.250.300.350 5 10 15 20Hang-up Frequency, (HUP/kt)P32, Fracture Area per Unit Volume, (m2/m3)69 Figure 4.12 Hang-up frequency and fracture intensity P32 for three HOD ranges. From DFN data generated by Munkhchuluun (2017) 4.8 Influence of Veining on Fragmentation Size at New Afton Veining has been identified as a contributing factor to the cave fragmentation process by a number of publications, including Bewick and Keiser (2016) and Brzovic et al. (2014). Quantitative data describing the quantity and type of veinlets in drill core pertaining to the East and West caves was not available. During inspection of collected drawpoint muck, rocks with calcite-coated faces were present in most samples. Since calcite is a common in-fill material at New Afton, it is assumed that the presence of calcite-coating is due to preferential breakage having taken place along calcite veins. Many rocks were also found with intact veins, similar to that shown in Figure 4.13. 0.000.100.200.300.400.500.600.700 5 10 15 20Hang-up Frequency, (HUP/kt)P32, Fracture Area per Unit Volume, (m2/m3)20 - 30 m HOD (Blasting influence) 80 - 90 m HOD 150 - 160 m HOD70 The presence of vein in-fill material on rock faces indicates that veining does play a role in the fragmentation process at New Afton. However, the extent of its influence on muck size is unclear. Figure 4.13 Drawpoint sample E13S (West Cave) taken November, 2015 (+50 mm) 71 4.9 Discussion and Conclusions Results indicated that fragmentation at New Afton is primarily controlled by faulting, as shown by Figure 4.4. The wide range in 80 percent passing size, from approximately 100 to 400 mm, for the eleven drawpoint muckpiles, was found to be strongly related to the abundance of carbonate and softer fault rock where the alteration has been obscured. Furthermore, this hypothesis is supported by the lower hang-up frequencies and finer fragmentation sizes that were recorded for the B1 cave, which had a significantly larger number of fault structures intersecting the in situ rock mass. Although a sizeable quantity of fragmentation size measurements and point load strength data was available, there was no evidence of a strong correlation between the two parameters. Prior to this study, it was expected that coarser muckpiles would be associated with higher point load strengths. The lack of correlation is attributed to the fact that competent carbonate rocks, for which an average Is(50) value of 5.3 MPa was measured, are also present in faults that have a direct bearing on fragmentation size. The impact of HOD on fineness, shown by fragmentation measurements and hang-up logs, was found to be lower for the B1 cave. At the El Teniente mine, the influence of HOD on fragment size was not significant for rock masses that had high volumetric fracture intensity, defined by P32 (Brzovic et al., 2016). This is inline with findings from the work done by Munkchuluun (2017) for this Cave-to-Mill research program, where fracture intensity was linked to hang-up frequency. The phenomenon may explain the lower degree of size degradation with HOD that was observed for the B1 cave. For the B2 cave, the relationship between fragmentation size and HOD was similar to results presented by Moss (2004) for the Palabora cave, where the coarsest fragmentation was measured for samples originating at the zone of primary fragmentation followed by a reduction in size with increasing HOD. The DFN component of the fragmentation study by Munkhchuluun (2017) showed that fracture intensity P32 is relevant to fragmentation size and hang-up frequency. Similar to the findings of Brzovic et al. (2016) at the El Teniente Mine, the largest reduction in hang-ups was found during 72 the earlier stages of drawpoint extraction. For future application of DFN-based fragmentation models, two refinement stages are proposed: 1) repopulation of DFN models as access to exposed ore increases during project development 2) calibration with size measurements of caved ore and hang-up logs when the project moves towards production. In the case that macro blocks are planned for a deposit, calibrated fragmentation models for an initial macro block can be treated as valuable design and planning tools for future lifts. For the New Afton B2 cave, BCF-based percentage passing predictions for size fractions coarser than 0.5 m were within 5% of image-based measurements, where the average of approximately eight image-based size distributions was compared to corresponding BCF predictions. Similar to the findings published by Brunton et al. (2016), BCF software was found to under predict the content of fines. This may be due to the proportion of faulting assumed in the B2 cave when running the BCF simulations. Since fault and carbonate rocks can be identified on the basis of point load strength, XRF response and Magnetic Susceptibility readings, information which is contained in the New Afton block model, fragmentation predictions can be made for future caves at the New Afton deposit. The effect of HOD on fragmentation size for future lifts can likely be inferred from the data shown for the B1 and B2 caves. However, more analysis on the influence of the cave boundary on hang-up frequency and fragmentation size is required. Image-based size analysis was found to be practical for particle sizes coarser than 100 mm. The percentage passing of aperture sizes below 100 mm is of relevance to material handling systems, production potential of the mill and the potential of sensor-based sorting systems to upgrade ore. The full-size distribution of gyratory crusher feed was back-calculated using the results of sieving and a crusher model in the JKSimMet simulation package. The method assumed that significant breakage does not occur in the orepasses, which are approximately 25 metres in length. Overall, secondary fragmentation at New Afton was found to be sensitive to faulting, fracture intensity, HOD and boundary effects. In small caves, such as the B1 and B2 caves, a large 73 proportion of the drawpoints is influenced by boundary effects, making fragmentation predictions particularly challenging. 74 Chapter 5: Relating Fragmentation Size to Mill Performance 5.1 Introduction Cave fragmentation is a key cave-to-mill parameter as it has implications on the productivity of both mining and milling processes. Chapter 4 showed that hang-up frequency and caved ore size at the New Afton Mine can be related to the geology, fracture intensity, proximity of drawpoints to the cave boundary and the amount of material drawn through a drawpoint. The focus of this chapter is to determine the impact of fluctuations in caved material size on mill performance. Fragmentation measurements of drawpoint muck, comminution tests and calibrated mill models were used to assess the impact of variations in feed size and hardness on New Afton mill performance. Image-based size analyses of drawpoint muck and comminution tests showed that coarser material generally contained harder rock. 5.2 Analysis of Mine and Mill Data Four months of production data, September to December 2016, were analysed to identify relationships between the location of the ore being mucked, mill feed size and mill throughput. The extraction level layout and the drawpoints that were active during the operating period are shown in Figure 5.1. 75 Figure 5.1 New Afton drawpoint layout including drawpoints in operation during the four-month period analysed for this study. B1 Cave drawpoints that had reached over 50 m in height of draw by the end of the 4-month operating period are shown in red. Mill feed size, as measured online by a camera mounted on the mill feed conveyor, was compared to mill throughput during periods of operation when the mill was throughput constrained. These periods were defined as one-hour intervals of operation when the mill did not achieve the throughput set by the mill operator due to control system interlocks engaging to maintain product size fineness, reduce pebble recirculating load and/or adhere to other control system criteria. Throughput data for mill-constrained operation were selected from the four-month operating period as they represented the maximum potential of the mill for the size and hardness of feed material being processed. Comparisons indicated that productivity of the New Afton mill is sensitive to the size of mill feed, as shown in Figure 5.2. Decreases in the 80% passing size of SAG mill feed (F80) from 90 mm to 70 mm result in improvements in throughput rate from approximately 570 to 660 tph. A 1,9001,9401,9802,0202,0602,1002,1403,200 3,300 3,400 3,500 3,600 3,700 3,800 3,900 4,000North, [m]East, [m]Drawpoints in Operation Layout East Cave (+50m HOD)76 Figure 5.2 Mill feed size and throughput when circuit is constrained by grinding circuits for operation during Sep to Dec, 2016 To relate the source of mill feed to mill feed size and throughput, the amount of material mucked from each drawpoint for each 12-hour shift was compared to mill feed conveyor camera data and logs for mill throughput; camera and throughput data were available as one-hour averages. Based on the size of orepasses, bins and stockpiles, a six-hour offset was applied to camera and mill data to account for the time taken for material to travel from drawpoints to the mill. An analysis of underground production data showed that mill feed size and throughput was sensitive to the proportion of run of mine material that was taken from drawpoints that have reached over 50 metres height of draw (HOD), located in the B1 Cave. Their locations are also indicated in Figure 5.1. This is in line with results from Chapter 4 of this thesis, which showed that: 1) B1 Cave drawpoint muck was significantly finer than B2 Cave drawpoint muck, and 2) for material originating from HOD heights above the undercut level (approximately 30 m), fragmentation size reduces as more material is drawn. The change in feed F80 size due to increases in the proportion of material taken from drawpoints with HOD greater than 50 m within the B1 Cave is presented in Figure 5.3. The size of the confidence intervals is attributed to fluctuations in material residence time and size segregation 40045050055060065070075020 30 40 50 60 70 80 90 100 110 120SAG Mill Throughput, [mtph]SAG Mill Feed F80 Size, [mm]𝑦 = 732(1 − 𝑒−10.39𝑒−0.02𝑥) R2 = 0.58 77 occurring within the material handling system, and instrument errors. In comparison to the throughput and feed size data presented in Figure 5.2, the range of mill feed F80 sizes shown in Figure 5.3 is significantly lower. Each datapoint represents an average of approximately 23 days of operation; averaging over extended periods dampens the extreme variations observed over shorter time periods. Figure 5.3 Mill feed size measured by an online camera on the mill feed conveyor and proportion of material from drawpoints within B1 Cave where HOD is greater than 50 m (including 90% confidence intervals). Each datapoint represents the average of 22 to 23 operating days (24 hours) The analysis of production data also showed an increase in mill throughput when a larger proportion of material was sourced from B1 Cave drawpoints with HOD greater than 50 m, shown in Figure 5.4. The average mill throughput increased from 670 to 700 tph when material drawn from these B1 Cave drawpoints increased from 14 to 26 % of overall mill feed. y = -0.18x + 65.57R² = 0.765054586266700 5 10 15 20 25 30Crusher product P80, [mm]Proportion of production from B1 Cave (+50m HOD), [%] 78 Figure 5.4 Mill throughput when the mill is throughput constrained and the proportion of material from drawpoints within the B1 Cave where HOD is greater than 50 m (including 90% confidence intervals). Each datapoint represents the average of 15 to 16 operating days (24 hours) The analysis of mine and mill data did support the hypothesis that the mine production schedule has an impact on the size of mill feed and subsequently affects mill throughput performance. It is important to note that during periods of mine production where larger proportions of material were taken from B1 Cave drawpoints with HOD greater than 50 m, less material was being taken from other areas of the mine, such as the B2 Cave which was found to have significantly coarser material (as presented in Chapter 4). 5.3 Cave-to-Mill Modelling and Simulation Methodology Mine and mill data analysis in the previous section was linked to the F80 size of mill feed, as measured online by conveyor-mounted cameras. To better understand the role of the hardness and size of caved material on mill throughput performance, the New Afton mill was modelled with calibrated crushing and grinding models using inputs from a comprehensive test program. The flowchart presented in Figure 3.5 shows the steps taken to calibrate mill models and collect feed size and hardness data for use as model inputs for simulation. The results of two mill surveys were used to validate the mill models. y = 2.51x + 637.26R² = 0.986206406606807007200 5 10 15 20 25 30Mill throughput, [mtph]Proportion of production from B1 Cave (+50m HOD), [%]79 Figure 3.5 Test regimen for the core of the test program (repeated for convenience) 5.4 Size Analysis of Drawpoint Samples Over the course of two days, eleven drawpoints were sampled after photos had been taken for size distribution analysis using WipFrag software. The drawpoints were chosen to ensure that the major ore types and end-members of the New Afton deposit were included. Further detail on the image-based size analyses is included in Section 3.3. Samples weighing approximately 50 kg for each drawpoint were shovelled from along the width of the base and mid-section of each drawpoint muckpile, described in greater detail in Section 3.4. To characterise the fines content of drawpoint muck, the -25 mm fractions from each drawpoint were sieved using a sieve shaker to a final screen size of 0.5 mm. The goodness of fit achieved with several mathematical models was compared and it was found that the Gaudin-Schuhmann equation, shown in equation 5.1, provided the best fit. 80 𝑀(𝑥)𝑀0= (𝑥𝐾)𝑚 (5.1) where x is the particle size, M(x) is the weight of undersize, K is the size modulus and m is the slope parameter (Gaudin et al., 1962). As presented in Chapter 4, image-based size measurements of muckpiles at drawpoints were found to be reliable for particle sizes coarser than approximately 100 mm. Since sieving results of muckpile samples were only available for -25 mm material, an approach for estimating the percentage passing aperture sizes from 25 to 100 mm was developed. The Gaudin-Schuhmann equation was fitted to WipFrag measurements for size fractions of 125 mm and above. The fitted model was then used to estimate the percentage passing for size fractions between 100 and 25 mm. Similarly, a Gaudin-Schuhmann model was fitted to sieving results for collected -25 mm drawpoint muck samples using a size modulus, K, value of 25 mm. The determined slope, m, was then applied to determine the percentage passing for size fractions below 25 mm. An example of the approach is shown in Figure 5.5 for drawpoint D7S. The method was carried out for eleven drawpoint muckpiles to estimate the size of gyratory crusher feed that would result for each case. 81 Figure 5.5 Example of estimated gyratory crusher feed size for sample D7S 5.5 Material Hardness Characterisation Point load, DropWeight (JK Tech, 2016), low-energy abrasion tests (JK Tech, 2016) and Bond ball mill work index tests were carried out for the eleven drawpoint samples and a mill feed sample, which was taken for model fitting. Table 5.1 shows a summary of the material properties and measured F80 muckpile sizes. 01020304050607080901000 0 1 10 100 1,000Percentage Passing, [%]Particle Size, [mm]D7S WipFrag Fines (Gaudin-Schuhmann)Coarse (Gaudin-Schuhmann) D7S Gyratory Feed82 Table 5.1 New Afton mill feed and geological unit characteristics Sample & Cave (B1/B2) Specific Gravity F80 Size of Muck DropWeight Index , kWh/m3** ta BBWI [-] [mm] [-] [-] [kWh/mt] SAG Mill Feed 2016 2.75 370* 6.98 0.29 18.4 #1, Hypogene B2 2.70 394 8.31 0.41 19.3 #2, Hypogene B2 2.67 99 6.91 0.39 18.4 #3, Hypogene B2 2.72 213 6.86 0.40 18.0 #4, Hypogene B2 2.72 117 8.08 0.27 19.4 #5, Hypogene B2 2.73 137 7.79 0.36 19.3 #6, Monzonite B2 2.72 263 6.91 0.28 17.9 #7, Phyllic-Altered B2 2.69 303 5.82 0.44 18.8 #8, Hypogene B1 2.76 276 7.08 0.38 20.8 #9, Secondary Hyp. B1 2.66 411 6.47 0.50 20.1 #10, Picrite B1 2.64 386 3.91 0.66 17.9 #11, Supergene B1 2.72 166 4.45 0.92 18.6 * Estimated from crusher modelling and sieving of crusher product ** Calculated using method of Doll (2016) to convert density and Axb values, from DropWeight testing, to DropWeight Index (kWh/m3) 5.6 Mill Surveys, Model Calibration and Validation A mill survey, for which 2.2 tonnes of mill feed was collected, was carried out in March, 2016 to generate data to fit established comminution models. Sampling points are shown in the mine and mill flowsheet (Figure 3.6). A summary of the operating performance of the mill during the survey is shown in Table 5.2. Data from two additional mill surveys which were used to validate fitted mill models have also been included. For the 2015 and 2017 surveys, approximately 700 and 1,100 kg of mill feed sample was collected from the SAG mill fresh feed conveyor, respectively. Further detail regarding the mill surveys is presented in Section 3.5. 83 Table 5.2 Summary of New Afton mill surveys Area Parameter Units Survey (2016), Model Fitting Survey (2015), Validation #1 Survey (2017), Validation #2** Feed Properties Fresh feed rate, dry [mtph] 700 651 728 Specific gravity [-] 2.75 2.72 2.72 Feed size, F80 [mm] 60 35 59 Feed size, F50 [mm] 18 12 18 Feed size, F20 [mm] 3 2 4 DropWeight, A [-] 61 70.3 58 DropWeight, b [-] 0.64 0.51 0.70 Abrasion parameter, ta [-] 0.29 0.34 0.29 Bond ball mill work index [kWh/mt] 18.4 18.9 19.4 SAG Mill Circuit SAG mill speed, % of critical [%] 64.0 63.8 63.8 SAG mill diameter, inside liners [m] 8.36 8.39 8.32 SAG mill ball load [%] 15 15 17 SAG mill power draw [kW] 4,766 4,324 4,914 Effective screen aperture size [mm] 13.2 9.5 9.5 Pebble crusher bypass [%] 0 19*** 0 Pebble port size (width) [mm] 114.3 88.9 114.3 Media diameter [mm] 158.8 152.4 158.8 Pebble crusher closed side-set [mm] 20 12 20 SAG mill screen undersize F80* [mm] 3.31 NA 2.59 Ball Mill Circuit Ball mill power draw [kW] 5,255 5,446 5,415 Media diameter [mm] 63.5 50 63.5 Medial load [%] 38 38 37 Circulating load [%] 255 372 307 Ball mill cyclone overflow, P80 [µm] 289 226 287 Operating Work Index, Wio [kWh/mt] 16.5 NA 17.2 Bond Energy Factor, Wio/ BBWI [-] 0.99 NA 0.97 Vertimill Circuit Vertimill power draw [kW] 2,090 Not Installed 2,165 Media diameter [mm] Circulating load [%] 143 165 Vertimill cyclone overflow, P80 [µm] 150 152 * Not directly sampled. Determined by mass balances around SAG screen and ball mill cyclone feed sump ** From Cebeci et al. (2017) *** Bypass observed during survey due to magnetite or tramp metal in pebble stream triggering metal detectors prior to pebble crusher 84 5.7 Mill Model Validation To model the SAG mill circuit, the variable rates SAG mill model within JKSimMet (JK Tech, 2015a), a software package for simulation of comminution circuits, was used. Recommendations by Bailey et al. (2009) for application of the model for fully-ported SAG mills, such as the New Afton mill, were followed. Anderson-Arachie-Whiten crusher models (JK Tech, 2015b), were used for pebble crusher simulation. For estimation of ball mill and Vertimill performance, Bond efficiency methods, which are based on the ratio of the operating work index to the Bond ball mill work index of sampled mill feed, were applied. The modelling method is described in more detail in Section 3.6. The model fitted to 2016 survey data was used to simulate the operating conditions of two separate mill surveys. Comparisons of simulated and measured data showed that predictions were within +/-10% for product size, throughput performance and power draw, as presented in Table 5.3. Performance predictions were more accurate for the 2017 survey, during which mill operating conditions were similar to the base case (2016) used to fit the models. Table 5.3 Comparison of measured and simulated mill performance Parameter Units Measured Simulated Percent Error, % Simulation Method Survey 2015 SAG mill power [kW] 4,324 4,710 8.9 JKSimMet Pebble recycle [mtph] 149 159 6.7 SAG screen under size F80 [µm] NA 2,792 - Ball mill power [kW] 5,446 - - Input into Eq. 3.1 Ball mill cyclone overflow P80 [µm] 226 247 9.3 Eq. 3.1 Survey 2017 SAG mill power [kW] 4,914 4,955 0.8 JKSimMet Pebble recycle [mtph] 196 208 5.7 SAG screen under size F80 [µm] 2,587 2,736 5.8 Ball mill power [kW] 5,415 5,419 0.1 Eq. 3.1 Ball mill cyclone overflow P80 [µm] 287 299 4.3 Eq. 3.1 Vertimill power [kW] 2,165 - - Input into Eq. 3.3 Vertimilll cyclone overflow P80 [µm] 152 153 0.3 Eq. 3.3 85 Simulation of Mill Performance Fitted models were used to determine the mill throughput potential associated with the eleven drawpoint samples, each having a unique size distribution and competency. The boundary limits for modelling were defined as starting at the feed to the primary crusher and ending at the cyclone overflow of the Vertimill operating in a tertiary grinding role. Product from the Vertimill is fed to a rougher flotation circuit. Table 5.4 shows the equipment and process constraints that were adhered to when calculating maximum throughput performance. Table 5.4 Equipment and process constraints for simulations Equipment Units Value Constraint Type Max SAG mill motor power draw [kW] 5,000 Drivetrain Max Vertimill motor power draw [kW] 2,200 Drivetrain Max Ball mill motor power draw [kW] 5,450 Drivetrain Max pebble crusher tph [mtph] 250 Material handling Max SAG volume load [%] 30 Operational Fixed product size, Vertimill cyclone overflow [µm] 150 Flotation Recovery A comparison of SAG mill feed size and mill throughput potential is shown in Figure 5.6. Simulated throughputs varied from 561 to 735 tph for F80 feed sizes of 73 to 44 mm. For feed F80 sizes coarser than 64 mm, simulations indicated that throughput was SAG mill constrained. For finer feed sizes, the bottleneck would shift to the ball and Vertimill circuits, at which point ball and Vertimill power draws were at their operational maximums. 86 Figure 5.6 Simulated mill throughput rates and mill feed size for 11 drawpoint samples. Filled markers indicate SAG mill limited operation. Empty markers indicate ball and Vertimill limited operation. For the range of available hardness parameters that were determined, the DropWeight index most closely correlated to the simulated throughput rates, shown in Figure 5.7. Simulations showed that throughput performance was constrained by the SAG mill for more competent ore types having DropWeight Index values greater than approximately 6.6 kWh/m3. y = -5.52x + 817.54R² = 0.86y = -9.30x + 745.54R² = 0.724004505005506006507007500 20 40 60 80 100Mill throughput rate, [mtph]SAG Mill feed size [mm]Feed F80 size Feed F50 Size Feed F20 Size𝑦 = 730(1 − 𝑒−160𝑒−0.07𝑥) R2 = 0.92 87 Figure 5.7 Simulated mill throughput rates and DropWeight Index parameters Simulations provided an understanding of the sensitivity of mill throughput to changes in ore hardness and feed size. Figure 5.8, which shows the relationship between impact hardness and the size distribution of the muckpile, indicates that hardness and feed size are not independent of each other. Coarser sample is generally harder. Therefore, drawpoint muck size, which directly relates to mill feed size, and ore hardness are considered to be material specific parameters that affect mill throughput and the location of bottlenecks. Figure 5.8 DropWeight Index and measured F80 size of sample drawpoint muckpiles y = -35.52x + 887.08R² = 0.714004505005506006507007503 4 5 6 7 8 9Simulated mill throughput, [mtph]DropWeight Index, [kWh/m3]Hypogene (B2) Hypogene (B1) Picrite (B1) Sec. Hypogene (B1)Phyllic (B2) Monzonite (B2) Supergene (B1)y = 0.01x + 4.18R² = 0.6301234567890 100 200 300 400 500DropWeight Index, [kWh/m3 ]Drawpoint muck F80 size, [mm]Hypogene (B2) Hypogene (B1) Picrite (B1) Sec. Hypogene (B1)Phyllic (B2) Monzonite (B2) Supergene (B1)SAG Mill limited operation Ball & Vertimill limited operation 88 5.8 Influence of Feed Size on Simulated Mill Performance In many power-based comminution models (Bond, 1962, Morrell, 2014 and Doll, 2015) mill feed size and throughput comparisons rely on the F80 as the main indicator of feed size. Esen et al. (2007) mention that in addition to mill feed top size and F80, the proportion of -10 mm material in mill feed affects SAG mill throughput performance (higher proportions improve throughput). An analysis of feed size distributions and simulated mill throughput provided results that are in line with this statement. Using the set of eleven mill simulations, the Pearson correlation coefficient for the relationship between throughput and the proportion of feed material retained on each sieve was calculated. A correlation coefficient of 1 implies that a linear equation perfectly describes the relationship between simulated mill throughput and the proportion retained on sieve size x. While a value of -1 indicates a perfectly inverse relationship. Figure 5.9 shows that the coarser (+27 mm) size fractions of mill feed negatively correlated with simulated mill throughput, while the proportion of material retained on sieves smaller than the SAG screen aperture size (13.2 mm) positively correlated with simulated throughput. Essentially, these fine sizes exit the SAG mill circuit in one pass with minimal contribution to the volume load (and power draw) of the SAG mill. The primary crusher at New Afton, which operates at a closed-side setting of 90 mm, does not noticeably increase the content of -13.2 mm material. Mill throughput forecasting stands to improve should better methods for predicting the content of fines (-10 mm) in drawpoint muck be developed. 89 Figure 5.9 Pearson correlation coefficient between proportion of material retained on each sieve and simulated mill throughput. 5.9 Relationships Between Point Load Index, Impact Breakage and Abrasion Results from the geotechnical and metallurgical test program showed that correlations exist between the results from New Afton Point Load Index IS(50) testing and milling parameters describing the resistance of material to breakage by impact and abrasion, represented by the DropWeight Index and ta abrasion parameter, respectively. Figure 5.10 and Figure 5.11 show a comparison of Point Load Index IS(50) results and the DropWeight Index and ta abrasion parameters. The graphs indicate that there is potential for use of the New Afton database of drill core point load strength results to forecast mill performance. Similarly, Burger et al. (2006) found useful correlations between Point Load Index and DropWeight strength. Figure 5.12 shows a minor correlation between muckpile 80% passing size and the resistance of the samples to abrasion, as determined by an abrasion test, ta. A weak correlation between muckpile 80% passing size and point load strength was apparent, presented in Figure 5.13. -1.0-0.8-0.6-0.4-0.20.00.20.40.60.81.01 10 100 1000Correlation coeff. between % retained in feed & tphSieve size, [mm]Drawpoint muck size Gyratory Crusher ProductEffective SAG screen aperture size90 Figure 5.10 Point Load Index IS(50), including 90% confidence intervals, and DropWeight Index results for twelve samples Figure 5.11 Point Load Index IS(50), including 90% confidence intervals, and Abrasion, ta, results for twelve samples 38 samples27 samples74 samplesy = 0.77x - 0.28R² = 0.620123456783 4 5 6 7 8 9Point Load Index I S(50), [MPa]DropWeight Index, [kWh/m3]Hypogene (B2) Hypogene (B1) Picrite (B1) Sec. Hypogene (B1)Phyllic (B2) Monzonite (B2) Supergene (B1) Mill Feed (2016)38 samples27 samples74 samplesy = 2.22x-0.84R² = 0.850123456780.0 0.2 0.4 0.6 0.8 1.0Point Load Index I S(50), [MPa]abrasion, taHypogene (B2) Hypogene (B1) Picrite (B1) Sec. Hypogene (B1)Phyllic (B2) Monzonite (B2) Supergene (B1) Mill Feed (2016)91 Figure 5.12 Abrasion Index, ta, and measured F80 size of drawpoint muck Figure 5.13 Point Load Index IS(50), including 90% confidence intervals, and measured F80 size of drawpoint muck y = 0.748e-0.002xR² = 0.540.00.10.20.30.40.50.60.70.80.91.00 100 200 300 400 500Abrasion index, taDrawpoint muck F80 size, [mm]Hypogene (B2) Hypogene (B1) Picrite (B1) Sec. Hypogene (B1)Phyllic (B2) Monzonite (B2) Supergene (B1)52 samples39 samples27 samplesy = 0.01x + 2.58R² = 0.580123456780 50 100 150 200 250 300 350 400 450Point Load Index I S(50), [MPa]Drawpoint muck F80 size, [mm]Hypogene (West) Hypogene (East) Picrite (East)Sec. Hypogene (East) Phyllic (West) Monzonite (West)92 5.10 Discussion Analysis of mine and mill data indicated that finer mill feed and increased mill throughput was achieved when higher proportions of mucked material were sourced from B1 Cave drawpoints that had an HOD greater than 50 metres. Simulations showed how mill throughput varies due to changes in mill feed size and hardness. The question then arises: what levers are available to mining engineers to generate finer mill feed? Aside from preconditioning methods, one lever is the adjustment of the closed-side setting of primary crushers to an operational minimum, as carried out at New Afton. A second lever is the inclusion of fragmentation data in the decision making that goes into planning underground production schedules. An example of the second lever being put into practice is the blending of material at the Grasberg block cave operation to ensure that the fines content in the material handling system is below levels that cause problems at the haulage level. The fines content observed at different areas of the extraction level is one of the inputs that is considered when setting the production schedule. Although blending practices at Grasberg are not directed at improving mill performance, the example does show that sufficient flexibility exists within production schedules, even though they are usually nominated to meet grade targets and constrained by the requirement to maintain similar draw rates between neighbouring drawpoints. Installation of secondary crushers ahead of existing SAG mill circuits also provides opportunity to improve mill throughput productivity at existing operations and reduce fluctuations in performance. High Pressure Grinding Roll (HPGR) based circuits are increasingly popular alternatives to SAG mill-based circuits due to their improved energy performance. HPGR-based circuits are known to be less sensitive to variations in ore size and hardness (Klymowsky, 2011) and their implementation would substantially offset the impact of variation in caved ore size on throughput performance. The collection of fragmentation data was limited due to the time-consuming process of manually correcting images of drawpoint muckpiles. Placement of cameras at draw-points, orepasses or crusher excavations is a solution to continuously measuring the size of caved ore. At the LKAB 93 Kiruna sub-level caving operation, Thurley et al. (2015) used a 3D-imaging system to produce automated size estimates of material in LHD buckets. The technology to monitor fragmentation using 3D imaging already exists; a critical step forward is its implementation to provide online measurements of cave fragmentation (Nadolski et al., 2015). 5.11 Conclusions Block and panel caving methods are being pushed to their extremes. Caving of deeply situated deposits, more competent rock masses and heterogeneous ore bodies presents greater challenges to predicting cave fragmentation size accurately. The initial lifts of block caving projects provide a valuable opportunity to gather cave-to-mill information to improve evaluation, design and operation of future lifts. Chapter 4 showed how fragmentation size and hang-up frequency relate to geology, fracture intensity, proximity to the cave boundary and the height of draw at the New Afton mine. The second part showed how variation in fragmentation size affects mill performance and the location of mill bottlenecks. The sensitivity of mill throughput to the content of fines (-10 mm) warrants more research on developing practical methods to measure fines content in muckpiles to inform approaches to improve fragmentation and flow models. Further work on defining the relationship between muckpile 80% passing size and DropWeight Index would help characterise ore domains in terms of mine and mill productivity. For the eleven samples tested in this study, there was an apparent correlation between muckpile 80% passing size and the resistance of the samples to abrasion, as determined by an abrasion test, ta. The study made use of results from the initial New Afton lift, comprising of the B1 and B2 caves. It is envisaged that mill feed will continue to decrease in size as the B2 cave, a source of coarser ore, depletes and finer material from maturing drawpoints of the B1 cave is sent to the mill. The improvement in mill size fineness may warrant changes to be made to equipment in the mill to increase throughput, such as increasing the duty of the SAG mill by reducing the aperture size of the SAG mill screen or changing the size of steel media used in the SAG and ball mills. 94 Chapter 6: Heterogeneity Analysis for Assessment of Sorting Potential 6.1 Introduction Sensor-based sorting was identified earlier as a key focus area of the proposed Cave-to-Mill approach as it address the lack of selectivity and potential for dilution associated with the block cave mining method. Online monitoring and bulk-sorting systems hold the potential to alleviate this problem and improve the productivity of caving operations while reducing energy and water requirements for each tonne of concentrate produced. The potential to sort is limited by the natural grade heterogeneity of the material being assessed. Where heterogeneity exists, there is an opportunity to use sensors to discriminate between ore and waste. For this reason, heterogeneity assessment is typically the initial component of a sorting study. The definitions for Constitutional (CH) and Distributional (DH) heterogeneity, presented by Gy (2004), have been adopted to assess particle and bulk sorting, respectively. CH was used to represent grade variation within particle sorter feed sized particles (-100, + 10 mm) that make up an ore domain. While DH was used to represent grade variation within individual groups of ore reporting to drawpoints; the sum of the groups represents a lot. The lot is a special domain that can represent an entire draw column, in the case that heterogeneity within an individual drawpoint is being assessed, or a horizontal slice of caved ore, in the case that heterogeneity during a period of mine production is being determined. Both Constitutional and Distributional Heterogeneity are affected by the various comminution and blending stages that take place from the onset of caving through to delivery of ore to the mill, as shown in Table 6.1. From the table it can be seen that as ore blocks travel away from their original in situ locations towards the surface stockpile, CH increases and DH decreases. In other words, the opportunity to sort particles increases and, conversely, the opportunity to bulk sort material decreases. It is important to note that the limitations of state-of-the-art particle sorting machines relate to throughput rate and particle size, approximately 10 mm being the finest particle size that can be economically sorted. Therefore, comminution steps that result in sub 10 mm particles being generated do not aid in increasing the sortability of ore. 95 The decrease in DH suggests that bulk sorting of material reporting to drawpoints would be most effective in the case that extraction and haulage levels are designed for separate handling of ore and waste. Table 6.1 Chronological mixing / comminution events and their impact on heterogeneity Chronology Blending / Comminution Event Blending or Comminution Event Impact on heterogeneity Constitutional Heterogeneity Distributional Heterogeneity 1 Fragmentation at cave back Fragmentation (primary) - 2 Collapse onto caved ore muckpile Blending & minor fragmentation 3 Waste entering draw cones Blending (new ore type generated) - 4 Flow through draw cones Blending & secondary fragmentation 5 Haulage to orepass (through grizzly) Blending & minor fragmentation 6 Mixing in orepass Blending - 7 Primary crushing Mechanical comminution - 8 Material flow through crusher product silo Blending - 9 Material transport to surface stockpile Minor blending - 10 Flow through stockpile Blending - Minor Change Significant Change 96 6.2 Evaluation of Distributional Heterogeneity and Implications on Bulk Sorting An analysis of caved material heterogeneity was carried out to evaluate the potential to bulk sort at the New Afton mine. Mine personnel regularly sample drawpoints to support production scheduling decisions and carry out mine and mill reconciliation. As part of the regular sampling campaign, a grab sample of approximately 5 kg is collected for every ~350 tonnes of material that is excavated from drawpoints. Although the sampling procedure is biased by the size of rock that can be collected by samplers (+100 mm rocks not being collected), assays from collected samples provide valuable grade trends. Ross (2012) states that collecting 10 kg of drawpoint sample for every 2,500 tonnes of production yielded statistically relevant grade data at the Palabora block cave mine. New Afton grade data for underground drawpoint samples were analyzed to predict the potential to bulk sort using a grade sensor and diverter system mounted on the underground-to-surface conveyor. For two months of operation, 2,478 copper grade results were analyzed. A cut-off grade of 0.3% Cu was nominated when calculating the proportion of run of mine (ROM) material that would have been diverted by the bulk sorting system prior to the mill. The Distribution Heterogeneity of the mill feed was calculated using Equation 6.1, from Gy (1995), for every day of production. 𝐷𝐻𝐿 = 𝑁𝐺 ∑ [(𝑎𝑛−𝑎𝐿)𝑀𝑛𝑎𝐿𝑀𝐿]2𝑛 (6.1) where Ng is the number of groups in a lot (equivalent to the number of drawpoints mucked during a day of production), L, an and Mn are the grade and mass of an individual group (equivalent to the muck from an individual drawpoint), n, and aL and ML are the weighted-average grade and mass of the entire lot of material (equivalent to the total mill feed for one day of operation). Figure 6.1 shows simulated sorter performance for varying ROM grade and heterogeneity; the figure shows an inverse relationship between DH and ROM grade. Over the two months of 97 operation, 9.7% of material was below the cut-off grade of 0.30 % Cu. Efficient operation of a bulk sorter that is able to divert lots of 300 tonnes and greater, would yield an improvement in mill feed grade from 1.09 to 1.18 % Cu at a copper recovery of 98.5%. Figure 6.1 Distribution heterogeneity of caved muck during two months of operation in 2014 and theoretical bulk sorter performance for a cut-off grade of 0.3% Cu 6.3 Analysis of Heterogeneity at the Extraction Level The extraction level of a cave mine refers to the network of drifts that provide access to drawpoints. Developments in bucket-mounted grade sensors for loaders prompted the analysis of grade heterogeneity within the columns of ore above individual drawpoints. To take full advantage of grade sensors mounted on load haul dumps (LHD), which transport material from drawpoints to orepasses, dedicated ore and waste material handling systems would be required; ideally these would be located in close proximity to drawpoints that experience high variability in ore grade. Figure 6.2 shows the grade heterogeneity calculated using Equation 6.1 for the 176 01020304050607080901000.000.200.400.600.801.001.201.401.6022/Dec 1/Jan 11/Jan 21/Jan 31/Jan 10/Feb 20/Feb 2/Mar 12/MarProportion Milled, [%]Mill Feed Grade, Cu [%], DHMill Feed Grade DH ROM Grade Proportion Milled98 drawpoints that were in production during the period of January 1st to March 3rd, 2014. The figure shows that a greater degree of heterogeneity occurs at the boundary of the cave. Figure 6.2 Distribution heterogeneity of caved muck during two months of operation in 2014 Erdenebat (2017) evaluated bulk material heterogeneity at the New Afton mine as part of his Master’s research within the Cave-to-Mill research project. Grade data from regular drawpoint sampling was compared to the average grade of drawpoints during their life of production (effectively a grade estimate for the draw column), presented in Figure 6.3. The graph shows that bulk-scale heterogeneity generally decreased with grade. 99 Figure 6.3 Distribution heterogeneity and average draw point grade (Erdenebat, 2017) 6.4 Variation of Grade with Size The deportment of gold to different size fractions has a strong influence on the performance of particle sorting systems. In the case that copper-gold mineralization significantly reports to fine fractions, economic upgrading of mill feed could be carried out through size classification alone. Analysis using drawpoint samples showed that the fines (-2 mm) were generally lower in grade than the coarser fractions of (-50, +25 mm) and (-150, + 50 mm). Primary and secondary hypogene samples showed that grade increased with rock size, shown in Figure 6.4. Samples that were diluted with phyllic, picrite and/or monzonite material, presented in Figure 6.5, were found to have higher grades in the middling fraction of -50, +25 mm. From the graphs it can be seen that the variation in copper grade was lower for sub 2 mm material. Since fines are known to flow preferentially within caves, it is expected that they are well mixed in comparison to other size fractions and therefore more homogenous in terms of grade. 01234560.0 0.5 1.0 1.5 2.0 2.5Distributional HeterogeneityDrawpoint Copper Grade, [%]100 Figure 6.4 Copper grade of different size fractions in hypogene drawpoint samples. Grades of overall drawpoint samples from ICP analysis of representative sub-samples are shown in the legend Figure 6.5 Copper grade of different size fractions in drawpoint samples with identified dilution/mineralization type. Grades of overall drawpoint samples from ICP analysis of representative sub-samples are shown in the legend 0.000.501.001.502.002.500 20 40 60 80 100Cu, [%]Geometric Mean, [mm]G40N (2nd Hyp.)0.68% CuE15N, 1.46% CuD7S, 1.09% CuD11S, 1.09% CuD11N, 0.58% CuE13S, 0.37% CuE30N, 0.35% Cu0.000.200.400.600.801.001.200 20 40 60 80 100Cu, [%]Geometric Mean, [mm]B13S (Phyllic),0.44% CuF33S (Picrite),0.40% CuE23N (Monzonite),0.09% CuD38N (Supergene),0.34% Cu-2 mm -50 +25 mm mm -150 +50 mm mm -2 mm -50 +25 mm mm -150 +50 mm mm 101 6.5 Discussion and Conclusions The heterogeneity analysis supported the use of bulk sorting in the proposed Cave-to-Mill approach. When comparing the grade of material reporting to individual drawpoints, higher variations in grade were found to occur at the perimeter of the cave. At New Afton and most cave mines, orepasses are used to handle both ore and waste. This results in ore dilution occurring when material is transported from drawpoints and through the material handling system to surface. The results suggest that bulk sorting systems, such as the use of bucket-mounted grade sensors with production LHDs, should be focused on areas near the perimeter of the cave. Designing extraction and haulage levels to include dedicated ‘waste’ passes may also hold value to future cave operations. Grade data for different size fractions indicates that screening of ROM material prior to particle sorters will generate a lower grade fines stream and a higher grade coarse stream for particle sorting or direct processing. The grade distribution in different size fractions shows that the size bias associated with collecting samples for grade monitoring of drawpoints will affect the accuracy of grade estimates. . 102 Chapter 7: Bulk and Particle Sorting System Evaluation 7.1 Introduction Successful implementation of sensor-based sorting systems stands to reduce milling requirements, improve the quality of mill feed and increase the size of the economic footprint. Following the heterogeneity analysis, presented in Chapter 6, an evaluation of bulk and particle sorting technologies was carried out at the New Afton block caving operation. It is envisaged that implementation of sensor-based sorting as part of the proposed Cave-to-Mill approach has considerable potential to improve the productivity of existing and future block cave operations. From outcomes of the cave fragmentation (Chapter 4) and mill performance (Chapter 5) evaluations, a summary of key rock types and their implications on productivity is shown in Table 7.1 Identification of unique sensor responses for these rock types provides an opportunity to apply sorting strategies according to grade, and mill comminution and recovery performance. 103 Table 7.1 Main lithology and alteration types and implications on the New Afton operation Lithology / Alteration Ore / Dilution Implications on Mine Implications on Mill Hypogene Ore (high grade) Competent rock Highest value mill feed Secondary Hypogene Ore (high grade) Increasing tennantite concentration Supergene Ore Present geotechnical challenges due to low competence. Fine cave fragmentation Lower copper recovery when treated by flotation. Increasing copper recovery by applying gravity concentration. Picrite Dilution Negatively affects froth flotation performance Fault Rock Dilution Low work index Monzonite Waste (monzonite) Competent rock associated with coarse fragmentation Difficult to grind due to high work index Carbonates Dilution Competent rock on a lump scale. Associated with faults 7.2 PGNAA Bulk Sensor Evaluation As discussed in the Methodology Section (3.8), nine samples of approximately 220 kg each were prepared for PGNAA testing. Samples were collected from drawpoints, mill feed and waste 104 stockpiles. Overall, a wide range of copper grades and rock types, as outlined in Table 7.1, were prepared. The nine samples were sent to Scantech testing facilities in South Australia for testing with a Geoscan-M unit, which was fitted with four detectors. Each sample was measured for a period of at least three hours, during which individual instantaneous analysis periods were varied throughout the test work to achieve a number of results for different measurement periods. The expected accuracy of a Geoscan-M unit when installed on a conveyor could then be estimated for different measuring times (which are directly related to the quantity of conveyed material). Figure 7.1 shows a comparison of copper grades measured by the Geoscan-M test unit and ICP analysis of representative sub-samples. The relative standard deviation associated with sampling and ICP analysis for copper content was less than 5% of the mean copper grade of each sample. Included in the figure are 95% confidence intervals based on the sub-sampling and assaying procedures used. Figure 7.1 Predicted copper grades of bulk samples using PGNAA sensors and results from ICP analyses for copper including 95% confidence intervals for sub-sampling and ICP analysis y = x0.000.300.600.901.201.501.800.00 0.30 0.60 0.90 1.20 1.50 1.80PGNAA sensor measurement, Cu, [%]ICP analysis, Cu, [%]Calibration Samples Blind Samples105 The repeatability of copper grade measurements in relation to analysis time and number of detectors, reported by Balzan (2016), is shown in Table 7.2. At New Afton, an underground-to-surface conveyor transports ore and waste at approximately 1,000 tonnes per hour. Based on this conveying rate, the quantity of material transported during each analysis interval has been estimated and included in the table. Selection of an analysis interval represents a trade-off between grade measurement accuracy and sorting lot size, assuming that the conveyor is fitted with a means to segregate the increment. Essentially, the sorting lot size should be nominated with consideration of the heterogeneity of conveyed material. The measurement accuracy of a PGNAA sensor unit can be improved up to a certain limit by increasing the number of detectors at the expense of additional sensor cost. Table 7.2 Measurement repeatability in relation to the material handling system Analysis Period Tonnes Scanned Equivalent Material Handling Lots Number of Detectors Copper Measurement, Standard Deviation [minutes] [tonnes] [#] [%] 2 33 One haul truck load 4 0.035 2 33 6 0.029 2 33 8 0.025 2 33 12 0.020 0.5 8 One LHD bucket 4 0.070 0.5 8 6 0.057 0.5 8 8 0.050 0.5 8 12 0.041 The gold grades associated with New Afton ROM material are below the threshold measurement value of 500 ppm, equivalent to 500 grams per tonne, of the Geoscan-M unit. Applications of the technology for gold grade measurement typically rely on proxy elements being used in a regression equation to estimate gold content. For the case of New Afton, drill core data and 106 grades from regular drawpoint samples show that the average ratio of gold (g/t) grade to copper (%) grade is approximately 0.8. Figure 7.2 shows how gold grade predictions and fire assay results compared. An R-squared value of 0.76 resulted. The 95 % confidence intervals representing sampling and fire assay error ranged from 0.11 to 0.39 g/t. Figure 7.2 Predicted gold grades of bulk samples using PGNAA sensors and results of fire assays 7.3 XRF Particle Sensor Evaluation XRF measurements were carried out on the surfaces of 1,226 rocks in the (-150, +25 mm) size range using a measuring time of 10 seconds for samples taken from drawpoints and mill feed stockpiles. Four surface readings were taken per rock. From the master set of 1,226 rocks, 291 rocks were selected based on rock type and subjected to crushing and pulverizing to determine the grade. More information is provided in the Methodology Section (3.8) and the test work flow is shown in Figure 3.5. y = x0.000.200.400.600.801.001.201.401.601.800.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80Predicted Au, [g/t]Fire assay, Au, [g/t]Calibration Samples Blind Samples107 Matlab R2016b was used to carry out stepwise regression analysis of pXRF readings for rock surfaces and powder material after pulverizing. Model terms with an entrance tolerance less than a p-value of 0.05 and an exit tolerance less than a p-value of 0.10 were accepted into the model. Equation 7.1 shows the resulting model for copper. Figure 7.3 shows a plot of predicted and observed copper grades, where an R-squared of 0.79 was achieved. 𝐶?̂? = 2.41𝐶𝑢𝑎𝑣𝑔 − 0.46𝐶𝑢𝑚𝑎𝑥 − 0.15𝐶𝑢𝑎𝑣𝑔2 (7.1) where 𝐶?̂? is the predicted copper content, Cuavg and Cumax are the average and maximum Cu grades of four pXRF surface readings. The root mean squared error for each rock was weighted by the copper grade from readings of pulverized material. The coefficients in Equation 7.1 were fitted by minimising the sum of the weighted root mean squared error. Figure 7.3 Predicted copper grades from surface readings and observed copper grades for 291 particles Results from the dataset of 291 particles, having a weighted average grade of 0.86% Cu, were used to simulate the sorting performance of a sorter. Grade and recovery curves in relation to sorter mass pull are presented in Figure 7.4. The figure shows that surface XRF readings can be y = x012345670 1 2 3 4 5 6 7Predicted Cu grade from surface pXRF, [%]pXRF Cu grade of pulverized rocks, [%]R2=0.79 108 used to recover 90% of the copper in 40 % of the material, which has a resulting grade of almost 2% Cu. Ideal grade-recovery data, determined using the copper content of each rock, shows that a recovery of approximately 92% is theoretically achievable for the same mass pull. Results from PGNAA and XRF test work confirmed that New Afton material is amenable to these technologies for sensor-based sorting. Figure 7.4 Particle sorting performance based on pXRF testing of 291 particles The XRF measurement time that is achievable during sorter operation is shorter than the 10 second measuring time that was used for portable XRF testing. Test data for a Steinert pXRF sorting unit from an independent study by Mazhary (2018) were analysed to observe the difference in sensing performance. The sorter tests were carried out on 94 particles that were conveyed individually past the XRF sensor and measured in millisecond time-scales. Following testing, each particle was pulverised and sub-sampled for copper grade measurement by ICP analysis. Equation 7.2 shows the resulting model for copper. 𝐶?̂? = 𝛽0. Cu − 𝛽1. 𝐶𝑢2 (7.2) 01234567891001020304050607080901000 10 20 30 40 50 60 70 80 90 100Concentrate Grade, Cu %Cu Recovery, %Mass Pull, %Recovery Ideal RecoveryConcentrate Grade, Cu % Ideal Grade109 Where β0 and β1 are fitted constants and Cu is the measured copper under dynamic conditions. Figure 7.5 shows a plot of predicted copper grades under dynamic conditions and observed copper grades. An R-squared of 0.76 was achieved, which is slightly lower than the R-squared value of 0.79 achieved with the portable XRF unit using longer measurement times for multiple surfaces. Figure 7.5 Predicted copper grades from surface readings taken under dynamic conditions using a sorter-mounted XRF and observed copper grades for 94 particles Grade and recovery curves in relation to sorter mass pull were calculated for sorter operation and are presented in Figure 7.6. Sorting performance estimates based on portable XRF testing and dynamic testing from Mazhary (2018) were similar. For example, at a mass pull of 30%, dynamic sorting tests showed a recovery of 81.5 % while using portable XRF test results a recovery of 80.3% was determined. y = x012345670 1 2 3 4 5 6 7Predicted Cu grade from sorter-mounted XRF, [%]Cu grade of pulverized rocks from ICP, [%]R2=0.76 110 Figure 7.6 Particle sorting performance from testing under dynamic conditions using a sorter-mounted XRF with 94 particles 7.4 Sorting System Evaluation The envisaged combination of bulk and particle sorting systems at the New Afton block cave mine is presented in Figure 7.7. All caved material, irrespective of grade, is conveyed to surface past a PGNAA bulk sensor. Based on the grade measurement, a mechanized diverter separates the material into high grade, intermediate grade and waste stockpiles. Low grade material that is transported on the underground-to-surface conveyor is dumped to a waste stockpile and high-grade material is sent directly to the mill. Intermediate grade material is fed to a double-deck screen with aperture sizes of 100 and 37.5 mm. Oversize (+100 mm) is crushed in closed-circuit with the screen and the middlings are sent to a coarse (-100, +37.5 mm) particle sorting unit. Screen undersize is conveyed to a single-deck screen with a cut-point of 12.5 mm. The oversize is processed by a fine (-37.5, +12.5 mm) particle sorting unit and the fines are either milled or dumped to a waste stockpile depending on 01234567891001020304050607080901000 10 20 30 40 50 60 70 80 90 100Concentrate Grade, Cu %Cu Recovery, %Mass Pull, %Recovery Ideal RecoveryConcentrate Grade, Cu % Ideal Grade111 the head grade measured by the bulk sensor at the time when it was conveyed from underground or depending on the measured grade should an additional PGNAA sensor be installed. Screening is carried out with wet sprayers to clean the surfaces of the rocks. Both coarse and fine particle sorting units are equipped with (1) XRF sensors that determine copper grades from surface readings, and (2) pneumatic ejectors, that send above cut-off grade material to a mill feed stockpile. The ratios of the maximum and minimum particle sizes for the coarse and fine units are within the guideline range of 3:1 for particle sorting; a guideline suggested by Salter and Wyatt (1991). Figure 7.7 Simplified flowsheet of the mine and sorting systems 7.5 XRF Particle-Sorting Models The amount of ore available for particle sorting depends on the proportion of fines (-12.5 mm) contained in the primary crushed ore. Sieving of primary crushed material was carried out as part of the fragmentation study described in Chapters 4 and 5. Results showed that approximately 112 34% of primary crushed material would pass a 12.5 mm screen. The proportion of fines was found to increase as the cave matures. Analysis of copper grades within different size fractions, presented in Section 6.4, showed that the grade of finer size fractions (-2 mm) was lower than the coarser size fractions of (-50, + 25 mm) and (-150, +50 mm). Based on the results it was assumed that within primary crushed material, the grade of the -12.5 mm size fraction was 75% of the head grade. Based on the assumptions for caved ore size and grade variation with size, the grade of ore sorter feed (+12.5 mm), is calculated by the following equation: 𝑓𝑠𝑜𝑟𝑡𝑒𝑟 𝑓𝑒𝑒𝑑 (%) =𝑓.(1−𝑓𝑖𝑛𝑒𝑠 𝑑𝑖𝑙𝑢𝑡𝑖𝑜𝑛.𝑃12.5𝑚𝑚)(1−𝑃12.5𝑚𝑚) (7.3) where f is the copper grade of caved material, P12.5 mm is the proportion of feed that passes a 12.5 mm screen and fines dilution is the grade of fines (-12.5 mm) as a percentage of head grade. The Constitutional Heterogeneity (CH) of underground drawpoint samples and a surface sample, which is a blend of material from underground drawpoints, were calculated based on pXRF surface readings for copper and pXRF readings of the same rocks after pulverizing using equation 7.4. 𝐶𝐻𝐿 = 𝑁𝐹 ∑ [(𝑎𝑖−𝑎𝐿)𝑀𝑖𝑎𝐿𝑀𝐿]2𝑖 (7.4) where NF is the number of particles in a lot, L, ai and Mi are the grade and mass of an individual particle within the lot, and aL and ML are the weighted-average grade and mass of the entire lot of particles. Figure 7.8 shows that the degree of grade heterogeneity, represented by CH, determined from surface readings is in-line with the grade heterogeneity determined from pulverized rocks. Therefore, pXRF surface measurements and rock masses could be used to determine the CH of each muckpile. The head grade of each muckpile sample was determined by crushing all collected rocks and fines, sub-sampling and carrying out ICP analysis. A comparison of CH and 113 muckpile grade is shown in Figure 7.9. Variations in copper grade were found to be higher for lower grade samples. Figure 7.8 Constitution heterogeneity based on surface readings of rocks and pulverized samples. Bubble widths are based on the number of rocks in each sample. Figure 7.9 Constitution heterogeneity from surface pXRF measurements and Cu grade of underground samples (determined from ICP analysis of all collected sample including fines). Bubble widths are based on the number of rocks in each sample 10 rocks19 rocks47 rocksy = x0.00.51.01.52.02.53.03.54.04.55.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0CH, Surface Cu pXRFCH, Pulverized Cu pXRFUnderground Samples Mill Feed Sample131 rocks100 rocks60 rocksy = 8.11e-1.19xR² = 0.730123456789100.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80CH, pXRF of Surface CuCu Grade of Bulk Sample from ICP Analysis, [%]Sample with picrite dilution 114 To simulate the performance of an XRF particle sorter operating with different head grades, the master data set for 291 rocks that had surface and pulverized sample measurements taken was separated into data sub-sets representing head grades ranging from 0.02 to 1.4 % Cu. When selecting rocks for a target heard grade, a grade-weighted random selection function was used in Excel. Objectives of the selection function were to select rocks such that (1) the weighted average grade of the rocks corresponded to the target head grade and (2) the head grade and CH were within 15% of the trend line included in Figure 7.9. Equation 7.3 was used to determine the head grade based on the weighted average grade of rocks and the fines dilution. For example, for a head grade of 0.22% Cu, surface and powder pXRF measurements for 218 rocks were randomly selected and the resulting CH was 6.8. The weighted average grade of all 218 rocks was 0.25% Cu. Equation 7.1 was then applied to determine the sorting performance achievable with the surface pXRF measurements. Particle sorting performance could then be simulated for a range of head grades while accounting for the heterogeneity occurring within lots of New Afton caved material. 7.6 Process and Cost Assumptions Table 7.3 shows the assumptions used to evaluate the sorting system. The mechanical sorting efficiency, which is defined as the efficiency of the ejection system to recover above-grade material that had been correctly identified by the XRF sensors, was assumed to be 95%. Milling and material handling costs were determined from demonstrated site cost data. The cost for screening, crushing and sorting was nominated to be $0.80 per tonne of screen feed. Lessard et al. (2016) assumed a cost of $0.10 to $0.13 for XRT particle sorting of pebbles from a semi-autogenous mill circuit. 115 Table 7.3 Cost and process assumptions Process Units Value Copper price $/lb 2.90 Gold price $/oz 1,290 Silver price $/oz 17 Ratio of Cu (%) to Au (g/t) - 0.81 Loading (screening plant cost) $/t 0.65 Milling cost $/t of mill feed 9 Waste disposal cost $/t 2.48 Screening, crushing & sorting cost $/per tonne of screen feed 0.80 Percentage passing 12.5 mm % 34 Mechanical sorting efficiency, effsort % 95 Grade of -12.5 mm material (fines dilution) % of head grade 75 The following flotation recovery model for copper in B1 and B2 hypogene ores was determined by New Afton metallurgists from site data and applies for mill process rates of 15,500 tonnes per day: 𝐶𝑢 𝐹𝑙𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 (%) = 89.1(−0.37𝑓𝑚𝑖𝑙𝑙 𝑓𝑒𝑒𝑑2 + 0.80fmill feed + 0.57) (7.5) where fmill feed is the copper feed grade in %. Recovery models based on historical mill data were also used for gold and silver recovery calculations. All cost and mill recovery assumptions were combined with the particle sorting model to determine the mass pull that provides the highest value for each head grade. For the case where material is screened, the oversize is processed by a particle sorter and the fines are processed by the mill, recovery of copper contained in screen feed is calculated by the following equation: 116 𝐶𝑜𝑝𝑝𝑒𝑟 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦, 𝑅 (%) = 𝑅𝑠𝑜𝑟𝑡 . 𝑒𝑓𝑓𝑠𝑜𝑟𝑡 . (1 − 𝑓𝑖𝑛𝑒𝑠 𝑑𝑖𝑙𝑢𝑡𝑖𝑜𝑛. 𝑃12.5𝑚𝑚) + 𝑃12.5𝑚𝑚. 𝑓𝑖𝑛𝑒𝑠 𝑑𝑖𝑙𝑢𝑡𝑖𝑜𝑛 (7.6) where Rsort is the proportion of copper in sorter feed that is recovered by the sorter and effsort is the efficiency of particle sorting systems downstream of the sensors. The proportion of ROM material that is directed to the mill is given by: ϒ (%) = ϒ𝑠𝑜𝑟𝑡𝑒𝑟 . (1 − 𝑃12.5𝑚𝑚) + 𝑃12.5𝑚𝑚 (7.7) where ϒ sorter is the proportion of sorter feed mass, or mass yield, that reports to sorter concentrate. The grade of mill feed is calculated as: 𝑓𝑚𝑖𝑙𝑙 𝑓𝑒𝑒𝑑 (%) =𝑅.𝑓ϒ (7.8) where f is the head grade of caved material. Similar calculations are used for the case where caved material is screened and the oversize is processed by a particle sorter while the fines are dumped to waste stockpiles. 𝑅𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑓𝑖𝑛𝑒𝑠 (%) = 𝑅𝑠𝑜𝑟𝑡 . 𝑒𝑓𝑓𝑠𝑜𝑟𝑡. (1 − 𝑓𝑖𝑛𝑒𝑠 𝑑𝑖𝑙𝑢𝑡𝑖𝑜𝑛. 𝑃12.5𝑚𝑚) (7.9) The mass pull, ϒ, of caved material that reports to sorter concentrate is: ϒ (%) = ϒ𝑠𝑜𝑟𝑡𝑒𝑟 . (1 − 𝑃12.5𝑚𝑚) (7.10) For both particle sorting scenarios, the value of recovered copper in copper concentrate is determined by the following equation: 𝐶𝑢 𝑣𝑎𝑙𝑢𝑒 ($𝑡) = 𝑓. 𝑅. 𝑅𝑓𝑙𝑜𝑡𝑎𝑡𝑖𝑜𝑛(%). 𝐶𝑢_𝑝𝑟𝑖𝑐𝑒 ($𝑡) (7.11) 117 where R is the copper recovery of the screening and sorting circuit which is determined from either Equation 7.6 or 7.9 (depending on whether fines are milled or dumped to waste), and Rflotation is the flotation copper recovery. Similarly, the gold and silver values are determined by applying their respective flotation recoveries and market values. The overall value per tonne of ore is: 𝑁𝑆𝑅 ($𝑡) = 𝐶𝑢 𝑣𝑎𝑙𝑢𝑒 + 𝐴𝑢 𝑣𝑎𝑙𝑢𝑒 + 𝐴𝑔 𝑣𝑎𝑙𝑢𝑒 − 𝑀𝑖𝑙𝑙𝑖𝑛𝑔 𝑐𝑜𝑠𝑡($𝑡). ϒ(%) − 𝑆𝑜𝑟𝑡𝑖𝑛𝑔 𝑐𝑜𝑠𝑡($𝑡) (7.12) An example of how the two sorting options compare for a feed grade of 0.22% Cu is shown in Figure 7.10. A mass pull of 31% provides the best value for both sorting options. The highest value of $6.52 per tonne was calculated for the option of particle sorting and processing of fines. This result represents an improvement over the value of $6.12 per tonne associated with direct milling. Figure 7.10 NSR and copper recovery in relation to mass pull for a material copper grade of 0.22% The modelling process was repeated for all sub-sets so that the optimum mass pull target (maximum sorting NSR) for a range of copper grades could be determined, which is shown in Figure 7.11. The corresponding sorter recoveries are included in the graph. 6.126.526.01012345670 20 40 60 80 100Net Smelter Return [$/t]Mass Pull (% of sorter feed)Sorting (fines milled) Sorting (fines dumped) No particle sortingNSR Lost 118 Figure 7.11 Economic mass pull and corresponding copper recovery in relation to material grade Models for economic sorter mass pull and the corresponding metal recovery were fitted to results using the following equations: 𝐸𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑆𝑜𝑟𝑡𝑒𝑟 𝑀𝑎𝑠𝑠 𝑃𝑢𝑙𝑙 = 40.04𝑒0.64𝑓𝑠𝑜𝑟𝑡𝑒𝑟 𝑓𝑒𝑒𝑑 − 40.04𝑒−3.00𝑓𝑠𝑜𝑟𝑡𝑒𝑟 𝑓𝑒𝑒𝑑 (7.13) 𝐸𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑅𝑆𝑜𝑟𝑡𝑒𝑟 = 88𝑒0.11𝑓𝑠𝑜𝑟𝑡𝑒𝑟 𝑓𝑒𝑒𝑑 − 88𝑒−15.09𝑓𝑠𝑜𝑟𝑡𝑒𝑟 𝑓𝑒𝑒𝑑 (7.14) By combining Equations 7.13 and 7.14 with equation 7.3, which relates the grade of caved material to the grade of sorter feed, sorting performance can be applied to a grade block model to determine the value associated with each processing option. 7.7 Application of Sorter Models to Grade Block Models For each block of ore within a grade block model, the value associated with direct milling (contained NSR), particle sorting with and without fines dumping (ore sorting NSR), versus 010203040506070809010001020304050607080901000.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40Sorter Mass Pul for Max NSR, [%]Cu Recovery, [%]Feed Grade,Cu [%]Mass Pull Recovery Mass Pull Model Recovery Model119 disposal as waste can be determined and compared using the calculations presented. Running production schedules for a nominated footprint allows the capacity requirements of sorting equipment to be estimated. Incorporating assumptions for contained gold and silver ratios, a comparison of ore value and feed grade is presented in Figure 7.12, which shows that for grades from 0.10 to 0.20% Cu, the highest value is achieved by particle sorting +12.5mm particles and sending -12.5 mm material to waste dumps. For ore grades of 0.20 to 0.36% Cu, the most economic option is to sort +12.5 mm material and process -12.5 mm in the mill. For grades above 0.36% Cu, the highest value is achieved by sending all material to the mill. At these grades, reductions in milling cost due to sorting are outweighed by the value of copper lost due to particle sorting inefficiencies. In caving operations, there are periods of operation where waste is caved and transported to surface due to: the requirement of achieving a suitably large footprint to ensure that the cave initiates and propagates, to access ore that is separated from the extraction level by blocks of waste and due to unplanned dilution. Once sub-economic material has been brought to surface, the value added by particle sorting may make it a preferable option to moving material to a waste dump. For this reason, mining costs have not been included in the value calculation in Figure 7.12. The graph shows that the cut-off grade for on-surface material is 0.11% Cu. 120 Figure 7.12 Net smelter return in relation to material grade for three sorting options. Grade ratios of Au (g/t): Cu (%) and Ag (g/t):Cu (%) of 0.82 and 2.6 assumed based on grade data 7.8 Discussion and Conclusions The sorting evaluation relied on key outcomes from the fragmentation and heterogeneity studies, and mill recovery models. An integrated approach, as proposed with Cave-to-Mill, is required for proper assessment of the potential to add value through implementation of sorting systems. The presented method for assessing the value of bulk and particle sorting systems was developed using material from an operating block cave mine. For block cave mines that are in the project development stage, drill core samples could be used in place of drawpoint samples. Grade-recovery results from alternative particle sorting systems, such as XRT-based sorters, can be modelled using similar calculation methods. Capacity requirements of particle sorting units, irrespective of sensing technology, would stand to be significantly reduced by the presence of the proposed bulk sorting system. 05101520253035400.00 0.10 0.20 0.30 0.40 0.50 0.60Net Smelter Return, [$/t]Feed Grade, Cu [%]Sorting (fines milled) Sorting (fines dumped) Directly Milled121 The value of sensing systems extends beyond the opportunity to sort ore according to grade. Both pXRF and PGNAA test results showed that picrite, a waste rock that negatively affects copper recovery in the flotation circuit, can be identified by elevated concentrations of nickel and chromium, shown in Appendix C. Additionally, PGNAA could be used to identify ore with native copper mineralization due to its distinct high copper to sulfur ratio, shown in Figure C.5. There is an opportunity to use information from the sensors in a feed-forward system to trigger adjustments to grinding and reagent addition for flotation or to direct material to alternative processing circuits. The installation of bulk sorting systems prompts reconsideration of the approach to nominating underground production schedules. Although production schedules are constrained by cave stresses, flow geometry and cave geometry, they do provide the possibility to blend ore types in a fashion that increases mill productivity. At many block cave mines, production schedules are nominated to provide a consistent grade of mill feed. This approach encourages mixing of grades in the crushing and material handling systems. However, in the case where a bulk sorting system is in place, production schedules should be nominated to minimize mixing so that high-grade and waste material can be campaigned to surface. This provides more opportunity to beneficiate the ore prior to the mill. When calculating the cost performance of the sorting systems, it was apparent that a significant cost benefit was the improvement in flotation recovery due to increases in mill feed grade. For a feed grade of 0.20% Cu, an improvement in ore value of $0.97 per tonne was calculated for the case where bulk and particle sorting systems are installed. Further evolutions of the approach involve implementation of bucket mounted grade sensors and tracking of load haul dump (scooptram) movements along the extraction level. Such an approach would provide opportunity to take advantage of the heterogeneity of caved material and improve mill productivity. To strengthen the assessment of XRF-based particle sorting, more test work is required to increase the database of 291 characterised particles for use in Net Smelter Return calculations. Bulk particle testing will provide opportunity to update the model to allow for sorting 122 inefficiencies that are associated with material presentation to the sorters and fluctuations in throughput. 123 Chapter 8: A Methodology for Cave Evaluation with Sensor-Based Sorting 8.1 Introduction The addition of sensor-based sorting to caving projects is envisaged to affect the size of the footprint and overall project value. This chapter presents a method for including sensor-based sorting parameters in the outputs of the PCBC cave planning and production scheduling software. Results from the sorting study carried out at the New Afton block cave mine were used to demonstrate the method for a conceptual cave in a copper-porphyry deposit. The method highlights the importance of integrating mine and mill parameters in the form of Cave-to-Mill, as inputs ranging from geotechnical to mill recovery parameters are referenced to carry out the cave evaluation. 8.2 Description of the Cave Evaluation Method Sensor-based sorting parameters, described in Chapter 7, were combined with the outputs of the PCBC cave planning and production scheduling module within the GEMS GEOVIA, version 6.8.1 package (Dassault Systèmes, 2018). A summary of the proposed method for determining the economic cave footprint and associated value for a grade block model is shown in Figure 8.1. The key steps of the proposed method include: • Assessing the applicability of bulk and/or particle grade sensors to material from the deposit, described in Chapters 6 and 7. • Relating sorting models to material head grade, presented in Chapter 7. • Populating the block model with sorting parameters for integration with the PCBC production scheduling module • Nominating the footprint geometry, orientation and elevation using Footprint Finder • Generating a mixed slice file. Updating the grade block model with mixed slice data using inverse distance weighting • Populating the updated grade block model with sorting parameters 124 • Calculating the Best Height of Draw (HOD) for each drawpoint and running production schedules to estimate project value, waste rejection and throughput rates for sorting and milling equipment Figure 8.1 Summary of approach to cave-sorting evaluation 8.3 Relating Sorting Performance to Head Grade To predict the properties of ore above a proposed extraction level layout, PCBC applies cave mixing models to generate a slice file, where rock mass data is arranged according to draw cones, representing the flow geometry of ore above drawpoints. Each draw cone is split into slices that contain grade and other additive information for the caved rock. The size of each slice is related to drawpoint spacing and the height of blocks within the grade block model. Therefore, 125 nominating block size is an important parameter that will affect the quality of sorting predictions. The block size should be nominated with careful consideration to geostatistical implications, which are outside the scope of this thesis, and the sorting unit size of the bulk sorting technology being assessed. In the case of establishing linkages between particle sorting performance and head grade, various drill core or drawpoint materials from existing caves within the deposit should be collected to cover a range of grades and lithology/oxidation types including waste rock. Sensor-based sorting tests for the relevant technology can then provide critical parameters, such as metal recovery and mass pull, to relate to head grade. During sorter operation, the mass pull of the sorting system can be controlled by modifying the cut-off grade (by way of changing a sensor threshold) for accepting and rejecting particles. The optimum mass pull for a sorting unit is a function of several parameters including metal value and costs for processing and material handling. The same cost and process assumptions as used for the sensor-based sorting study, presented in Table 7.3, were used for the case study. 8.4 Nominating Footprint Geometry, Orientation and Elevation Once sorter models are available as a function of head grade, they can be applied to the grade block model to define the block NSR, and the sorter mass pull and recovery for each block. To run a high-level comparison of extraction level sizes and elevations, the Footprint Finder function in the PCBC module provides the value of each unit area ($/m2) for a range of elevations. The outputs can be exported in spreadsheet form so that a footprint area, horizontal location and elevation that represents a high project NPV can be identified in external packages (e.g. Matlab or Excel). Practically, the footprint geometry is constrained by the caveability of the rock mass and the nominated drawpoint spacing and drift dimensions. An example of costing, caveability and design inputs is shown in Table 8.1. 126 Table 8.1 Cave design and costing parameters for the copper-gold case study Item Units Value Mining cost $/t 11 Development cost per footprint area $/m2 5,800 Drawpoint spacing, minor m 15 Drawpoint spacing, major m 20.9 Drift, width x height m 4.2 x 4.2 Vertical caving rate m/year 40 Minimum aspect ratio (length to width) [-] 3 Minimum hydraulic radius [m] 30 Development rate Drawpoints per month 8 Discount rate % 8 Based on sorting data and conceptual cave data, Figure 8.2 shows the footprint area and elevation that provides maximum project value while adhering to the constraints shown in Table 8.1. Footprints for various sorting flowsheets can be compared to the nominal footprint that has been solved assuming sorting systems are not implemented. From the figure it can be seen that a proportion of the footprint has lower value (black) and is only developed to adhere to geometric constraints for hydraulic radius and aspect ratio. To reduce the risk of cave stall, a minimum height of draw of 50 metres was assigned to all drawpoints within the footprint. The cost of processing material from marginal drawpoints is reduced by rejecting low grade material prior to the mill. 127 Figure 8.2 Plan view of a conceptual cave footprint for a proposed extraction level elevation and sorting scenario. Red datapoints refer to high-value areas of the footprint 8.5 Applying Models to Mixed Material in Draw Cones Since PCBC scheduling software, in its current form, does not allow for input of sorter models within the production scheduling routine, sorter models are applied in a three-step approach. 1. A slice file is generated for all drawpoints in the proposed layout. During slice file creation, mixing models are applied based on estimates for the height of the mixing horizon and draw cone diameter, both of which relate to fragmentation size 2. The grade block model is updated using inverse distance weighting of data from the mixed slice file 3. Sorter models and NPV calculations are applied to blocks within the updated grade block model, which now includes a more accurate representation of material that will report to drawpoints To prepare for production scheduling, a final slice file is generated with no mixing applied. Development Sequence 128 8.6 Production Scheduling and Evaluation The best height of draw, acting as a limit for drawpoint tonnage, is calculated for each drawpoint based on the final slice data, the draw strategy using the production rate curve (PRC) and draw cone profiles. Figure 8.3 shows an example of a production schedule for the case study. Typical run of mine production would follow the blue ‘Actual’ curve, whereas mill feed production rates following sorting would follow the ‘Mill Feed’ curve. Results show that approximately 21% of low grade material could be rejected by bulk and particle sorters prior to the mill, while achieving a metal recovery of 97%. Table 8.2 shows the corresponding mass balance, where it can be seen that the use of bulk sorting significantly reduces the capacity requirements of screening and particle sorting equipment. Figure 8.3 Example production schedule runs for a conceptual cave The mass balance shown in Table 8.2 provides valuable input for sizing of screens, sorters and mill equipment. - 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,0000 20 40 60 80 100 120Tonnes per dayMonthActual Cave limit Mill Feed129 Table 8.2 Mass Balance for a conceptual cave Stream Mass Proportion, % Metal Proportion, % Bulk Sorter Feed (ROM) 100 100 Bulk Sorter, High Grade Stream (to mill) 66.7 92.2 Particle Sorter Concentrate 4.1 4.6 Particle Sorting Plant, Fines (to mill) 8.4 0.8 Particle Sorter Waste 12.1 2.1 Bulk Sorter, Waste 8.7 0.3 Total Mill Feed 79.2 97.6 8.7 Discussion and Conclusions The nature of cave operations makes them ideal for incorporation of sensor-based systems to reject rock and allow the operation to become more dynamic. The presented method, which relies on Cave-to-Mill parameters ranging from geotechnical rock mass characterization to mill performance, provides a means to design and evaluate a block or panel cave for the case where sensor-based sorting systems are included in the flowsheet. The resulting mass balance can be used for stockpile estimation and equipment sizing to estimate the capital cost savings associated with lower mill throughput requirements. Using the method, changes in footprint size, project value and life of mine metal production can be compared to alternative scenarios where sorting systems are not in place. The implementation of bulk sorting systems on underground-to-surface conveyors or integrated with production shafts prompts reconsideration of underground production scheduling. For example, campaigning of low-grade and high-grade ore from drawpoints will reduce blending within orepasses and downstream material handling systems, and improve sorting performance. The value opportunity associated with sensor-based particle sorting is strongly related to fragmentation size. For proper evaluation of particle sorting systems at cave mines, the variable 130 nature of cave fragmentation and the deportment of mineralization to different size fractions within caved ore need to be considered. 131 Chapter 9: Conclusions and Recommendations 9.1 Main Outcomes An initial outcome of the research was the definition of a Cave-to-Mill approach targeting block cave productivity. Key aspects of cave planning and operation were identified in the form of a Cave-to-Mill flowchart, Figure 2.9, which showed how a coordinated effort can be made to maximize productivity in the form of Net Present Value (NPV). A central component is the ore block model, which is continuously refined during project development and mine operation through input of geotechnical, geological and metallurgical information. It is envisaged that the flowchart will be a valuable reference for cave project planning and operations. Bulk and Particle Sorting To address the lack of selectivity of the block cave mining method and improve productivity, bulk and particle sensor-based sorting systems were evaluated using New Afton ore. A configuration where bulk and particle systems are used in series was found to improve the Net Smelter Return (NSR) of caved material and reduce the number of particle sorters required. To assess the change in reserve size and cave value associated with sorting of caved ore, a method was developed to incorporate sensor-based sorting parameters in cave planning and production scheduling software (Chapter 8). This methodology can now be used to assess the value of future cave mines and caving lifts. Variations of bulk and particle sorting configurations can be incorporated with the method. The resulting mass balance can be used for stockpile estimation and equipment sizing to estimate the capital cost savings associated with lower mill throughput requirements. Evaluation of XRF-based particle sorters led to development of a method to relate feed grade to particle sorting performance based on grade heterogeneity (Section 7.5). The method is applicable to all mining methods and allows sorting parameters to be included in block models to assess the size of reserve and mine value when sorting systems are included in the flowsheet. 132 The research assessed the use of a PGNAA sensor for online grade measurement of New Afton caved material. Based on results of the work, New Afton mine installed a PGNAA sensor on the underground-to-surface conveyor in late 2017. In mid-2018, an upgraded diverter was installed on the conveyor to divert caved material at intervals of 30 seconds or more based on outputs of the PGNAA sensor. The installation is the first of its kind in the caving industry. Bulk sensing and sorting is expected to become an industry standard in caving operations. In mid-2018, a PGNAA sensor was installed at Newcrest’s Cadia East Block Cave Mine. An analysis of grade heterogeneity on a bulk scale showed that drawpoints located at the New Afton cave boundary were more suitable for separate handling of ore and waste underground (Section 6.3). To take full advantage of grade sensors mounted on load haul dumps (LHD), which transport material from drawpoints to orepasses, dedicated ore and waste material handling systems would be required. Ideally these would be located in close proximity to drawpoints that experience high variability in ore grade. The outcomes are of value for designing alternative extraction and haulage level designs which facilitate bulk sorting with mobile equipment. Cave Fragmentation and Mill Productivity Analysis of underground production scheduling and mill productivity (Section 5.2) identified drawpoints that provided finer mill feed and are associated with increased mill throughput. Coarser (+27 mm) size fractions of mill feed negatively correlated with simulated SAG mill throughput, while the proportion of material retained on sieves smaller than the SAG screen aperture size (13.2 mm) positively correlated with simulated throughput. Opportunities for improving cave productivity by providing finer feed to the mill were identified as being (1) adjustment of the closed-side setting of primary crushers to an operational minimum and (2) inclusion of fragmentation and hardness data in the decision making that goes into planning underground production schedules. Installation of secondary crushers ahead of existing SAG mill circuits also provides opportunity to improve mill throughput productivity at existing operations and reduce fluctuations in performance. For future caving projects, High Pressure Grinding Rolls (HPGR), which are increasingly popular alternatives to SAG mill-based circuits and are known to be less sensitive to variations in 133 ore size and hardness, were identified as a solution to reducing the impact of cave size fluctuations on mill throughput performance. The cave fragmentation study showed that secondary fragmentation at New Afton is sensitive to faulting, fracture intensity, height of draw (HOD) and boundary effects. Similar to the findings published by Brunton et al. (2016), BCF software was found to under predict the content of fines within caved ore. The DFN component of the fragmentation study by Munkhchuluun (2017) showed that fracture intensity P32 is relevant to fragmentation size and hang-up frequency. As found by Brzovic et al. (2016) at the El Teniente Mine, the largest reduction in hang-ups due to increases in height of draw was found during the earlier stages of drawpoint extraction. Based on outcomes of the fragmentation study, the following two-stage method for application of DFN-based fragmentation models was proposed: 1) repopulation of DFN models as access to exposed ore increases during project development 2) calibration with size measurements of caved ore and hang-up logs when the project moves towards production. In the case that macro blocks are planned for a deposit, calibrated fragmentation models for an initial macro block can be treated as valuable design and planning tools for future lifts. Rock Characterization The main test regimen characterized rocks in terms of the terminal limits of the study, starting at the in situ rock mass and terminating at the rougher flotation cells of the mill. Lithology and alteration types were described in terms of their impact on mine and mill performance. For New Afton material, relationships were found between geotechnical and metallurgical test results, and cave fragmentation size. Carrying out a similarly structured test regimen at the cave planning stage should allow for more efficient rock characterization and improved valuation of blocks within block models (Section 5.9 and Table 7.1). 134 9.2 Recommendations for Future Work Size Analysis Image-based size analysis was found to be practical for particle sizes coarser than 100 mm. The percentage passing of aperture sizes below 100 mm is of relevance to material handling systems, production potential of the mill and the performance of sensor-based sorting systems. It is recommended that further research be focused on developing practical methods to measure fines content in muckpiles to inform approaches to improve fragmentation and flow models. Fragmentation Models BCF software was found to under predict the content of -0.5 m material within caved ore. Similar findings were found at the Cadia Mine by Brunton et al. (2016). Improved methods for measuring -100 mm material, as described in the previous recommendation, would provide data for developing fragmentation models that more accurately describe fines fragmentation. Linking Sensor Responses to Cave-to-Mill Characteristics At New Afton, finer cave fragmentation size was associated with the content of fault and carbonate rocks within muckpiles. Since fault and carbonate rocks can be identified on the basis of point load strength, XRF response and Magnetic Susceptibility readings, information which is contained in the New Afton block model, fragmentation predictions can be made for future caves at the New Afton deposit. It is recommended that further research be focused on core characterization in terms of cave and mill performance. Efforts to recognize relationships between geotechnical and metallurgical parameters would help streamline core analyses. Design of Underground Bulk Sorting Systems Results from the heterogeneity study prompts further research in the area of extraction and haulage level design for underground bulk sorting. Configurations where grade sensors are used on mobile equipment, orepasses and conveyors could be explored. Trials of bucket-mounted grade sensors with a participating cave mine would confirm readiness of available sensor technologies to underground conditions. In addition to sorting, grade sensors on mobile equipment could be used for drawpoint grade control, as suggested by Ross (2012). 135 Relating In Situ Fracture Intensity to Secondary Fragmentation Size The DFN component of the fragmentation study by Munkhchuluun (2017) showed that fracture intensity P32 is relevant to fragmentation size and hang-up frequency. It is posited that fragmentation predictions would improve should DFN approaches be combined with secondary fragmentation models, similar to those proposed by Dorador (2016). Production Scheduling with Sensor-Based Sorting Parameters To refine the proposed approach for evaluation of cave mines working with sensor-based sorting systems, it is recommended that work be carried out to update algorithms within the PCBC module of GEMS Geovia software. The presented method did not allow sensor-based sorting equations to be applied to the results of more advanced mixing and flow algorithms, such as Cellular Automaton, contained within the software. An update of the software where block parameters relating to head grade are accepted, would allow sensor-based sorting performance to be calculated with the results of more advanced mixing models. 136 Chapter 10: Claims of Original Contributions A novel approach for planning and operating block cave mines, termed Cave-to-Mill, was developed with data and samples from an operating cave mine and presented in this thesis. The proposed integration of cave mine and mill systems is the first published work of its kind. The following are outcomes of the research that are original and significant to the field. Cave-to-Mill Flowchart A comprehensive flowchart, which focuses on cave mine productivity, is novel due to the emphasis on block model definition and progressive refinement with geological, caving and milling parameters as a caving project progresses from planning to cave maturation stages. Cave Planning and Evaluation with Sensor-Based Sorting A method was developed to incorporate sensor-based sorting parameters in cave planning and production scheduling software. This methodology can now be used to assess the value of future cave mines and caving lifts. Variations of bulk and particle sorting configurations can be incorporated with the method. The resulting mass balance can be used for stockpile estimation and equipment sizing to estimate the capital cost savings associated with lower mill throughput requirements. Particle Sorting Assessment Evaluation of XRF-based particle sorters led to development of a method to relate feed grade to particle sorting performance. The method uses grade heterogeneity determined from surface measurements to simulate sorter performance for a range of head grades. The method is applicable to all mining methods and allows sorting parameters to be included in block models to assess the size of the reserve and mine value when sorting systems are included in the flowsheet. Evaluation of Bulk and Particle Sorting Systems for a Cave Mine The assessment of bulk and particle sorting systems to determine the upgrade to Net Smelter Return (NSR) value, was the first published sorting study for a block cave mine, where bulk and particle sorting systems are included in the flowsheet. The study considered the impacts of cave 137 fragmentation size and the grade distribution of copper minerals within different size fractions on particle sorting performance. Improvements in flotation performance, as a result of higher mill feed grades, were also included in the value calculation. Cave-to-Mill Rock Characterization The characterization of New Afton rock samples in terms of grade, mine and mill performance, including cave fragmentation, is the first published work of its kind. Original contributions include determined relationships between cave fragmentation size, DropWeight strength, resistance to abrasion and point load strength. Cave Fragmentation Size and Mill Performance Analysis of mine and mill data showed that increased mill throughput was achieved when higher proportions of mucked material were sourced from one area of an extraction level at a cave mine. Simulations showed how the locations of mill throughput bottlenecks change according to the size and hardness of caved material being processed. Grade Heterogeneity The presented research on grade heterogeneity is the first published research showing grade heterogeneity at the extraction level of an operating cave mine. Results were shown with a focus on separation of ore and waste for separate material handling. Abbreviated DropWeight Test Method The author is aware of one other abbreviated DropWeight testing method for energy-breakage characterization of rock sample, however, it relies on unpublished methodologies and calculation methods. To address the need for a public domain method, an abbreviated DropWeight test was developed and presented in Appendix B. 138 References Alfaro, M., & Saavedra, J. (2004). Predictive models for gravitational flow. Proceedings of the Fourth International Conference and Exhibition on Mass Mining (MassMin 2004) Pg. 179 -184. AMTEL (2006). Deportment of gold in in composites of ore styles from New Afton Mine. AMTEL Report 06/30. September 11, 2006. Van As, A. (2010). Caving terminology and glossary. Rio Tinto Copper Projects. Vancouver, Canada. ASTM D5731-08. (2013). Standard test method for determination of the point load strength index of rock and application to rock strength classifications. West Conshohocken, PA. Bailey, C., Lane, G., Morrell, S., & Staples, P. (2009). What can go wrong in comminution circuit design. Tenth Mill Operator’s Conference, October 12 – 14, 2009. Adelaide, South Australia Balzan. L. (2016). New Afton copper porphyry ore test report. November 25, 2016. Scantech, South Australia. Bamber, A.S., Klein, B., Pakalnis, R., & Scoble, M. J. (2008). Integrated Mining, Poressing and Waste Disposal Systems for Reduced Energy and Operating Costs at Xstrata Nickel’s Sudbury Operations, Institute of Materials, Mining Technology 117, no. 3: 142-153 Barratt, D., & Doll, A. (2008). Testwork programs that deliver multiple data sets of comminution parameters for use in mine planning and project engineering Bbosa, L. S., Mainza, A. N., & Powell M. (2014). Analysis of the Breakage Data of a Series of Devices. Proceedings of the XXVII International Mineral Processing Congress – IMPC 2014. 139 Bergen RD, Krutzelmann H., & Rennie DW. (2015). New Gold Inc. NI 43-101 Technical report on the New Afton Mine. RPA Inc. Toronto, Canada. March 23. Bewick, R., P., & Kaiser, P. K. (2016). Characterisation and classification of veined rocks for rock mass behaviour assessment. Proceedings of the 7th International Conference and Exhibition on Mass Mining - MassMin 2016, Sydney, Australia, May 9 -11, 2016. Bond, F. (1962). More accurate grinding calculations. Society of Mining Engineers. Annual Meeting of the American Institute of Mining, Metallurgical and Petroleum Engineers, Inc., New York February 18 to 22, 1962 Brown E.T. (2003). Block Caving Geomechanics. Indooroopilly, Queensland: Julius Kruttschnitt Mineral Research Centre. Brunton I, Lett J.L., & Thornhill T. (2016). Fragmentation prediction and assessment at the Ridgeway Deeps and Cadia East Cave operations. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. p. 151–159. Brzovic A., Vallejos J., Arancibia E., Burgos L., Delonca A., Gaete M., & Barrera, V. (2016). Analysis and modelling of the fragmentation process – case studies at El Teniente Mine. Seventh International Conference and Exhibition on Mass Mining, May 9 – 11. Sydney, Australia. p. 305–310. Brzovic A., Schachter, P., de los Santos, C., Vallejos, J. A., & Mas Ivars, D. (2014). Characterization and synthetic simulations to determine rock mass behaviour at the El Teniente Mine, Chile. Part I. Proceedings of the Third International Symposium on Block and Sublevel Caving 2014, Santiago, Chile. Burger, B., McCafferey, K., Jankovic, A., Valery, W., & McGaffin, I. (2006). Batu Hijau model for throughput forecast, mining and milling optimisation and expansion studies, Advances in Comminution, S. K. Kawatra Ed, 461 – 479, SME 2006. 140 Bye, A. (2011). Case studies demonstrating value from geometallurgy initiatives, First AusIMM International Geometallurgy Conference, 9-30, Brisbane, The Australasian Institute of Mining and Metallurgy. Campbell, D., & Thurley, M. J. (2017). Application of laser scanning to measure fragmentation in underground mines, Mining Technology, 126:4, 240-247 Catalan, A., Onederra, I., & Chitombo, G. (2017). Evaluation of intensive preconditioning in block and panel caving – Part I, quantifying the effect on intact rock, Mining Technology, 126:4, 209-220 Cebeci, T., Wang, C., Huang, M., & Liu, S. (2017). Study on variable speed drives for ball mills and vertimills at the New Afton Mine. UBC Mining report. November 28, 2017. Vancouver, BC. Chitombo, G. P. (2010). Cave mining: 1 years after Laubscher’s 1994 paper, ‘Cave mining – state of the art’, Transactions of the Institutions of Mining and Metallurgy Section A: Mining Technology, Vol. 119, Issue. 3, 132 – 141. Coghill, P., Cutmore, N., Lehmann-Horn, J., Lovric, B., McEwen, A., Milinkovic, D., Miljak, D., Roberts, G., & Yong, R. (2018), ‘Demonstration of a magnetic resonance analyser for bulk copper sorting’, Proceedings of the Eighth Conference on Sensor-Based Sorting & Control, Aachen, pp. 152–160. Condori, P., Fischer, D., Winnett, J., & Makgatho, J. (2011). From open cast to block cave and the effects on the autogenous milling circuit at Palabora Mining Copper. The Fifth International Conference on Autogenous Grinding, Semi-autogenous Grinding and High Pressure Grinding Roll Technology, September 25th - September 28th, 2011. Vancouver, Canada. 141 Cundall. P., Mukundakrishnan B., & Lorig. L. (2000). REBOP (Rapid Emulator Based on PFC3D) Formulation and User’s Guide, Itasca Consulting Group, Inc., Report to the International Caving Study, ICG00-099-7-20 Dance, A. (2016). Downstream effects of block caving. Cave Mining Forum. Workshop Series II. October 14th, 2016. Vancouver, BC, Canada. Dassault Systèmes. (2018). Geovia GEMS PCBCTM. Version 6.8.1. Dassault Systèmes. Vancouver, Canada. https://www.3ds.com/products-services/geovia/products/pcbc/ Davies A. G. L. (2015). Case study: Understanding the mechanics behind the rockmass deformation observed in an extraction strike drive at New Gold’s New Afton Mine block cave operation, 49th US Rock Mechanics / Geomechanics Symposium. June 28 – July 1. San Francisco, CA, USA. Diering, T., Richter, O., & Villa, D. (2010), ‘Block cave production scheduling using PC-BC’, SME Annual Meeting & Exhibit, Feb. 28 – Mar. 3, 2010, Phoenix. Diering, T. (2012). Quadratic programming applications to block cave scheduling and cave management, Proceedings of the Sixth International Conference and Exhibition on Mass Mining (MassMin 2012), Sudbury, Canada. Doll. A., & Barratt. D. (2010). Choosing the right motors for your mills. VII International Mineral Processing Seminar. December 8 – 10, 2010. Santiago, Chile. Doll. A. (2016). Calculating DWi from a drop weight test result. SAGMilling.com. Accessed Nov, 29, 2017. Doll. A. (2015). Technical Memorandum to the UBC Mine 331 (Unit Operations) Clas) on the subject of SAG mill + ball mill circuit sizing. February 16, 2015. 142 Dorador, L. (2016). “Experimental Investigation of the Effect of Broken Ore Properties on Secondary Fragmentation during Block Caving.” Electronic Theses and Dissertation Duncan, G. (2016). Development of a protocol to determine the sorting potential of particulate ore material. Master’s dissertation. University of Johannesburg. Dunstan, G., & Popa. L. (2012). Innovative cave establishment practices at Ridgeway Deeps. Proceedings of the Sixth International Conference and Exhibition on Mass Mining (MassMin 2012), Sudbury, Canada. Elshkaki, A., Graedel, T.E., Ciacci, L., & Reck, B. (2016). Copper demand, supply, and associated energy use to 2050, Journal of Global Environmental Change, Vol. 39, 2016, Pages 305-315. Encina, V., Méndez, D., Caballero, C., & Osorio, H. (2010). New approach for rapid preparation of block caving mines, Mining Technology, Volume 119 Erdenebat, E. (2017). Study of New Afton ore heterogeneity and its amenability to sensor based sorting. Master’s dissertation. August 201 . Vancouver, BC, Canada Esen, S., Rosa, D. La, Dance, A., Valery, W., Jankovic, A. (2007). Integration and optimisation of blasting and comminution processes, EXPLO Conference, Wollongong, NSW, September 2007, 1 – 10, The Australasian Institute of Mining and Metallurgy. Esterhuizen G. S. (2005). BCF Version 3.05. A program to predict block cave fragmentation. Technical reference and user’s guide. Venetia, PA. Ernst & Young, (2017). Productivity in mining: now comes the hard part - a global survey. Ernst and Young Global Mining & Metals Network. 143 Flores, G. (2014). Future challenges and why cave mining must change, Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago Chile. Conference proceedings. June 5/6, 2014, Pg. 23 – 52. Foggiatto, B. (2017). Modelling and simulation approaches for exploiting multi-component characteristics of ores in mineral processing circuits. Dissertation. University of Queensland. Sustainable Minerals Institute. François-Bongarçon, D. and Gy, P. (2002). The most common error in applying ‘Gy’s Formula’ in the theory of mineral sampling, and the history of the liberation factor. The Journal of the South African Institute of Mining and Metallurgy. November/December 2002. Pg. 475 – 480. Frenzel, M. (2016). New continuous production system in block cave mining – the Rock Flow System. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. Fuentes, S., & Villegas, F. (2014). Block caving using macro blocks, Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago Chile. Conference proceedings. Pg. 211 – 216. Gantumur, J., Godard, R., Prince, F., & Pascoe, C. (2016). Development and evaluation processes of cave production schedules for the Resolution Copper Project. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. Gaudin, A. M., Meloy, T. P. (1962). Model and a comminution distribution equation for single fracture. Society of Mining Engineers of AIME 223 (March 1962): 40. Global Mining Standards and Guidelines Group. (2015). Determining the Bond efficiency of industrial grinding circuits. August 20, 2015. 144 Global Mining Standards and Guidelines Group. (2016). Methods to survey and sample grinding circuits for determining energy efficiency. April 28, 2016. Golder Associates. (2017). Technical Report for the Kemess Underground Project and Kemess East Project, British Columbia, Canada. July 12, 2017. Vancouver, BC, Canada Gómez, R., Castro, R.L., Casali, A. (2017). A Comminution Model for Secondary Fragmentation Assessment for Block Caving Rock Mech Rock Eng (2017) 03 July 2017, 50: 3073 Gy, P. (2004). Sampling of discrete materials – a new introduction to the theory of sampling I. Qualitative approach, Journal of Chemometrics and intelligent laboratory systems, 74 (2004) 7 – 24. Gy. P. (1995). Introduction to the theory of sampling I. Heterogeneity of a population of uncorrelated units. Trends in Analytical Chemistry, Volume 14, Issue 2, 1995, Pages 67-76. Van Haarlem, Y. (2017). Ultragold: Real-time gold-in slurry analysis with 100 PPB accuracy. Proceedings of the Conference of Metallurgists. Aug 27-30, 2017. Vancouver, Canada. Hoek, E. (2007) Practical Rock Engineering. Chapter on Rock mass classification. Imperial Metals. (2017). Annual Information Form for the Year Ended December 31, 2016. Dated March 29, 2017. International Copper Association (ICA). (2017). The long-term availability of copper. NY, USA. October, 2017. Ivars, D M, Pierce, M. E., Darcel, C., Reyes-Montes, J, Potyondy, D. O, Young, R. P., & Cundall, P. A. (2011). The synthetic rock mass approach for jointed rock mass modelling, International Journal of Rock Mechanics and Mining Sciences, Volume 48, Issue 2, 2011, Pages 219-244. 145 JKTech Pty Ltd. (2016). Procedure for the JK Drop Weight Test. Indooroopilly, Queensland. JKTech Pty Ltd. (2015). JKSimMet, Version 6.1. User Manual. JKTech Pty Ltd. Indooroopilly, Queensland. JKTech Pty Ltd. (2015). JKSimMet Version 6.1. Build 29. JKSimMet. Indooroopilly, Queensland. http://jktech.com.au/jksimmet Kanchibotla, S. S., Vizcarra, T.G., Musunuri, S. A. R., Tello, S., Hayes, A., & Moylan, T. (2015). Mine to Mill optimisation at Paddington Gold Operations. The 6th International Conference on Semi-Autogenous and High Pressure Grinding Technology, September 20-23. Vancouver, Canada. Khodayari F. & Pourrahimian, Y. (2014). Determination of the best height of draw in block cave sequence optimization, Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago Chile. Conference proceedings. June 5/6, 2014, Pg. 457 – 465. Klein, B., & Bamber, A. S. (2018). Chapter on mineral sorting. SME Mineral Processing and Extractive Metallurgy Handbook. Editors: Dunne, R., Young, C., Kawatra, S. K. Society for Mining, Metallurgy and Exploration. CO, USA. Published date: 2018. Klymowsky, R. (2009). HPGR – Accepted technology in mining. Mining Magazine Congress. Oct. 8 -9, 2009. Niagara-on-the-lake, Canada. Kobzev, A. (2014). History of Sensor-Based Sorting in CIS, Proceedings of the Sixth Conference on Sensor-Based Sorting, Aachen, pp. 39–48. Kobzev, A. (2018). Methodology for the evaluation of ore sorting potential: approach in CIS, Proceedings of the Eighth Conference on Sensor-Based Sorting & Control, Aachen, pp. 152–160. 146 Kojovic. T. (2016). HIT – a portable field device for rapid testing at site. 13th AUSIMM Mill Operator’s Conference. Perth, WA, 10–12 October, 2016. Kurth, H. (2017). Suitability of on-belt elemental analysis for real-time ore quality measurement and bulk sorting. Proceedings of the Conference of Metallurgists. Aug 27-30, 2017. Vancouver, Canada. La Rosa, D., Caron., K., Valery, W., Diaz, R., & Mamani, H. (2015). Blast fragmentation impacts on downstream processing at Goldfields Cerro Corona. 11th International Symposium on Rock Fragmentation by Blasting. August 24-26, 2015. Sydney, NSW. Australia. Laubscher D. H. (2000). A practical manual on block caving. International Caving Study. Julius Kruttschnitt Mineral Research Centre, Isles Road, Indooroopilly, Queensland 4068, Australia. Laubscher DH. 2011. Cave Mining. SME Mining Engineering Handbook, Third Edition. Lessard, J., Sweetser, W., Bartram, K., Figueroa, J., & McHugh, L. (2016). Bridging the gap: Understanding the economic impact of ore sorting on a mineral processing circuit. Minerals Engineering 91. P92-99. Lovitt M. B. & Degay, B. (2004). Single shot drawbell blasting with electronic detonators at Freeport. Proceedings of the thirtieth annual Conference on Explosives and Blasting Technique. International Society of Explosives Engineers. New Orleans, Louisiana USA Liu Y., Nadolski S., Elmo D., Klein B., & Scoble M. (2015). Use of digital imaging processing techniques to characterise block caving secondary fragmentation and implications for a proposed Cave-to-Mill approach. 49th US Rock Mechanics / Geomechanics Symposium. June 28 – July 1. San Francisco, CA, USA. Lotter, N. (2011). Modern process mineralogy: an integrated multi-disciplined approach to flowsheeting, Minerals Engineering, Vol. 24, Issue 12, Pg. 1229 – 1237. 147 Manca, L. (2013). Modern planning practices for cave mining, 3rd International Seminar on Mine Planning, Santiago Chile, 24/26 July 2013, Chile. Matthews. D., & du Toit, T. (2011). Real-time online analysis of iron ore, validation of material stockpiles and roll out for overall elemental balance as observed in the Khumani iron ore mines, South Africa. Iron ore Conference. July 11-13, 2011. Western Australia. Mazhary, A., & Klein, B. (2015). Heterogeneity of low-grade ores and amenability to sensor-based sorting. Proceedings of the 47th Annual Canadian Mineral Processors Operators Conference, Ottawa, Ontario, January 20-22, 2015. Mazhary, A. (2018). Sensor response data and assay results for New Afton ore testing with a Steinert sorter. Test data, Vancouver, British Columbia, January 5, 2018. Mckee, D. (2013). Understanding Mine to Mill. The Cooperative Research Centre for Optimising Resource Extraction (CRC ORE). Queensland, Australia. Mindat (2018). Definition of monzonite. https://www.mindat.org/glossary/monzonite. Accessed: November 11th, 2018 Moose Mountain Technical Services. (2017). NI43-101 Technical Report for the Kwanika Project. April, 28, 2017. Cranbrook, BC. Canada. Morrell, S. (2014). Using the SMC Test® to predict comminution circuit performance. https://www.smctesting.com/documents/Using_the_SMC_Test.pdf Accessed, June 6th. 2018. Morrell, S., & Valery, W. (2001). Influence of feed size on AG/SAG Mill performance. The Third International Conference on Autogenous and Semiautogenous Grinding Technology, September 30th - October 3rd, 2001. Vancouver, Canada. Moss, A. (2017) Personal Communication 148 Moss A. (2004). Caving and fragmentation at Palabora: Prediction to fragmentation. Fourth International Conference and Exhibition on Mass Mining, August 22 – 25. Santiago, Chile. p. 585 – 590. Motion Metrics Inernational Corp. (2017). Vancouver, Canada. [June, 15th, 2017] http://www.motionmetrics.com/portable/. PortaMetrics™ Portable Fragmentation analysis. Narendrula, R. (2004). Effect of fragmentation on productivity of loading equipment. Master’s dissertation. Laurentian University. Newcrest Mining Limited. (2012). Panel Session on Technology and Delivery, 2012. Oancea, D. (2013). Underground Block Caving: A Guide for Investors. Resource Investing News. April 16, 2013. Orozco, P. (2010). Operating Designs in Early Stage Panel Caving with Sizers, PhD Dissertation, University of Chile. Paredes, P., Gajardo, D., Leaño, T., Peñaloza, S., & Tramón, C. (2016). Simplifying the ore handling system through the use of current technology at Chuquicamata Underground Mine Project. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. Paredes, P. S., & Pineda, M. F. (2014). An analysis of the lateral dilution entry mechanisms in panel caving. Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago Chile. Pierce, M. (2010). A model for gravity flow of fragmented rock in block caving mines, PhD dissertation, The University of Queensland, Brisbane, Australia. 149 Rahal, D., Smith, M., van Hout, G., & von Johannides, A. (2003). The use of mixed integer linear programming for long-term scheduling in block caving mines, Application of Computers and Operations Research in the Mineral Industries, South African Institute of Mining and Metallurgy, pp. 123 -131. Robben, C. (2014). Characteristics of sensor-based sorting technology and implementation in mining. PhD thesis dissertation. RWTH Aachen, Germany. Rogers. S., Elmo, D., Webb, G., & Catalan, A. (2010). A DFN based approach to defining in situ, primary and secondary fragmentation distributions for the Cadia East panel cave project, Proceedings of the Second International Symposium on Block and Sublevel Caving, April 20/22 2010, Perth, Australia. Ross, I. (2012). Sampling in block cave mines. Sampling Conference. August 21 to 22, 2012. Perth, WA, Australia. Rowland, C. A. (1982). Selection of rod mills, ball mills, pebble mills and regrind mills. Ch. 23. Designa and Installation of Comminution Circuits. SME, pp. 393 – 438. Rule, C., Fouchee, R., & Stewart, W. (2015). Run of Mine Ore Upgrading – Proof of Concept Plant for XRF Ore Sorting, in Proceedings of the 2015 SAG Conference, Vancouver, BC Canada. Sagawa, Y., and Yamatomi, J. (2009). Technology of block caving mining method in Northparkes E48 Mine, Australia. Journal of MMIJ. The Mining and Minerals Processing Institute of Japan. Vol. 125. Pg. 409 – 419. Salter, J. D., & Wyatt, N. P. G. (1991). Sorting in the minerals industry: past, present and future. Minerals Engineering Journal. Vol.4, Nos 7 – 11, pp. 779-796. United Kingdom. 150 Sandvik Automine System – Rio Tinto Northparkes Mine. (2012). Maquinaria Magazine, Dec 21, 2012. Video presentation. Singer D. A. (2017). Future copper resources. Ore Geology Reviews. Vol. 86, June 2017. Pg. 271-279. Srikant A., Nicholas D. E., & Rachmad, L. (2004). Visual estimation of fragment size distributions in the DOZ block cave. Fourth International Conference and Exhibition on Mass Mining. August 22 – 25. Santiago, Chile. p. 286 – 290. Srikant, A. (2006). Fragment size measurement and estimation in the DOZ block cave. Mining Engineering, Volume 58, Issue 10. Stark, S., Perkins, T., & Napier-Munn, T. J. (2008). JK Drop Weigh Parameters – a statistical analysis of their accuracy and precision, and the effect on SAG mill comminution circuit simulation. Proceedings of the Metallurgical Plant Design and Operating Strategies Conference, Perth, WA, 18-19 August, 2008. Stewart, C. A., & Butcher, R. J. (2016). Block cave evaluation. Proceedings of the 7th International Conference and Exhibition on Mass Mining - MassMin 2016, Sydney, Australia, May 9 -11, 2016. Talu, S., Van As, A., Henry, R., Hilton, J., & Whiteman D. (2010). Installation of ‘Smart Markers’ to monitor Lift 2 North Extension ore flow behavior. Proceedings of the Second International Symposium on Block and Sublevel Caving, Australian Centre for Geomechanics, Perth. Teck. (2018). ‘Building a smarter shovel’, https://www.teck.com/news/stories/2018/building-a-smarter-shovel. Published March 29, 2018, Accessed May 5th, 2018. 151 Tetra Tech. (2016). KSM Prefeasibility study update and preliminary economic assessment. October 6th, 2016. Vancouver, BC. Canada. Thurley, M. J., Wimmer, M., & Nordqvist, A. (2015). Blast fragmentation measurement based on 3D imaging in sublevel caving draw-points and LHD buckets at LKAB Kiruna. 11th International Symposium on Rock Fragmentation by Blasting. 24 - 26 August, 2015. Sydney, Australia. Paper No. 161. Tokimatsu, K., Murakami, S., Adachi, T. Ryota I., Rieko Y., & Masahiro N. (2017). Long-term demand and supply of non-ferrous mineral resources by a mineral balance model. Mineral Economics (2017) 30: 193. https://doi.org/10.1007/s13563-017-0109-8. Tollenaar, R. N. (2008). Characterization of discrete fracture networks and their influence on caveability and fragmentation, Master of Applied Science Dissertation, The University of British Columbia, Vancouver, Canada. Tong, L., Khoshaba, B., Bamber, A., Klein, B. (2015). Correlation and regression analysis in the X-ray fluorescence sorting of a low grade copper ore. The 6th International Conference on Semi-Autogenous and High Pressure Grinding Technology, September 20-23. Vancouver, Canada. Whiteman D., Talu, S., Wilson, M., Watt, G., Van As, A. & Kuiper, P. (2016). Cave tracker flow monitoring system installation at Argyle Diamond Mine. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. Whiteman D. (2010). Smart Marker System – a new tool for measuring underground orebody flow in block and sublevel mines, Mining Technology, 119:3, 182-192. Wyman, R. A. (1966). The application of electronic sorting to minerals beneficiation. Mineral Processing Division of the Department of Energy, Mines and Resources. Mines Branch. Ottawa. July 1966. 152 Viera, E., & Diez, E. (2014). Analysis of hangup frequency in bloque 1-2, Esmeralda Sur Mine. Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago Chile. Conference proceedings. Villa, D., Farías, F. (2016). Remnant ore model generation and categorisation post-block caving extraction at Salvador Mine. Proceedings of the 7th International Conference and Exhibition on Mass Mining - MassMin 2016, Sydney, Australia, May 9 -11, 2016. WipWare Incorporated. (2017). WipFragTM, 2.7.28, North Bay, Ontario, www.wipware.com Woo, K., Eberhardt, E., Elmo, D., & Stead, D. (2013). Empirical investigation and characterization of surface subsidence related to block cave mining. International Journal of Rock Mechanics and Mining Scieences.Vol. 61. Pg. 31-42. de Wolfe, C. & I, Ross. (2016). Super caves – benefits, considerations and risks. Seventh International Conference and Exhibition on Mass Mining; May 9 – 11. Sydney, Australia. 153 Appendices Vertimill Operational Data 154 Figure A.1 Vertimill operating performance and calculated ball mill main motor requirements for equivalent duty based on a Bond ball mill work index of 18.4 kWh/t, Rowland’s (1982) efficiency factors for mill diameter (> 3.81 m) and low reduction ratio, and overall drivetrain efficiency of 0.93 0408012016020024028032036040004008001,2001,6002,0002,4002,8003,2003,6004,000February 10, 2016 February 15, 2016 February 20, 2016 February 25, 2016 March 1, 2016 March 6, 2016 March 11, 2016Feed and product 80% passing size, [µm]Main motor power draw, [kW] and throughput, [mtph]Fresh feed mtph Vertimill main motor power Calculated ball mill main motor powerBall mill cyclone overflow F80 Vertimill cyclone overflow P80155 An Abbreviated DropWeight Test Method for Lump Rock B.1 Introduction This technical note presents a method for generating impact breakage parameters using lump rock with lower sample and time commitments than the standard JKDropWeight test. The method does not rely on confidential calculation methods or privately-owned databases for calibration factors, which makes it a distinguishing feature from other abbreviated DropWeight test methods that are in use within industry. The results of DropWeight impact breakage tests are used in crusher and SAG mill models included in the JKSimMet simulation package. Critical test outcomes include A and b parameters, which describe the proportion of product that passes a tenth of the feed size for a specific energy input, and, an appearance function, which describes the shape of the product size distribution. Current versions of JKSimMet use the A and b breakage parameters with the assumption that particles of a different size will break in a similar way. The standard DropWeight test and JKSimMet simulation package has been recognized by industry as being sufficiently accurate to generate useful predictions of crushing and grinding circuit performance. B.2 Background The standard JKDropWeight test method involves testing of three energy levels and five size fractions ranging from 13.2 to 63 mm in size and specific energy inputs of 0.25 to 2.5 kWh/mt (Stark et al., 2008). Overall, 345 particles in different size classes are tested. One abbreviated method is the SMC test, where five energy levels are tested using 100 particles from the same size fraction; 20 particles per energy level (Morrell, 2012). The SMC method relies on confidential calculation methods or privately-owned databases for calibration factors. The standard DropWeight test has been reported to be accurate to ±3.6 per cent (Kojovic, 2016). Stark et al. (2008) analyzed the errors associated with the steps of standard DropWeight testing 156 and found that Axb can be estimated to an accuracy of ±2.68 at 95% confidence. The analysis showed that the dominant sources of error are attributed to particle selection and the test itself, which includes setting of the drop weight height and sieving of the products. Based on the analysis, a focus of designing the abbreviated test method was to reduce the error associated with particle selection and drop-weight height specification. B.3 Methodology The trade-off between accuracy and test time was considered when designing the test method for lump rock. Table B.1 shows the test methodology, where one size fraction is tested at five different energy levels. The size fraction tested is the median size used for the full DropWeight test, therefore no size correction is applied. As shown in the table, particles are grouped according to their mass and designated as light, medium and heavy. Eleven lump rock samples, representing different locations at the New Afton block cave mine, were tested using the presented method. Table B.1 Abbreviated DropWeight Test Regimen Target specific energy level, kWh/mt Size range, mm Particle weight grouping 2.5 -31.5, + 26.5 Heavy Medium Light 1.8 -31.5, + 26.5 Heavy Medium Light 1 -31.5, + 26.5 Heavy Medium Light 0.5 -31.5, + 26.5 Heavy Medium Light 0.25 -31.5, + 26.5 Heavy Medium Light 157 B.3.1 Particle selection according to mass The full DropWeight test involves random sub-sampling to generate 15 lots of particles for testing using a unique size and energy input. For each lot containing 10 to 30 particles, one drop-weight height and weight is assigned based on the target specific energy input. The natural variation in particle masses due to differences in density and shape results in individual particles being impacted at specific energies that are higher or lower than the setpoint. An example of the mass distribution of a set of particles is shown in Figure B.1. One approach to minimizing the error would be to adjust the drop-weight height according to the individual mass of each particle such that the target energy input is achieved. However, this would result in a test that would take too long to complete to be practical. A compromise of the two extremes is put forward, where particles are placed into three groups according to their masses. The range of particle masses accepted into each group is calculated such that the specific energy is a minimum, as shown in equation B.1. 𝑇𝑜𝑡𝑎𝑙 𝑀𝑎𝑠𝑠 𝑉𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 = ∑ (1𝑚𝑎𝑠𝑠ℎ𝑒𝑎𝑣𝑦̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅−1𝑚𝑎𝑠𝑠𝑖ℎ)2𝑛ℎ𝑖ℎ=1+ ∑ (𝟏𝒎𝒂𝒔𝒔𝒎𝒆𝒅𝒊𝒖𝒎̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅−𝟏𝒎𝒂𝒔𝒔𝒊𝒎)𝟐𝒏𝒎𝒊𝒎=𝟏+ ∑ (𝟏𝒎𝒂𝒔𝒔𝒍𝒊𝒈𝒉𝒕̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅−𝟏𝒎𝒂𝒔𝒔𝒊𝒍)𝟐𝒏𝒊𝒊𝒍=𝟏 (B.1) Where nh, nm and ni are the number of particles in the heavy, medium and light weight groupings. massheavy, massmedium and masslight are the average mass of individual particles in each respective group. massih, Massim and massil are individual masses of particles within each heavy, medium and light weight grouping. The calculation is constrained by the requirement of having integer values for the resulting number of particles in each group (as fractions of a particle can not be tested). Figure B.1 shows the resulting solution for sample #1. 158 Figure B.1 Distribution of particle masses for sample #1 and example of mass groupings for minimum error in specific energy input B.4 Comparison of Testing Methods After particles have been selected for breakage, the full DropWeight test method relies on the average particle mass to select the potential energy required to achieve the setpoint specific energy (Bbosa et al., 2014). Figures B.2 and B.3 show the distribution of specific energies in the case where thirty particles are randomly sampled and a drop-weight height and weight are set to target a specific energy consumption of 1 kWh/mt. For Figure B.2, the average particle mass was used to specify the drop-weight energy, whereas for Figure B.3, three dropweight setpoint energies were specified for application to heavy, medium and light particles as determined using the proposed method. The calculation was carried out for eleven samples, representing eleven spatial domains in a block cave mine. The box plots show that use of the proposed method reduces the deviation of 159 specific energy from the target of 1 kWh/mt. For both the standard and proposed method, it can be seen than the average specific energy for each sample is higher than the target of 1 kWh/mt. The calculation methods for both cases results in an average mass-weighted specific energy consumption that is equal to the setpoint. Figure B.2 Distribution of specific energies for 11 samples, where 30 particles are randomly sub-sampled as per the standard DropWeight test method. For each sample, the average particle mass was used to specify a drop-weight height which corresponds to a specific energy input of 1 kWh/mt 160 Figure B.3 Distribution of specific energies for 11 samples, where 30 particles are randomly sub-sampled as per standard DropWeight test method. For each sample, particles were grouped into heavy, medium and light weight groupings and treated separately at a target specific energy input of 1 kWh/mt The reduction in variation of specific energy input was assessed using random sampling algorithms in Matlab 2018a. For each of the 11 geometallurgical samples, 150 particles were weighed and the process of sub-sampling particles into five lots of thirty, representing a sample for an energy-breakage setpoint, was simulated. The mean squared error was calculated assuming a setpoint of 1 kWh/mt. To simulate a typical DropWeight test approach, a case was evaluated where each lot of 30 particles is impacted using the one drop-weight potential energy, which is based on the setpoint of 1 kWh/mt and the average particle mass. To simulate the proposed abbreviated method, a case was evaluated where each sample set of 150 particles is separated into heavy, medium and light particle groupings according to equation B.1. Particles from heavy, medium and light mass groupings are then randomly sub-sampled to generate five lots of 30 particles (while keeping heavy, medium and light particles separate). For each sample containing 30 particles, three drop-weight potential energies are applied based on their mass grouping, as shown in Table B.1. The drop-weight potential energy to be used is calculated using the setpoint of 1 kWh/mt and the average particle mass in the heavy, medium and light groups. 161 Although a typical test approach would involve testing of different specific energy levels, to illustrate the error associated with the two test methods, a specific energy input of 1 kWh/mt was assumed for all five sub-samples (containing 30 particles). The simulation was repeated a thousand times to determine the distribution of mean squared error associated with each method. Figures B.4 and B.5 show the resulting mean squared errors for one thousand simulations of sampling based on the standard approach and the proposed method, respectively. The median of the error in specific energy consumption for the standard method ranged from 0.097 to 0.156 for the eleven samples. For the proposed method, the median of the error ranged from 0.011 to 0.020. Overall, the mean squared error of specific energy consumption decreased by approximately 85%. Figure B.4 Mean squared error of specific energy for 11 samples using a typical DropWeight approach where 30 particles are sub-sampled. The mean particle mass was used to specify a drop-weight potential energy which corresponds to a specific energy input of 1 kWh/t 162 Figure B.5 Mean squared error of specific energy for 11 samples using the proposed DropWeight method, where heavy, medium and light weight particles are randomly sub-sampled to generate a 30 particle sub-sample. Table B.2 shows a comparison of the Axb values resulting from the standard DropWeight test procedure and the proposed abbreviated method. For the two samples, the difference in DWI was less than 5%. Table B.2 Comparison of Axb results Sample Standard Method Abbreviated Method A b Axb DWI (kWh/m3) A* b Axb DWI (kWh/m3) Copper Mountain #1 62.7 0.47 29.5 9.52 80 0.381 30.5 9.20 Copper Mountain #2 62.4 0.48 30.0 9.59 80 0.377 30.1 9.53 * A was limited to 80% during fitting 163 B.5 Conclusions and Recommendations The abbreviated DropWeight method allows energy-breakage parameters A and b to be determined for material characterization using less sample than the full DropWeight method. The variation in specific energy was reduced by systematically separating particles into heavy, medium and light groups, and adjusting the dropweight height according to the mean of their masses. The difference in DWI results, from the standard DropWeight procedure and the abbreviated method, was within 5% for the two samples that were used for the final comparison. References Bbosa, L. S., Mainza, A. N. & Powell M. (2014). Analysis of the breakage data of a series of devices. Proceedings of the XXVII International Mineral Processing Congress – IMPC 2014. Kojovic. T. (2016). HIT – a portable field device for rapid testing at site. 13th AUSIMM Mill Operator’s Conference. Perth, WA, 10–12 October, 2016. Stark, S., Perkins, T., & Napier-Munn, T. J. (2008). JK Drop Weight Parameters – a statistical analysis of their accuracy and precision, and the effect on SAG mill comminution circuit simulation. Proceedings of the Metallurgical Plant Design and Operating Strategies Conference, Perth, WA, 18-19 August, 2008. 164 PGNAA and XRF Test Results Figure C.1 Rock type and pXRF measurement for Ni. Maximum of four readings. 1,226 rocks in presented dataset. Figure C.2 Rock type pXRF surface measurement for Cr. Maximum of four surface readings. 1,226 rocks in presented dataset. 165 Table C.1 Bulk samples used for PGNAA testing. Sample weights were approximately 220kg each. Mill feed was hypogene with minor amounts of secondary hypogene alteration Sample Abbreviation Copper grade (ICP), (%) Calibration/Blind Mill feed, low Grade Blend, LG 0.56 Calibration Mill feed, medium Grade Blend, MG 0.68 Blind Mill feed, high Grade Blend, HG 0.89 Calibration Monzonite Monzonite 0.23 Blind Picrite Picrite 0.31 Blind Hypogene Hypogene 1.55 Calibration Surface Supergene Surface Supergene 0.27 Calibration Underground Supergene UG Supergene 0.68 Calibration Hypogene “Waste” Hypogene Waste 0.19 Calibration Figure C.3 PGNAA and ICP measurements for Ni. Nine bulk samples of approximately 220kg each 050100150200250300350Ni, (ppm)PGNAA ICP166 Figure C.4 PGNAA and ICP measurements for Cr. Nine bulk samples of approximately 220kg each Figure C.5 Ratio of copper to sulfur from PGNAA measurements. Nine bulk samples of approximately 220kg each 050100150200250300350400450Cri, (ppm)PGNAA ICP012345678Ratio of copper to sulfur167 Figure C.6 Copper grade of drawpoint sample, based on ICP analysis of a representative sub-sample, and copper grade of fines (-2 mm), measured using pXRF Figure C.7 Fe content in fines (-2 mm), measured by pXRF, and the proportion of carbonate rock in drawpoint samples 0.000.200.400.600.801.001.201.401.600.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60ICP of rep-split of bulk sample. Cu (%)pXRF of fines (-2mm), Cu (%) R² = 0.8201234567890 1 2 3 4 5 6 7 8 9 10Fe content in fines (-2 mm), [%]Proportion of identified carbonates in drawpoint sample [%]Samples from boundary (H-Wall & F-Wall) 168 Point Load Test Results Point load tests were carried out according ASTM D5731-08 (2013) standards for lump rock. Lithology and alteration data are reported according to guideline shown in Table 3.1. Based on discussion with New Afton representatives, the following conversion factors were applied to Is(50) results to estimate an equivalent UCS value: Hypogene, Mesogene and DMZ (demineralized): 17 Supergene ore: 15 Note: blank values indicate an unsuccessful test result due to breakage occurring that does not meet ASTM guidelines (eg. chipping etc.) 169 Table D.1 Point load test results for mill feed sample taken in March 2016 Specimen # Size Lithology/ Alteration Group Magnetic (Yes = 1, No = 0) Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] Mill #1 -100,+75 1 1 2.81 58 63 37 22.6 2,732 52.3 8.3 1.0 8.4 Mill #2 -100,+75 1 0 2.75 87 88 29 30.5 3,212 56.7 9.5 1.1 10.0 Mill #3 -100,+75 1 0 2.76 101 80 25 18.6 2,546 50.5 7.3 1.0 7.3 Mill #4 -100,+75 1 0 2.67 88 91 36 12.0 4,034 63.5 3.0 1.1 3.3 Mill #5 -100,+75 1 0 2.77 97 75 39 24.1 3,724 61.0 6.5 1.1 7.1 Mill #6 -100,+75 1 0 2.87 66 93 27 9.7 2,269 47.6 4.3 1.0 4.2 Mill #7 -100,+75 1 0 2.73 89 56 29 12.7 2,068 45.5 6.1 1.0 5.9 Mill #8 -100,+75 1 0 2.74 73 61 57 13.7 4,427 66.5 3.1 1.1 3.5 Mill #9 -100,+75 1 0 2.68 54 74 42 15.2 2,888 53.7 5.3 1.0 5.5 Mill #10 -100,+75 1 0 2.69 91 96 55 16.9 6,373 79.8 2.6 1.2 3.3 Mill #11 -100,+75 1 0 2.88 64 57 41 32.5 2,976 54.5 10.9 1.0 11.4 Mill #12 -100,+75 1 0 2.73 90 83 34 28.5 3,593 59.9 7.9 1.1 8.6 Mill #13 -100,+75 3 0 2.72 77 50 50 16.4 3,183 56.4 5.2 1.1 5.5 Mill #14 -100,+75 3 0 2.67 54 92 43 12.0 2,956 54.4 4.1 1.0 4.2 Mill #15 -100,+75 3 0 2.64 89 45 47 16.3 2,693 51.9 6.0 1.0 6.1 Mill #16 -100,+75 3 0 2.66 71 67 55 10.7 4,692 68.5 2.3 1.2 2.6 Mill #17 -100,+75 3 0 2.68 86 84 58 15.2 6,203 78.8 2.4 1.2 3.0 Mill #18 -100,+75 3 1 2.72 54 89 49.5 11.1 3,403 58.3 3.3 1.1 3.5 Mill #19 -100,+75 3 1 2.76 53 68 54.5 34.4 3,678 60.6 9.4 1.1 10.2 Mill #20 -100,+75 3 0 2.78 42 79 50 4.8 2,674 51.7 1.8 1.0 1.8 Mill #21 -100,+75 3 0 2.68 89 71 51 16.0 4,610 67.9 3.5 1.1 4.0 Mill #22 -100,+75 3 1 2.70 82 81 41 15.0 4,228 65.0 3.6 1.1 4.0 Mill #23 -100,+75 3 0 2.69 65 68 49 23.9 4,055 63.7 5.9 1.1 6.6 Mill #24 -100,+75 3 0 2.77 53 69 37 13.0 2,497 50.0 5.2 1.0 5.2 Mill #25 -100,+75 7 0 2.88 78 95 70 14.1 6,952 83.4 2.0 1.3 2.6 Mill #26 -100,+75 6 0 2.85 84 69 73 27.6 6,413 80.1 4.3 1.2 5.3 Mill #27 -100,+75 6 0 2.66 41 55 39 12.9 2,036 45.1 6.4 1.0 6.1 170 Specimen # Size Lithology/ Alteration Group Magnetic (Yes = 1, No = 0) Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] Mill #28 -100,+75 5 0 2.62 103 50 Mill #29 -100,+75 5 0 2.78 53 78 70 39.6 4,724 68.7 8.4 1.2 9.7 Mill #30 -100,+75 5 0 2.69 62 83 59 33.4 4,658 68.2 7.2 1.2 8.2 Mill #31 -100,+75 5 0 2.70 87 79 Mill #32 -100,+75 5 0 2.65 78 80 65 32.5 6,455 80.3 5.0 1.2 6.2 Mill #33 -100,+75 5 1 2.76 79 86 Mill #34 -100,+75 5 0 2.74 78 83 65 39.1 6,455 80.3 6.1 1.2 7.5 Mill #35 -100,+75 5 0 2.74 77 40 Mill #36 -100,+75 5 0 2.71 86 65 74 16.0 6,124 78.3 2.6 1.2 3.2 Mill #37 -100,+75 1 0 2.74 103 94 45 21.5 5,386 73.4 4.0 1.2 4.7 Mill #38 -100,+75 1 1 3.03 79 49 48 20.2 2,995 54.7 6.7 1.0 7.0 Mill #39 -100,+75 1 0 2.74 72 84 62 18.3 5,684 75.4 3.2 1.2 3.9 Mill #40 -100,+75 1 1 2.82 61 75 54 23.7 4,194 64.8 5.7 1.1 6.4 Mill #41 -100,+75 1 0 2.77 90 90 46.5 40.3 5,329 73.0 7.6 1.2 9.0 Mill #42 -100,+75 1 1 2.75 100 63 51 11.0 4,091 64.0 2.7 1.1 3.0 Mill #43 -100,+75 1 1 2.70 62 30 45.5 17.0 1,738 41.7 9.8 0.9 9.0 Mill #44 -100,+75 1 0 2.75 78 74 35 28.3 3,298 57.4 8.6 1.1 9.1 Mill #45 -100,+75 1 0 2.82 65 55 59 22.5 4,132 64.3 5.4 1.1 6.1 Mill #46 -100,+75 1 1 2.87 81 86 38 20.6 3,919 62.6 5.3 1.1 5.8 Mill #47 -75, +63 7 - - 78 74 29 11.6 2,732 52.3 4.3 1.0 4.3 Mill #48 -75, +63 7 - - 46 53 49.5 15.1 2,899 53.8 5.2 1.0 5.4 Mill #49 -75, +63 7 - - 66 46 52 19.5 3,046 55.2 6.4 1.0 6.7 Mill #50 -75, +63 7 - - 84 55 48 8.3 3,361 58.0 2.5 1.1 2.6 Mill #51 -75, +63 7 - - 65 44 37 13.1 2,073 45.5 6.3 1.0 6.1 Mill #52 -75, +63 7 - - 49 58 44 39.4 2,745 52.4 14.4 1.0 14.7 Mill #53 -75, +63 7 - - 51 58 44 22.0 2,857 53.5 7.7 1.0 7.9 Mill #54 -75, +63 6 - - 64 64 33 11.3 2,689 51.9 4.2 1.0 4.3 Mill #55 -75, +63 6 - - 56 59 30 5.3 2,139 46.2 2.5 1.0 2.4 171 Specimen # Size Lithology/ Alteration Group Magnetic (Yes = 1, No = 0) Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] Mill #56 -75, +63 6 - - 74 57 37 15.9 2,685 51.8 5.9 1.0 6.0 Mill #57 -75, +63 6 - - 69 43 33 8.6 1,807 42.5 4.8 0.9 4.4 Mill #58 -75, +63 6 - - 75 44 53 12.1 2,969 54.5 4.1 1.0 4.2 Mill #59 -75, +63 6 - - 70 50 37 14.3 2,355 48.5 6.1 1.0 6.0 Mill #60 -75, +63 6 - - 61 64 44 9.2 3,417 58.5 2.7 1.1 2.9 Mill #61 -75, +63 1 - - 81 59 29 19.6 2,179 46.7 9.0 1.0 8.7 Mill #62 -75, +63 1 - - 67 57 27 19.1 1,960 44.3 9.8 0.9 9.2 Mill #63 -75, +63 1 - - 63 42 45 15.9 2,406 49.1 6.6 1.0 6.6 Mill #64 -75, +63 1 - - 57 63 30 23.7 2,177 46.7 10.9 1.0 10.6 Mill #65 -75, +63 1 - - 58 57 37 15.1 2,685 51.8 5.6 1.0 5.7 Mill #66 -75, +63 1 - - 54 67 43 23.9 2,956 54.4 8.1 1.0 8.4 Mill #67 -75, +63 3 - - 50 74 31 20.9 1,974 44.4 10.6 0.9 10.0 Mill #68 -75, +63 3 - - 63 63 29 11.9 2,326 48.2 5.1 1.0 5.0 Mill #69 -75, +63 3 - - 58 81 35 15.3 2,585 50.8 5.9 1.0 5.9 Mill #70 -75, +63 3 - - 61 56 45 31.8 3,209 56.6 9.9 1.1 10.5 Mill #71 -75, +63 3 - - 54 75 30 20.9 2,063 45.4 10.1 1.0 9.7 Mill #72 -75, +63 3 - - 57 50 38 21.1 2,419 49.2 8.7 1.0 8.7 Mill #73 -75, +63 5 - - 64 58 47 14.8 3,471 58.9 4.2 1.1 4.6 Mill #74 -75, +63 5 - - 65 65 36 22.5 2,979 54.6 7.6 1.0 7.9 Mill #75 -75, +63 5 - - 62 60 44 18.3 3,361 58.0 5.4 1.1 5.8 Mill #76 -75, +63 5 - - 62 56 45 15.7 3,209 56.6 4.9 1.1 5.2 Mill #77 -75, +63 5 - - 57 52 47 18.2 3,112 55.8 5.8 1.1 6.1 Mill #78 -75, +63 5 - - 81 80 29 17.4 2,954 54.3 5.9 1.0 6.1 Mean Is(50): 6.2 MPa Median Is(50) 6.0 MPa Standard deviation 2.5 MPa Equivalent UCS (NX 54mm Core) 106 MPa 172 Figure D.1 Point load test results for Mill Feed (sample collected March 2016) y = 0.0051x1.010.0100.01,000 10,000 100,000Load P, [kN]De2 [mm2]173 Table D.2 Point load test results for drawpoint sample B13S taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 2 2.70 57 67 67 15.6 5,289 72.7 3.0 1.2 3.5 4 2.66 62 70 63 14.5 5,294 72.8 2.7 1.2 3.2 8 2.72 9 2.55 104 99 84 1.0 10,856 104.2 0.1 1.4 0.1 11 2.63 76 77 36 5.4 3,507 59.2 1.5 1.1 1.7 12 2.76 27 81 72 10.4 4,950 70.4 2.1 1.2 2.4 13 2.68 82 88 14 8.5 1,515 38.9 5.6 0.9 5.0 14 2.61 23 67 73 3.5 4,183 64.7 0.8 1.1 0.9 16 2.75 54 54 33 6.4 2,269 47.6 2.8 1.0 2.8 19 2.73 36 53 57 19.4 3,230 56.8 6.0 1.1 6.4 21 2.70 49 70 35 4.2 2,652 51.5 1.6 1.0 1.6 22 2.86 85 101 41 12.8 4,855 69.7 2.6 1.2 3.1 23 2.65 117 83 46 4.6 5,857 76.5 0.8 1.2 0.9 24 2.72 60 42 63 3.6 4,091 64.0 0.9 1.1 1.0 25 2.63 35 56 36 4.0 2,086 45.7 1.9 1.0 1.9 26 2.80 41 50 34 16.2 1,970 44.4 8.2 0.9 7.8 29 2.77 31 2.71 81 82 58 3.2 6,019 77.6 0.5 1.2 0.6 32 2.68 63 56 53 3.8 4,015 63.4 0.9 1.1 1.0 33 2.58 47 46 44 4.4 2,605 51.0 1.7 1.0 1.7 36 2.72 60 55 40 16.5 2,928 54.1 5.6 1.0 5.9 39 2.71 72 72 28 17.6 2,567 50.7 6.8 1.0 6.9 40 2.84 44 50 33 21.3 1,975 44.4 10.8 0.9 10.3 41 2.72 26 61 62 9.1 3,434 58.6 2.6 1.1 2.8 42 2.74 51 65 31 13.8 2,289 47.8 6.0 1.0 5.9 43 2.76 45 53 48 11.4 2,995 54.7 3.8 1.0 4.0 46 2.71 83 84 52 32.8 5,528 74.4 5.9 1.2 7.1 174 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 49 2.73 23 61 60 26.5 3,209 56.6 8.3 1.1 8.7 50 2.73 46 63 47 19.7 3,261 57.1 6.0 1.1 6.4 51 2.77 55 63 66 8.6 4,958 70.4 1.7 1.2 2.0 52 2.65 48 44 36 8.3 2,108 45.9 3.9 1.0 3.8 B13S Mean Is(50): 3.8 MPa B13S Median Is(50) 3.1 MPa Standard deviation 2.7 MPa Equivalent UCS (NX 54mm Core): 64 MPa 175 Figure D.2 Point load test results for drawpoint sample B13S taken Nov 2015 y = 0.002247x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]176 Table D.3 Point load test results for drawpoint sample D7S taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 1 2.81 95 91 76 25.0 8,999 94.9 2.8 1.3 3.7 2 2.71 59 83 84 13.4 7,594 87.1 1.8 1.3 2.3 3 2.82 87 76 67 43.8 6,953 83.4 6.3 1.3 7.9 4 2.73 22 85 56 6.5 3,815 61.8 1.7 1.1 1.9 5 2.82 18 82 55 27.4 3,501 59.2 7.8 1.1 8.5 7 2.82 13 45 42 28.3 1,551 39.4 18.3 0.9 16.4 8 2.69 50 45 48 20.7 2,903 53.9 7.1 1.0 7.4 10 2.79 51 112 63 35.8 6,537 80.9 5.5 1.2 6.8 13 2.67 40 82 43 17.8 3,340 57.8 5.3 1.1 5.7 14 2.73 78 84 22 7.2 2,269 47.6 3.2 1.0 3.1 15 2.83 32 56 44 18.9 2,465 49.6 7.7 1.0 7.6 17 2.74 47 43 30 5.3 1,719 41.5 3.1 0.9 2.8 18 2.76 67 53 42 19.8 3,209 56.6 6.2 1.1 6.5 19 2.72 48 52 42 12.0 2,674 51.7 4.5 1.0 4.6 21 2.67 55 43 35 30.4 2,184 46.7 13.9 1.0 13.5 22 2.80 53 57 24 10.0 1,681 41.0 5.9 0.9 5.4 23 2.68 40 78 74 31.5 5,559 74.6 5.7 1.2 6.8 24 2.73 42 54 49 29.4 2,995 54.7 9.8 1.0 10.2 25 2.74 52 44 25 13.2 1,528 39.1 8.6 0.9 7.7 26 2.64 50 52 36 7.1 2,338 48.3 3.0 1.0 3.0 27 2.76 24 50 29 6.5 1,366 37.0 4.8 0.9 4.2 30 2.71 25 48 40 16.5 1,859 43.1 8.9 0.9 8.3 31 2.73 29 82 36 13.6 2,544 50.4 5.4 1.0 5.4 32 2.81 54 74 48 4.5 3,911 62.5 1.1 1.1 1.3 33 2.66 56 42 45 11.3 2,807 53.0 4.0 1.0 4.1 34 2.82 41 59 47 13.2 2,992 54.7 4.4 1.0 4.6 35 2.82 31 42 41 6.0 1,905 43.7 3.2 0.9 3.0 177 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 36 2.81 57 61 41 5.1 3,080 55.5 1.7 1.0 1.7 37 2.73 32 60 41 22.8 2,401 49.0 9.5 1.0 9.4 38 2.89 61 58 39 5.5 2,955 54.4 1.9 1.0 1.9 39 2.64 54 40 37 4.2 2,214 47.1 1.9 1.0 1.8 43 2.61 30 54 40 3.4 2,139 46.2 1.6 1.0 1.5 44 2.82 41 49 36 3.1 2,063 45.4 1.5 1.0 1.5 45 2.63 47 59 35 12.7 2,362 48.6 5.4 1.0 5.3 46 2.64 26 43 29 8.8 1,274 35.7 6.9 0.9 5.9 47 2.72 57 52 30 11.0 2,082 45.6 5.3 1.0 5.1 48 2.64 57 51 40 7.5 2,750 52.4 2.7 1.0 2.8 51 2.68 35 56 36 22.0 2,086 45.7 10.5 1.0 10.1 54 2.83 44 67 40 22.1 2,827 53.2 7.8 1.0 8.1 55 2.69 40 40 40 6.0 2,037 45.1 3.0 1.0 2.8 56 2.80 34 53 53 4.9 2,935 54.2 1.7 1.0 1.7 58 2.74 46 39 49 8.5 2,652 51.5 3.2 1.0 3.2 59 2.79 54 46 38 10.5 2,419 49.2 4.3 1.0 4.3 64 2.88 43 73 40 25.3 2,954 54.3 8.6 1.0 8.9 66 2.69 50 59 50 15.1 3,470 58.9 4.3 1.1 4.7 D7S Mean Is(50): 5.4 MPa D7S Median Is(50) 4.7 MPa Standard deviation 3.3 MPa Equivalent UCS (NX 54mm Core): 92 MPa 178 Figure D.3 Point load test results for drawpoint sample D7S taken Nov 2015 y = 0.004504x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]179 Table D.4 Point load test results for drawpoint sample D11N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 2 2.79 54 54 34 3.9 2,338 48.3 1.7 1.0 1.7 3 2.72 68 81 57 41.6 4,935 70.3 8.4 1.2 9.8 5 2.72 44 68 48 40.4 2,689 51.9 15.0 1.0 15.3 7 2.73 76 38 43 22.1 2,080 45.6 10.6 1.0 10.2 8 2.81 58 81 46 24.9 3,397 58.3 7.3 1.1 7.9 9 2.75 45 70 61 22.9 3,495 59.1 6.6 1.1 7.1 10 2.75 54 48 29 3.5 1,772 42.1 2.0 0.9 1.9 11 2.61 104 77 48 14.7 4,706 68.6 3.1 1.2 3.6 12 2.66 43 48 32 20.7 1,752 41.9 11.8 0.9 10.9 13 2.69 56 47 33 10.9 1,975 44.4 5.5 0.9 5.2 14 2.65 85 93 26 16.0 2,814 53.0 5.7 1.0 5.8 15 2.68 80 90 39 5.4 3,973 63.0 1.4 1.1 1.5 16 2.85 93 61 45 20.8 3,495 59.1 6.0 1.1 6.4 17 2.69 60 50 53 10.3 3,374 58.1 3.1 1.1 3.3 18 2.65 19 2.69 54 35 30 5.8 1,337 36.6 4.4 0.9 3.8 22 2.67 83 63 54 15.4 4,332 65.8 3.5 1.1 4.0 23 2.73 83 84 74 18.0 7,820 88.4 2.3 1.3 3.0 25 2.70 69 70 22 14.9 1,933 44.0 7.7 0.9 7.3 26 2.75 47 45 35 8.3 2,005 44.8 4.2 1.0 4.0 28 2.68 103 70 59 25.0 5,258 72.5 4.8 1.2 5.6 29 2.95 80 77 45 30.1 4,412 66.4 6.8 1.1 7.7 30 2.77 52 45 47 22.1 2,693 51.9 8.2 1.0 8.4 32 2.63 40 45 23 5.3 1,171 34.2 4.5 0.8 3.8 33 2.68 44 63 38 15.7 2,129 46.1 7.4 1.0 7.1 34 2.67 36 2.75 55 58 38 22.6 2,661 51.6 8.5 1.0 8.6 180 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 37 2.72 52 56 42 17.2 2,781 52.7 6.2 1.0 6.3 41 2.70 53 54 38.4 15.3 2,591 50.9 5.9 1.0 5.9 42 2.57 59 45 28 10.2 1,604 40.1 6.3 0.9 5.7 43 2.68 81 70 43 11.4 3,832 61.9 3.0 1.1 3.3 44 2.68 53 45 29 9.6 1,662 40.8 5.8 0.9 5.3 48 2.68 29 53 34 6.0 1,255 35.4 4.8 0.9 4.1 49 2.58 42 36 44 10.4 2,185 46.7 4.8 1.0 4.6 50 2.67 45 55 39 8.0 2,483 49.8 3.2 1.0 3.2 52 2.79 58 58 49 16.4 3,619 60.2 4.5 1.1 4.9 54 2.64 43 70 38 11.8 2,734 52.3 4.3 1.0 4.4 56 2.79 61 55 49 11.5 3,619 60.2 3.2 1.1 3.4 57 2.57 46 47 26 5.9 1,539 39.2 3.8 0.9 3.4 58 2.72 51 52 47 17.7 3,082 55.5 5.7 1.0 6.0 59 2.69 36 44 31 18.0 1,579 39.7 11.4 0.9 10.3 60 2.68 74 88 45 13.7 4,641 68.1 2.9 1.1 3.4 61 2.68 68 59 37 17.0 2,991 54.7 5.7 1.0 5.9 62 2.73 54 53 51 15.4 3,474 58.9 4.4 1.1 4.8 D11N Mean Is(50): 5.7 MPa D11N Median Is(50) 5.3 MPa Standard deviation 2.8 MPa Equivalent UCS (NX 54mm Core): 97 MPa 181 Figure D.4 Point load test results for drawpoint sample D11N taken Nov 2015 y = 0.004855x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]182 Table D.5 Point load test results for drawpoint sample D11S taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 3 2.70 53 37 37 6.0 1,743 41.8 3.5 0.9 3.2 4 2.71 81 49 35 7.3 2,184 46.7 3.3 1.0 3.2 5 2.77 63 54 49 12.9 3,369 58.0 3.8 1.1 4.1 7 2.78 76 45 58.5 9.1 3,352 57.9 2.7 1.1 2.9 8 2.69 53 40 41 13.0 2,088 45.7 6.2 1.0 6.0 11 2.73 79 75 62 26.5 5,921 76.9 4.5 1.2 5.4 12 2.61 38 73 42 12.9 2,032 45.1 6.4 1.0 6.1 13 2.69 37 34 33.7 4.8 1,459 38.2 3.3 0.9 2.9 14 2.66 57 51 27 6.1 1,753 41.9 3.5 0.9 3.2 15 2.78 35 44 36.1 20.6 1,609 40.1 12.8 0.9 11.6 16 2.64 54 59 39 10.9 2,681 51.8 4.1 1.0 4.1 18 2.71 45 60 65 8.8 3,724 61.0 2.4 1.1 2.6 19 2.70 40 59 40 3.4 2,037 45.1 1.7 1.0 1.6 21 2.71 42 53 30 13.7 1,604 40.1 8.5 0.9 7.7 23 2.69 71 63 32 19.8 2,567 50.7 7.7 1.0 7.8 24 2.74 45 65 66 6.5 3,782 61.5 1.7 1.1 1.9 25 2.73 42 35 30 10.4 1,337 36.6 7.8 0.9 6.8 26 2.71 27 2.71 66 55 39.6 16.2 2,773 52.7 5.9 1.0 6.0 28 2.74 58 59 50 9.5 3,692 60.8 2.6 1.1 2.8 29 2.74 71 74 42 7.9 3,797 61.6 2.1 1.1 2.3 30 2.70 77 47 57 17.8 3,411 58.4 5.2 1.1 5.6 31 2.76 49 71 31 15.4 1,934 44.0 8.0 0.9 7.5 32 2.75 75 80 51 8.2 4,870 69.8 1.7 1.2 1.9 33 2.74 72 78 49 10.2 4,492 67.0 2.3 1.1 2.6 34 2.70 38 35 35 20.2 1,560 39.5 12.9 0.9 11.6 37 2.76 71 71 66 15.1 5,966 77.2 2.5 1.2 3.1 183 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 40 2.85 55 73 47 23.5 3,291 57.4 7.1 1.1 7.6 42 3.27 59 42 23 7.7 1,230 35.1 6.3 0.9 5.3 43 2.87 24 61 29 14.9 886 29.8 16.8 0.8 13.3 44 2.75 35 41 46 45 2.74 77 65 59.3 13.8 4,908 70.1 2.8 1.2 3.3 46 2.76 70 76 45.4 7.9 4,046 63.6 2.0 1.1 2.2 48 2.74 59 55 54.3 32.3 3,803 61.7 8.5 1.1 9.3 49 2.73 57 45 35 5.3 2,005 44.8 2.6 1.0 2.5 54 2.76 45 49 25.4 8.4 1,455 38.1 5.8 0.9 5.1 55 2.74 79 50 42 59 2.72 35 33 33.2 11.6 1,395 37.3 8.3 0.9 7.3 61 2.76 47 63 46 7.1 2,753 52.5 2.6 1.0 2.6 62 2.71 37 47 34.5 6.7 1,625 40.3 4.1 0.9 3.8 63 2.73 41 51 85 63.2 22.5 4,104 64.1 5.5 1.1 6.1 D11S Mean Is(50): 5.1 MPa D11S Median Is(50) 4.1 MPa Standard deviation 2.9 MPa Equivalent UCS (NX 54mm Core): 86 MPa 184 Figure D.5 Point load test results for drawpoint sample D11S taken Nov 2015 y = 0.003894x1.010.0100.01,000 10,000 100,000P [kN]De2 [mm2]185 Table D.6 Point load test results for drawpoint sample D38N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 1 2.71 53 60 31 5.3 2,230 47.2 2.4 1.0 2.3 2 2.69 47 61 35 2.3 2,406 49.1 0.9 1.0 0.9 3 2.66 75 66 36 5.2 3,231 56.8 1.6 1.1 1.7 4 2.72 29 63 33 3.5 1,933 44.0 1.8 0.9 1.7 5 2.60 56 39 38 3.1 2,298 47.9 1.3 1.0 1.3 6 2.63 44 53 55 3.5 3,396 58.3 1.0 1.1 1.1 9 2.74 48 51 29 13.0 1,828 42.8 7.1 0.9 6.6 11 2.79 18 57 37 4.0 1,767 42.0 2.3 0.9 2.1 13 2.75 64 60 53 13.0 4,184 64.7 3.1 1.1 3.5 14 2.70 67 45 27 3.4 1,925 43.9 1.8 0.9 1.7 15 2.66 75 55 29 3.6 2,400 49.0 1.5 1.0 1.5 16 2.79 38 44 34 6.5 1,775 42.1 3.7 0.9 3.4 18 2.77 90 85 44.2 10.4 4,924 70.2 2.1 1.2 2.5 20 2.69 36 65 48 4.6 3,074 55.4 1.5 1.0 1.6 21 2.76 60 40 38 6.2 2,419 49.2 2.6 1.0 2.5 26 2.72 105 77 50 16.5 5,793 76.1 2.8 1.2 3.4 28 2.78 43 45 29 7.2 1,625 40.3 4.4 0.9 4.0 30 2.69 55 45 39 8.4 2,483 49.8 3.4 1.0 3.4 32 2.81 64 20 34 3.4 1,818 42.6 1.9 0.9 1.7 33 2.72 55 29 50 5.6 2,674 51.7 2.1 1.0 2.1 34 2.83 56 53 38 7.2 2,637 51.4 2.7 1.0 2.8 35 2.78 51 20 40 2.1 1,808 42.5 1.1 0.9 1.1 36 2.76 48 50 35 5.2 2,184 46.7 2.4 1.0 2.3 37 2.70 75 29 27 7.9 1,788 42.3 4.4 0.9 4.1 39 2.70 50 29 44 3.1 2,213 47.0 1.4 1.0 1.4 41 2.65 44 49 34 0.9 2,013 44.9 0.4 1.0 0.4 43 2.53 56 45 33 3.0 2,122 46.1 1.4 1.0 1.3 186 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 44 2.70 57 54 34 5.1 2,403 49.0 2.1 1.0 2.1 47 2.73 46 19 31 0.6 1,283 35.8 0.5 0.9 0.4 49 2.71 61 68 28 2.4 2,299 48.0 1.0 1.0 1.0 D38N Mean Is(50): 2.2 MPa D38N Median Is(50) 1.9 MPa Standard deviation 1.3 MPa Equivalent UCS (NX 54mm Core): 37 MPa 187 Figure D.6 Point load test results for drawpoint sample D38N taken Nov 2015 y = 0.002245x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]188 Table D.7 Point load test results for drawpoint sample E13S taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 2 3.15 60 78 61.4 23.4 5,386 73.4 4.3 1.2 5.2 4 2.80 68 72 25.2 11.5 2,240 47.3 5.2 1.0 5.0 6 2.82 62 60 46.5 17.7 3,612 60.1 4.9 1.1 5.3 7 2.76 61 60 36.6 7.9 2,819 53.1 2.8 1.0 2.9 8 2.76 56 22 9 2.75 57 58 47.4 12.8 3,470 58.9 3.7 1.1 4.0 10 2.76 68 60 63 14.8 5,134 71.6 2.9 1.2 3.4 11 2.74 54 49 40 10.5 2,623 51.2 4.0 1.0 4.0 13 2.68 71 66 49.8 26.9 4,343 65.9 6.2 1.1 7.0 14 2.73 78 54 62.5 17.5 5,252 72.5 3.3 1.2 3.9 15 2.63 64 61 35.9 9.4 2,857 53.4 3.3 1.0 3.4 18 2.78 50 51 32 13.5 2,058 45.4 6.6 1.0 6.3 19 2.69 38 61 35.8 11.1 2,256 47.5 4.9 1.0 4.8 20 2.76 59 63 35.2 24.7 2,734 52.3 9.1 1.0 9.2 22 2.66 58 17 47.9 4.5 2,287 47.8 2.0 1.0 1.9 23 2.69 80 70 50.5 10.0 4,822 69.4 2.1 1.2 2.4 24 2.71 52 76 45 17.9 3,667 60.6 4.9 1.1 5.3 25 2.61 65 63 34.9 14.1 2,844 53.3 5.0 1.0 5.1 26 2.86 21 53 48.5 7.4 2,285 47.8 3.2 1.0 3.2 27 2.63 38 61 50 13.5 3,151 56.1 4.3 1.1 4.5 30 2.68 68 79 65 8.6 6,083 78.0 1.4 1.2 1.7 31 2.71 60 56 36 22.4 2,659 51.6 8.4 1.0 8.5 32 2.68 56 17 34 10.9 1,580 39.8 6.9 0.9 6.2 35 2.92 100 90 85.3 22.4 10,318 101.6 2.2 1.4 3.0 36 2.68 63 82 70 19.7 6,462 80.4 3.1 1.2 3.8 40 2.69 79 64 41.3 15.5 3,760 61.3 4.1 1.1 4.5 41 2.71 43 70 42.4 18.6 3,050 55.2 6.1 1.0 6.4 189 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 42 2.67 58 61 47.3 15.3 3,583 59.9 4.3 1.1 4.6 43 2.68 33 71 37.7 13.4 2,496 50.0 5.4 1.0 5.4 44 2.73 67 60 52.4 20.1 4,230 65.0 4.7 1.1 5.3 45 2.73 66 22 47 2.78 49 50 35.7 10.4 2,250 47.4 4.6 1.0 4.5 48 2.67 68 91 41.4 22.2 4,191 64.7 5.3 1.1 5.9 49 2.64 71 84 35.9 9.5 3,542 59.5 2.7 1.1 2.9 50 2.69 46 69 33 23.3 2,416 49.2 9.7 1.0 9.6 51 2.65 52 60 42.7 10.4 3,045 55.2 3.4 1.0 3.6 52 2.73 44 48 42 4.5 2,460 49.6 1.8 1.0 1.8 53 2.68 51 20 35 22.3 1,582 39.8 14.1 0.9 12.7 54 2.75 31 54 38 20.4 2,056 45.3 9.9 1.0 9.5 56 2.61 53 18 39.1 11.1 1,767 42.0 6.3 0.9 5.8 57 2.83 46 52 32.4 20.0 2,021 45.0 9.9 1.0 9.4 58 2.80 22 45 59 2.66 32 50 41.5 10.8 2,174 46.6 5.0 1.0 4.8 60 2.72 58 61 57 10.4 4,318 65.7 2.4 1.1 2.7 62 2.61 52 41 54 17.7 3,197 56.5 5.5 1.1 5.8 63 2.66 53 32 37.2 8.1 2,013 44.9 4.0 1.0 3.8 64 2.82 34 38 23 5.9 1,054 32.5 5.6 0.8 4.6 66 2.74 54 43 31.3 21.6 1,933 44.0 11.2 0.9 10.5 68 2.69 43 54 39.1 9.4 2,415 49.1 3.9 1.0 3.8 69 2.76 47 37 43 6.8 2,299 48.0 3.0 1.0 2.9 70 3.01 65 21 29.4 11.5 1,610 40.1 7.2 0.9 6.5 71 2.78 19 57 31 13.0 1,500 38.7 8.7 0.9 7.7 73 2.61 32 30 44.2 18.9 1,742 41.7 10.9 0.9 10.0 74 2.81 32 56 50.6 9.6 2,835 53.2 3.4 1.0 3.5 76 2.73 42 33 37 12.7 1,767 42.0 7.2 0.9 6.7 190 E13S Mean Is(50): 5.3 MPa E13S Median Is(50) 4.8 MPa Standard deviation 2.5 MPa Equivalent UCS (NX 54mm Core): 90 MPa Figure D.7 Point load test results for drawpoint sample E13S taken Nov 2015 y = 0.003900x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]191 Table D.8 Point load test results for drawpoint sample E15N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 1 2.71 75 79 68 36.6 6,667 81.6 5.5 1.2 6.8 2 2.67 73 58 59 5 2.68 58 50 27 9.1 1,856 43.1 4.9 0.9 4.6 6 2.69 27 46 44 20.0 2,045 45.2 9.8 1.0 9.3 7 2.70 52 64 24 18.6 1,772 42.1 10.5 0.9 9.7 8 2.80 45 39 30 18.2 1,604 40.1 11.3 0.9 10.2 9 2.81 34 53 38 20.4 2,105 45.9 9.7 1.0 9.3 10 2.82 42 30 30 12.5 1,375 37.1 9.1 0.9 8.0 11 2.63 55 47 31 6.1 2,013 44.9 3.1 1.0 2.9 13 2.74 51 59 34 17.6 2,381 48.8 7.4 1.0 7.3 14 2.75 42 43 65 13.4 3,517 59.3 3.8 1.1 4.1 15 2.67 65 69 46 30.4 3,924 62.6 7.7 1.1 8.6 16 2.70 34 64 50 12.2 3,119 55.9 3.9 1.1 4.1 17 2.70 86 78 50 19.5 5,220 72.3 3.7 1.2 4.4 18 2.66 30 53 50 7.2 2,642 51.4 2.7 1.0 2.8 20 2.68 50 64 26 3.7 1,887 43.4 2.0 0.9 1.8 22 2.66 53 59 40 17.5 2,852 53.4 6.1 1.0 6.3 23 2.70 50 71 35 6.8 2,696 51.9 2.5 1.0 2.6 24 2.85 55 66 51 19.9 3,929 62.7 5.1 1.1 5.6 25 2.80 48 56 30 10.3 1,986 44.6 5.2 0.9 4.9 27 2.68 88 77 73 11.7 7,668 87.6 1.5 1.3 2.0 31 2.67 50 57 24 10.4 1,635 40.4 6.4 0.9 5.8 32 2.75 37 56 43 13.9 2,546 50.5 5.5 1.0 5.5 33 2.83 50 51 45 14.7 2,893 53.8 5.1 1.0 5.3 34 2.74 46 52 33 12.2 2,059 45.4 5.9 1.0 5.7 35 2.77 81 90 56 27.8 6,096 78.1 4.6 1.2 5.6 36 2.64 33 49 38 9.8 1,984 44.5 5.0 0.9 4.7 192 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 37 2.73 57 63 41 17.2 3,132 56.0 5.5 1.1 5.8 38 2.70 37 50 31 5.4 1,717 41.4 3.2 0.9 2.9 41 2.74 37 48 52 9.9 2,814 53.0 3.5 1.0 3.6 43 2.75 16 64 36 4.8 1,833 42.8 2.6 0.9 2.4 44 2.60 50 52 27 11.4 1,753 41.9 6.5 0.9 6.0 45 2.69 51 55 29 13.2 1,957 44.2 6.7 0.9 6.4 46 2.83 59 63 36 18.3 2,796 52.9 6.6 1.0 6.7 48 2.74 39 75 49 14.5 3,556 59.6 4.1 1.1 4.4 49 2.69 44 52 32 14.0 1,956 44.2 7.2 0.9 6.8 50 2.70 36 55 35 26.1 2,028 45.0 12.9 1.0 12.3 51 2.66 50 32 37 16.4 1,932 43.9 8.5 0.9 8.0 52 2.69 43 56 47 12.1 2,962 54.4 4.1 1.0 4.2 53 2.68 57 65 30 13.0 2,330 48.3 5.6 1.0 5.5 54 2.70 53 58 24 13.7 1,696 41.2 8.1 0.9 7.4 55 2.69 44 54 29 9.1 1,809 42.5 5.0 0.9 4.7 57 2.66 72 74 57 17.5 5,298 72.8 3.3 1.2 3.9 58 2.76 63 60 27 11.8 2,114 46.0 5.6 1.0 5.4 59 2.73 38 37 32 17.7 1,528 39.1 11.6 0.9 10.4 62 2.65 34 55 38 14.3 2,153 46.4 6.6 1.0 6.4 63 2.71 45 46 43 8.2 2,491 49.9 3.3 1.0 3.3 64 2.75 22 48 50 14.8 2,228 47.2 6.7 1.0 6.5 68 2.79 67 63 28 14.0 2,317 48.1 6.1 1.0 6.0 E15N Mean Is(50): 5.8 MPa E15N Median Is(50) 5.6 MPa Standard deviation 2.3 MPa Equivalent UCS (NX 54mm Core): 98 MPa 193 Figure D.8 Point load test results for drawpoint sample E15N taken Nov 2015 y = 0.004727x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]194 Table D.9 Point load test results for drawpoint sample E23N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 2 2.74 102 100 62 20.9 7,973 89.3 2.6 1.3 3.4 4 2.74 39 108 44 28.8 4,118 64.2 7.0 1.1 7.8 5 2.71 51 53 31 19.5 2,052 45.3 9.5 1.0 9.1 7 2.72 39 39 33 12.4 1,639 40.5 7.5 0.9 6.9 9 2.73 37 68 42 11.6 2,807 53.0 4.1 1.0 4.2 10 2.71 31 69 29 13.8 1,846 43.0 7.5 0.9 7.0 11 2.75 61 61 51 29.7 3,961 62.9 7.5 1.1 8.3 12 2.67 68 66 45 11.1 3,839 62.0 2.9 1.1 3.2 13 2.64 74 41 34 15.0 2,489 49.9 6.0 1.0 6.0 14 2.70 60 16 40 15.0 1,935 44.0 7.7 0.9 7.3 15 2.71 62 26 24 15.5 1,345 36.7 11.5 0.9 10.0 16 2.77 57 51 43 25.6 2,956 54.4 8.7 1.0 9.0 17 2.74 38 61 39 18 2.76 52 42 58 18.4 3,471 58.9 5.3 1.1 5.7 20 2.69 43 58 47 13.3 3,022 55.0 4.4 1.0 4.6 21 2.70 28 65 58 15.6 3,434 58.6 4.6 1.1 4.9 22 2.77 51 58 24 13.2 1,665 40.8 7.9 0.9 7.2 23 2.74 16 55 40 11.4 1,808 42.5 6.3 0.9 5.8 24 2.74 20 33 33 5.4 1,113 33.4 4.9 0.8 4.0 27 2.64 29 51 43 6.2 2,190 46.8 2.8 1.0 2.7 28 2.68 36 54 56 5.4 3,209 56.6 1.7 1.1 1.8 29 2.76 63 59 42 10.7 3,262 57.1 3.3 1.1 3.5 30 2.71 71 72 31 9.7 2,822 53.1 3.4 1.0 3.5 32 2.71 71 72 36 13.6 3,277 57.2 4.1 1.1 4.4 33 2.76 33 62 45 14.1 2,722 52.2 5.2 1.0 5.3 37 2.78 45 78 37 5.2 2,897 53.8 1.8 1.0 1.9 38 2.74 80 61 35 25.4 3,142 56.1 8.1 1.1 8.5 195 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 40 2.75 46 77 42 33.5 3,289 57.3 10.2 1.1 10.8 41 2.66 113 84 64 16.6 8,027 89.6 2.1 1.3 2.7 42 2.75 74 79 59 30.1 5,747 75.8 5.2 1.2 6.3 43 2.71 33 63 42 44 2.68 67 51 46 17.7 3,456 58.8 5.1 1.1 5.5 45 2.69 76 103 72 17.3 8,205 90.6 2.1 1.3 2.7 47 2.75 70 45 40 20.3 2,928 54.1 6.9 1.0 7.2 49 2.72 75 79 47 22.2 4,608 67.9 4.8 1.1 5.5 50 2.67 54 72 40 17.8 3,209 56.6 5.6 1.1 5.9 51 2.75 58 76 45 23.9 3,839 62.0 6.2 1.1 6.9 52 2.76 47 66 28 15.6 2,014 44.9 7.8 1.0 7.4 53 2.72 50 45 40 15.3 2,419 49.2 6.3 1.0 6.3 56 2.70 50 56 37 8.6 2,497 50.0 3.4 1.0 3.4 57 2.67 18 41 45 58 2.72 95 101 45 18.1 5,615 74.9 3.2 1.2 3.9 60 2.74 33 61 57 14.2 3,411 58.4 4.2 1.1 4.5 61 2.83 59 28 32 8.2 1,772 42.1 4.6 0.9 4.3 62 2.66 29 55 26 6.2 1,390 37.3 4.5 0.9 3.9 63 2.70 26 42 39 66 2.66 40 52 33 7.2 1,933 44.0 3.7 0.9 3.5 68 2.90 42 59 41 26.0 2,636 51.3 9.9 1.0 10.0 69 2.67 66 55 30 8.5 2,311 48.1 3.7 1.0 3.6 70 2.70 56 47 44 19.9 2,885 53.7 6.9 1.0 7.1 71 2.63 28 54 37 15.9 1,932 43.9 8.2 0.9 7.8 196 E23N, Mean Is(50): 5.6 MPa E23N, Median Is(50) 5.5 MPa Standard deviation 2.3 MPa Equivalent UCS (NX 54mm Core): 96 MPa Figure D.9 Point load test results for drawpoint sample E23N taken Nov 2015 y = 0.004348x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]197 Table D.10 Point load test results for drawpoint sample E30N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 2 2.70 103 102 51 10.2 6,656 81.6 1.5 1.2 1.9 4 2.71 61 55 54 13.9 3,988 63.1 3.5 1.1 3.9 5 2.72 70 56 36 4.8 2,888 53.7 1.7 1.0 1.7 7 2.75 48 40 38 13.0 2,129 46.1 6.1 1.0 5.9 8 2.79 63 70 58 22.5 4,911 70.1 4.6 1.2 5.3 10 2.65 54 39 46 29.7 2,723 52.2 10.9 1.0 11.1 11 2.71 12 2.70 43 57 34 22.6 2,165 46.5 10.4 1.0 10.1 14 2.68 80 36 38 20.1 2,806 53.0 7.2 1.0 7.3 15 2.63 49 49 48 6.8 2,995 54.7 2.3 1.0 2.4 16 2.70 46 46 36 10.1 2,108 45.9 4.8 1.0 4.6 18 2.67 37 58 51 11.1 3,084 55.5 3.6 1.0 3.8 21 2.82 58 48 49 14.4 3,307 57.5 4.4 1.1 4.7 22 2.76 94 84 45 21.3 5,099 71.4 4.2 1.2 4.9 23 2.63 61 42 50 27.0 3,279 57.3 8.2 1.1 8.8 24 2.73 52 51 46 13.2 3,002 54.8 4.4 1.0 4.6 25 2.87 46 60 57 23.1 3,846 62.0 6.0 1.1 6.6 28 2.75 29 2.64 59 1 28 25.7 1,070 32.7 24.1 0.8 19.9 31 2.69 50 57 51 30.0 3,474 58.9 8.6 1.1 9.3 32 2.71 30 41 53 12.9 2,396 48.9 5.4 1.0 5.3 33 2.75 45 55 39 18.0 2,483 49.8 7.3 1.0 7.2 34 2.67 45 33 32 7.9 1,589 39.9 5.0 0.9 4.5 36 2.67 39 47 36 10.4 1,971 44.4 5.3 0.9 5.0 38 2.73 70 76 39 25.8 3,625 60.2 7.1 1.1 7.8 39 2.64 52 43 28 6.3 1,693 41.2 3.7 0.9 3.4 42 2.74 73 47 39 30.1 2,979 54.6 10.1 1.0 10.5 198 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 43 2.90 92 93 65 21.9 7,655 87.5 2.9 1.3 3.7 44 2.00 79 84 63 26.8 6,537 80.9 4.1 1.2 5.1 45 2.79 48 2.71 91 85 50 21.6 5,602 74.8 3.9 1.2 4.6 49 2.81 73 66 30 26.5 2,655 51.5 10.0 1.0 10.1 54 2.65 26 55 38 14.8 1,960 44.3 7.6 0.9 7.2 55 2.71 56 2.69 40 63 37 11.4 2,426 49.3 4.7 1.0 4.7 57 2.64 35 50 42 20.0 2,273 47.7 8.8 1.0 8.6 58 2.76 15 42 50 12.3 1,814 42.6 6.8 0.9 6.3 59 2.79 50 55 44 11.4 2,941 54.2 3.9 1.0 4.0 60 2.70 36 46 48 20.8 2,506 50.1 8.3 1.0 8.3 61 2.67 53 10 34 10.0 1,364 36.9 7.4 0.9 6.4 62 2.68 60 58 40 4.7 3,005 54.8 1.6 1.0 1.6 63 2.88 81 32 28 20.9 2,014 44.9 10.4 1.0 9.9 64 2.75 44 50 32 12.1 1,915 43.8 6.3 0.9 5.9 E30N Mean Is(50): 6.3 MPa E30N Median Is(50) 5.3 MPa Standard deviation 3.4 MPa Equivalent UCS (NX 54mm Core): 108 MPa 199 Figure D.10 Point load test results for drawpoint sample E30N taken Nov 2015 y = 0.004751x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]200 Table D.11 Point load test results for drawpoint sample F33S taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 3 2.52 87 70 52 7.0 5,197 72.1 1.3 1.2 1.6 4 2.65 81 74 53 7.0 5,230 72.3 1.3 1.2 1.6 7 2.90 88 77 53 9.2 5,567 74.6 1.7 1.2 2.0 9 2.70 99 85 77 11.1 9,020 95.0 1.2 1.3 1.6 13 2.89 72 68 62 15.1 5,526 74.3 2.7 1.2 3.3 17 2.58 124 131 44 7.6 7,143 84.5 1.1 1.3 1.4 21 2.65 36 33 35 0.8 1,537 39.2 0.5 0.9 0.5 23 2.81 47 46 34 13.0 2,013 44.9 6.5 1.0 6.2 24 2.48 65 42 47 1.6 3,202 56.6 0.5 1.1 0.5 25 2.69 51 58 35 16.5 2,429 49.3 6.8 1.0 6.7 26 2.63 48 50 33 4.0 2,059 45.4 2.0 1.0 1.9 30 2.48 43 51 34 1.2 2,035 45.1 0.6 1.0 0.5 31 2.70 66 79 56 9.9 5,169 71.9 1.9 1.2 2.3 32 2.73 61 73 56 21.2 4,777 69.1 4.4 1.2 5.1 34 2.49 63 63 52 0.8 4,171 64.6 0.2 1.1 0.2 39 2.46 82 88 70 0.8 7,576 87.0 0.1 1.3 0.1 1 2.73 43 41 32 8.6 1,711 41.4 5.0 0.9 4.6 6 2.71 36 32 29 8.7 1,255 35.4 7.0 0.9 6.0 7 2.76 41 41 30 13.4 1,566 39.6 8.6 0.9 7.7 12 2.67 42 44 35 2.9 1,916 43.8 1.5 0.9 1.4 16 2.45 31 2.72 46 43 31 10.5 1,756 41.9 6.0 0.9 5.5 44 2.66 32 27 32 8.5 1,202 34.7 7.1 0.8 6.0 46 2.66 50 40 30 12.6 1,719 41.5 7.3 0.9 6.8 64 2.67 38 34 34 8.8 1,558 39.5 5.6 0.9 5.1 72 2.73 42 43 36 2.1 1,948 44.1 1.1 0.9 1.0 76 2.85 31 31 31 4.0 1,224 35.0 3.3 0.9 2.8 201 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 89 2.62 34 40 31 8.0 1,460 38.2 5.5 0.9 4.9 F33S Mean Is(50): 3.2 MPa F33S Median Is(50) 2.3 MPa Standard deviation 2.4 MPa Equivalent UCS (NX 54mm Core): 55 MPa 202 Figure D.11 Point load test results for drawpoint sample F33S taken Nov 2015 y = 0.001724x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]203 Table D.12 Point load test results for drawpoint sample G40N taken Nov 2015 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 4 2.54 84 79 51 15.5 5,292 72.7 2.9 1.2 3.5 6 2.65 16.4 3,441 58.7 4.8 1.1 5.1 7 2.61 59 56 47 7.5 3,748 61.2 2.0 1.1 2.2 8 2.60 63 65 46 21.3 5,407 73.5 3.9 1.2 4.7 10 2.74 76 73 57 21.3 5,407 73.5 3.9 1.2 4.7 14 2.68 62 27 53 6.5 3,003 54.8 2.2 1.0 2.3 15 2.59 68 60 42 6.0 3,422 58.5 1.8 1.1 1.9 16 2.68 65 62 56 4.3 4,528 67.3 0.9 1.1 1.1 17 2.61 78 80 56 9.8 5,633 75.1 1.7 1.2 2.1 19 2.68 63 58 40 11.1 3,081 55.5 3.6 1.0 3.8 20 2.64 77 76 39 8.2 3,799 61.6 2.2 1.1 2.4 21 2.66 75 70 37 3.7 3,415 58.4 1.1 1.1 1.2 22 2.62 113 94 52 15.7 6,853 82.8 2.3 1.3 2.9 23 2.72 136 120 62 25 2.65 76 65 48 19.4 4,309 65.6 4.5 1.1 5.1 26 2.70 27 2.58 68 30 43 6.8 2,683 51.8 2.5 1.0 2.6 29 2.68 77 71 41 5.6 3,863 62.2 1.5 1.1 1.6 31 2.75 38 84 41 17.8 3,184 56.4 5.6 1.1 5.9 33 2.64 73 89 48 11.1 4,950 70.4 2.2 1.2 2.6 34 2.66 103 95 46 18.7 5,798 76.1 3.2 1.2 3.9 35 2.72 55 70 44 3.6 3,501 59.2 1.0 1.1 1.1 36 2.67 39 59 44 21.4 2,745 52.4 7.8 1.0 8.0 37 2.64 95 39 56 7.4 4,777 69.1 1.5 1.2 1.8 41 2.67 23 46 39 12.9 1,713 41.4 7.5 0.9 6.9 43 2.64 22 38 39 12.9 1,490 38.6 8.7 0.9 7.7 46 2.68 23 50 31.8 9.9 1,478 38.4 6.7 0.9 6.0 204 Specimen # Specific Gravity Width 1 Width 2 Diameter Load P De2 De Is Correction Factor F Is(50) [mm] [mm] [mm] [kN] [mm2] [mm] [MPa] [MPa] [MPa] 47 2.72 55 13 24 6.5 1,039 32.2 6.3 0.8 5.2 48 2.57 45 51 38 1.6 2,322 48.2 0.7 1.0 0.7 50 2.67 51 55 24 3.3 1,620 40.2 2.0 0.9 1.8 51 2.71 39 50 31 7.4 1,756 41.9 4.2 0.9 3.9 52 2.63 40 63 20 9.5 1,313 36.2 7.3 0.9 6.3 G40N Mean Is(50): 3.6 MPa G40N Median Is(50) 3.2 MPa Standard deviation 2.1 MPa Equivalent UCS (NX 54mm Core): 62 MPa 205 Figure D.12 Point load test results for drawpoint sample G40N taken Nov 2015 y = 0.002828x1.010.0100.01,000 10,000 100,000Load P [kN]De2 [mm2]206 DropWeight Test Results In addition to standard DropWeight parameters A and b, material parameters of an alternative energy-breakage equation from the work of Shi and Kojovic (2007) were also reported: 𝑡10 = 𝑀(1 − 𝑒−𝑓𝑚𝑎𝑡.𝑥𝑛(𝐸𝑐𝑠−𝐸𝑚𝑖𝑛)) (E.1) Where t10 is the percentage passing 1/10th of the feed size, M and fmat are material parameters, x is feed size, Ecs represents input energy, Emin is the energy threshold of the material and n describes the effect of feed size on breakage. A New Afton mill feed sample was tested using twenty-one energy-size setpoints to determine material parameters M, fmat and n, shown in Table F.1. The fitted parameter n, having a value of 0.37, was used to determine unique fmat and M parameters for each New Afton geometallurgical unit, which were tested using the abbreviated DropWeight method described in Appendix B. Equation (F.1) was rearranged to estimate Ecs for different t10 and feed size values x, so that a table of energy-breakage data could be generated for use in the JKSimMet crusher model. 𝐸𝑐𝑠 =ln (1−𝑡10𝑀)−𝑓𝑚𝑎𝑡.𝑥𝑛+ 𝐸𝑚𝑖𝑛 (E.2) Reported DWI values were calculated using the following equation from Doll (2016) 𝐷𝑊𝐼 (𝑘𝑊ℎ𝑚3) = ƿ.96.703(𝐴.𝑏)0.992 (E.3) Where ƿ is the average specific gravity of sample determined from the dry weight and apparent weight of samples after immersion in water. Table E.1 DropWeight test results for mill feed sample collected March, 2016 207 Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramProject: New GoldDeposit: New AftonSample ID:Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs57.8 17.0 0.40 9.2 0.25 4.2 0.1041.1 35.5 0.96 9.9 0.25 4.6 0.1028.9 54.8 3.33 50.7 2.49 31.8 1.00 15.9 0.50 9.4 0.2520.6 55.5 3.49 50.8 2.49 27.9 1.00 14.1 0.50 9.2 0.2514.5 53.3 3.49 44.4 2.50 23.7 1.00 13.4 0.50 6.6 0.25SAG/AG MILL PARAMETERSA: 61.3 b: 0.64 A*b: 39.2 Ta: 0.29DWi (kWh/m3): 6.98M: 65.2 fmat: 0.18 n: 0.37 Emin (kWh/t): 0.01CRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.36 0.32 0.28 0.25 0.2220 0.77 0.68 0.60 0.53 0.4730 1.29 1.14 1.01 0.89 0.78DENSITY DATAMean 2.75 Std Dev 0.07 Max 2.89 Min 2.63Ecs (kWh/t)Stefan NadolskiTest Date: 26-Jul-16Tester: Stefan NadolskiMill Feed, Collected 11 AM March 16th, 2016, Extended DropWeight Test (21 energy-size combinations)208 Figure E.1 Breakage index t10 and input energy with the standard A & b model Figure E.2 Breakage index t10 and input energy and fitted M and fmat model 209 Figure E.3 Breakage index t10 and tn for mill feed (21 energy-size combinations) 210 Table E.2 DropWeight test results for 2015 mill feed sample, SMC method (-31.5, 26.5 mm) Ecs, [kWh/t] 0.247 0.499 0.998 2.058 3.119 t10, [%] 9.05 14.50 27.54 47.17 55.05 Results taken from SMC test report provided by ALS Metallurgy KM4697 Note: Abrasion results, ta, were derived by undisclosed SMC methods, which are assumed to reference drop-weight test results and specific gravity 211 Table E.3 DropWeight test results for drawpoint sample B13S (collected Nov, 2015) Figure E.4 Breakage index t10 and energy with fitted A & b curve for drawpoint sample B13S Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample B13S (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 57.2 2.50 48.0 1.80 36.4 1.00 19.7 0.50 9.8 0.25SAG/FAG MILL PARAMETERSA: 71.5 b: 0.64 A*b: 46.1 Ta: 0.44DWi (kWh/m3): 5.82M: 65.0 fmat: 0.24 n*: 0.37 Emin (kWh/t): 0.05* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.32 0.28 0.26 0.23 0.2120 0.63 0.56 0.50 0.45 0.4030 1.02 0.90 0.80 0.71 0.64DENSITY DATAMean 2.69 Std Dev 0.12 Max 3.22 Min 2.48Ecs (kWh/t)Tester: Stefan212 Table E.4 DropWeight test results for drawpoint sample D7S (collected Nov, 2015) Figure E.5 Breakage index t10 and energy with fitted A & b curve for drawpoint sample D7S Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample D7S (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 51.9 2.43 47.7 1.80 30.4 1.00 17.9 0.50 8.4 0.25SAG/FAG MILL PARAMETERSA: 73.3 b: 0.54 A*b: 39.5 Ta: 0.40DWi (kWh/m3): 6.86M: 65.4 fmat: 0.20 n*: 0.37 Emin (kWh/t): 0.05* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.75 3.56 5.53 25.71 58.5120 5.49 7.12 11.06 46.62 83.3730 8.24 10.68 16.59 63.61 93.93POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.36 0.32 0.29 0.26 0.2420 0.73 0.65 0.58 0.51 0.4630 1.19 1.06 0.94 0.83 0.74DENSITY DATAMean 2.72 Std Dev 0.07 Max 2.86 Min 2.62Ecs (kWh/t)Tester: Stefan213 Table E.5 DropWeight test results for drawpoint sample D11N (collected Nov, 2015) Figure E.6 Breakage index t10 and energy with fitted A & b curve for drawpoint sample D11N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample D11N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 52.9 2.40 46.7 1.80 31.4 1.00 17.9 0.50 7.7 0.25SAG/FAG MILL PARAMETERSA: 77.9 b: 0.49 A*b: 38.4 Ta: 0.39DWi (kWh/m3): 6.91M: 64.4 fmat: 0.21 n*: 0.37 Emin (kWh/t): 0.08* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.75 3.56 5.53 25.71 58.5120 5.49 7.12 11.06 46.62 83.3730 8.24 10.68 16.59 63.61 93.93POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.37 0.33 0.30 0.27 0.2520 0.72 0.64 0.58 0.52 0.4630 1.16 1.03 0.92 0.82 0.73DENSITY DATAMean 2.67 Std Dev 0.17 Max 3.02 Min 2.01Ecs (kWh/t)Tester: Stefan214 Table E.6 DropWeight test results for drawpoint sample D11S (collected Nov, 2015) Figure E.7 Breakage index t10 and energy with fitted A & b curve for drawpoint sample D11S Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample D11S (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 59.3 3.14 55.0 2.50 29.1 1.00 14.1 0.50 6.8 0.25SAG/FAG MILL PARAMETERSA: 90.7 b: 0.36 A*b: 32.3 Ta: 0.41DWi (kWh/m3): 8.31M: 87.0 fmat: 0.11 n*: 0.37 Emin (kWh/t): 0.01* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.42 0.37 0.33 0.29 0.2620 0.89 0.78 0.69 0.61 0.5430 1.43 1.26 1.11 0.98 0.87DENSITY DATAMean 2.70 Std Dev 0.05 Max 2.81 Min 2.61Ecs (kWh/t)Tester: Stefan215 Table E.7 DropWeight test results for drawpoint sample D38N (collected Nov, 2015) Figure E.8 Breakage index t10 and energy with fitted A & b curve for drawpoint sample D38N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample D38N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 56.5 2.40 49.3 1.80 36.9 1.00 24.6 0.50 14.3 0.25SAG/FAG MILL PARAMETERSA: 60.1 b: 1.02 A*b: 61.2 Ta: 0.92DWi (kWh/m3): 4.45M: 60.1 fmat: 0.29 n*: 0.37 Emin (kWh/t): 0.00* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.23 0.20 0.18 0.16 0.1420 0.51 0.45 0.40 0.35 0.3130 0.88 0.77 0.68 0.60 0.53DENSITY DATAMean 2.72 Std Dev 0.06 Max 2.85 Min 2.59Ecs (kWh/t)Tester: Stefan216 Table E.8 DropWeight test results for drawpoint sample E13S (collected Nov, 2015) Figure E.9 Breakage index t10 and energy with fitted A & b curve for drawpoint sample E13S Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample: Drawpoint Sample E13S (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 51.4 2.50 44.9 1.80 28.3 1.00 16.5 0.50 6.4 0.25SAG/FAG MILL PARAMETERSA: 84.3 b: 0.40 A*b: 33.5 Ta: 0.27DWi (kWh/m3): 8.08M: 63.9 fmat: 0.20 n*: 0.37 Emin (kWh/t): 0.09* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.41 0.37 0.34 0.31 0.2820 0.80 0.71 0.64 0.57 0.5230 1.28 1.14 1.02 0.90 0.81DENSITY DATAMean 2.72 Std Dev 0.07 Max 2.89 Min 2.62Ecs (kWh/t)Tester: Stefan217 Table E.9 DropWeight test results for drawpoint sample E15N (collected Nov, 2015) Figure E.10 Breakage index t10 and energy with fitted A & b curve for drawpoint sample E15N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample: Drawpoint Sample E15N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 51.4 2.50 44.9 1.80 28.3 1.00 16.5 0.50 6.4 0.25SAG/FAG MILL PARAMETERSA: 84.3 b: 0.40 A*b: 33.5 Ta: 0.27DWi (kWh/m3): 8.08M: 63.9 fmat: 0.20 n*: 0.37 Emin (kWh/t): 0.09* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.41 0.37 0.34 0.31 0.2820 0.80 0.71 0.64 0.57 0.5230 1.28 1.14 1.02 0.90 0.81DENSITY DATAMean 2.72 Std Dev 0.07 Max 2.89 Min 2.62Ecs (kWh/t)Tester: Stefan218 Table E.10 DropWeight test results for drawpoint sample E23N (collected Nov, 2015) Figure E.11 Breakage index t10 and energy with fitted A & b curve for drawpoint sample E23N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample: Drawpoint Sample E23N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 56.0 2.80 48.0 2.00 32.5 1.00 15.4 0.50 7.1 0.25SAG/FAG MILL PARAMETERSA: 72.3 b: 0.54 A*b: 39.2 Ta: 0.28DWi (kWh/m3): 6.91M: 65.6 fmat: 0.20 n*: 0.37 Emin (kWh/t): 0.09* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.39 0.36 0.33 0.30 0.2720 0.76 0.68 0.61 0.54 0.4930 1.21 1.07 0.96 0.85 0.76DENSITY DATAMean 2.72 Std Dev 0.20 Max 3.71 Min 2.55Ecs (kWh/t)Tester: Stefan219 Table E.11 DropWeight test results for drawpoint sample E30N (collected Nov, 2015) Figure E.12 Breakage index t10 and energy with fitted A & b curve for drawpoint sample E30N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample: Drawpoint Sample E30N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 50.6 2.40 47.4 1.80 30.9 1.00 17.3 0.50 8.1 0.25SAG/FAG MILL PARAMETERSA: 73.1 b: 0.53 A*b: 38.8 Ta: 0.38DWi (kWh/m3): 7.08M: 62.5 fmat: 0.22 n*: 0.37 Emin (kWh/t): 0.07* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.37 0.33 0.30 0.27 0.2520 0.73 0.65 0.58 0.52 0.4730 1.19 1.05 0.94 0.83 0.74DENSITY DATAMean 2.76 Std Dev 0.11 Max 2.95 Min 2.50Ecs (kWh/t)Tester: Stefan220 Table E.12 DropWeight test results for drawpoint sample F33S (collected Nov, 2015) Figure E.13 Breakage index t10 and energy with fitted A & b curve for drawpoint sample F33S Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample: Drawpoint Sample F33S (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 56.2 2.35 52.1 1.80 39.9 1.00 26.2 0.50 15.0 0.25SAG/FAG MILL PARAMETERSA: 60.1 b: 1.12 A*b: 67.5 Ta: 0.66DWi (kWh/m3): 3.91M: 60.1 fmat: 0.32 n*: 0.37 Emin (kWh/t): 0.00* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.72 58.5120 5.5 7.1 11.1 46.62 83.3730 8.2 10.7 16.6 63.61 93.93POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.21 0.18 0.16 0.14 0.1320 0.46 0.41 0.36 0.32 0.2830 0.79 0.70 0.62 0.54 0.48DENSITY DATAMean 2.64 Std Dev 0.12 Max 2.91 Min 2.42Ecs (kWh/t)Tester: Stefan221 Table E.13 DropWeight test results for drawpoint sample G40N (collected Nov, 2015) Figure E.14 Breakage index t10 and energy with fitted A & b curve for drawpoint sample G40N Drop Weight Test Parameters for Crushers and SAG MillsNew Gold / UBC Cave-to-Mill ProgramDeposit: New AftonSample ID: Drawpoint Sample G40N (Collected Nov, 2015)Base Datat1 t10 Ecs t10 Ecs t10 Ecs t10 Ecs t10 Ecs28.9 53.2 2.20 42.8 1.60 30.9 1.00 19.3 0.50 9.9 0.25SAG/FAG MILL PARAMETERSA: 75.4 b: 0.54 A*b: 40.9 Ta: 0.50DWi (kWh/m3): 6.47M: 72.1 fmat: 0.17 n*: 0.37 Emin (kWh/t): 0.00* Value from calibration sampleCRUSHER PARAMETERSCRUSHER APPEARANCE FUNCTION DATAt10 t75 t50 t25 t4 t210 2.7 3.6 5.5 25.7 58.520 5.5 7.1 11.1 46.6 83.430 8.2 10.7 16.6 63.6 93.9POWER DATAMean Size (mm)14.5 20.6 28.9 41.1 57.8t1010 0.33 0.29 0.26 0.22 0.2020 0.72 0.63 0.56 0.49 0.4330 1.19 1.04 0.92 0.81 0.71DENSITY DATAMean 2.66 Std Dev 0.07 Max 2.88 Min 2.56Ecs (kWh/t)Tester: Stefan222 Bond Ball Mill Index Results Table F.1 Bond ball mill work index test results for mill feed sample collected March, 2016 BOND BALL GRINDABILITY TEST New Afton Mill Feed (March 2016)1379.5 g Aperture Test Sieve : 106µm394.1 g Percent Undersize : 17.6%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1379.5 200 362.4 242.4 120.0 0.602 362.4 551 538.5 63.7 474.8 0.863 538.5 347 421.2 94.6 326.6 0.944 421.2 341 409.9 74.0 335.9 0.995 409.9 327 389.8 72.0 317.8 0.976 389.8 335 396.0 68.5 327.5 0.98BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.98 gP = 80% Passing size of test product. 78 µmF = 80% Passing size of test feed. 2436 µm18.4 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 9.50 1.94 98.1 - - -7 Mesh 2800 42.80 8.75 89.3 - - -8 Mesh 2360 55.20 11.29 78.0 - - -9 Mesh 2000 38.20 7.81 70.2 - - -10 Mesh 1700 40.00 8.18 62.0 - - -12 Mesh 1400 35.40 7.24 54.8 - - -14 Mesh 1180 24.70 5.05 49.7 - - -20 Mesh 850 38.80 7.94 41.8 - - -28 Mesh 600 31.10 6.36 35.4 - - -35 Mesh 425 24.30 4.97 30.5 - - -48 Mesh 300 19.20 3.93 26.5 - - -65 Mesh 212 16.20 3.31 23.2 - - -100 Mesh 150 15.40 3.15 20.1 - - -150 Mesh 106 12.20 2.50 17.6 2.00 1.99 98.0170 Mesh 90 - - - 9.00 8.94 89.1200 Mesh 75 - - - 11.80 11.72 77.4270 Mesh 53 - - - 13.30 13.21 64.2325 Mesh 45 - - - 6.50 6.45 57.7400 Mesh 38 - - - 5.10 5.06 52.6TOTAL 488.9 100.00 ** 100.7 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2436µm K80 = 78µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS223 Table F.2 Bond ball mill work index test results for mill feed sample collected March, 2015 BOND BALL GRINDABILITY TEST New Afton Mill Feed (March 3rd, 2015)1337 g Aperture Test Sieve : 180µm382.0 g Percent Undersize : 14.6%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1336.5 100 329.8 195.7 134.1 1.342 329.8 249 337.1 48.3 288.8 1.163 337.1 286 389.5 49.4 340.1 1.194 389.5 273 387.2 57.0 330.2 1.215 387.2 269 390.0 56.7 333.3 1.246 390.0 262 389.0 57.1 331.9 1.2657 389.0 257 381.3 57.0 324.3 1.2638 381.3 258 382.0 55.8 326.2 1.263BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 180 µmGbp = Net undersize produced per revolution of mill. 1.26 gP = 80% Passing size of test product. 134 µmF = 80% Passing size of test feed. 2147 µm18.9 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 0.00 0.00 100.0 - - -7 Mesh 2800 40.70 3.50 96.5 - - -9 Mesh 2000 237.60 20.42 76.1 - - -12 Mesh 1400 219.30 18.85 57.2 - - -16 Mesh 1000 145.80 12.53 44.7 - - -24 Mesh 710 103.30 8.88 35.8 - - -32 Mesh 500 85.40 7.34 28.5 - - -42 Mesh 355 65.50 5.63 22.9 - - -60 Mesh 250 52.70 4.53 18.3 - - -80 Mesh 180 43.00 3.70 14.6 0.00 0.00 100.0100 Mesh 150 - - - 39.80 13.46 86.5115 Mesh 125 - - - 32.20 10.89 75.7170 Mesh 90 - - - 42.00 14.20 61.4250 Mesh 63 - - - 32.50 10.99 50.5325 Mesh 45 - - - 23.20 7.85 42.6Pan 170.40 14.64 -- - -- - -- - -- 126 42.61TOTAL 1163.7 100.00 ** 295.7 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2147µm K80 = 134µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS224 Table F.3 Bond ball mill work index test results for drawpoint sample B13S collected Nov, 2015 BOND BALL GRINDABILITY TEST B13S Drawpoint sample (Nov 2015)1277 g Aperture Test Sieve : 106µm364.9 g Percent Undersize : 11.4%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1277.0 200 252.9 145.4 107.5 0.542 252.9 625 542.2 28.8 513.4 0.823 542.2 369 414.9 61.8 353.1 0.964 414.9 332 363.7 47.3 316.4 0.955 363.7 339 366.8 41.4 325.4 0.966 366.8 337 363.2 41.8 321.4 0.95BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.96 gP = 80% Passing size of test product. 79 µmF = 80% Passing size of test feed. 2604 µm18.8 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 34.60 6.82 93.2 - - -7 Mesh 2800 40.60 8.00 85.2 - - -8 Mesh 2360 59.50 11.72 73.5 - - -9 Mesh 2000 42.30 8.33 65.1 - - -10 Mesh 1700 42.80 8.43 56.7 - - -12 Mesh 1400 38.40 7.57 49.1 - - -14 Mesh 1180 26.50 5.22 43.9 - - -20 Mesh 850 42.20 8.32 35.6 - - -28 Mesh 600 33.70 6.64 28.9 - - -35 Mesh 425 25.90 5.10 23.8 - - -48 Mesh 300 19.60 3.86 20.0 - - -65 Mesh 212 16.50 3.25 16.7 - - -100 Mesh 150 15.00 2.96 13.8 - - -150 Mesh 106 12.10 2.38 11.4 3.40 3.51 96.5170 Mesh 90 - - - 7.90 8.16 88.3200 Mesh 75 - - - 11.20 11.57 76.8270 Mesh 53 - - - 14.60 15.08 61.7325 Mesh 45 - - - 7.40 7.64 54.0400 Mesh 38 - - - 6.30 6.51 47.5TOTAL 507.5 100.00 ** 96.8 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2604µm K80 = 79µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS225 Table F.4 Bond ball mill work index test results for drawpoint sample D7S collected Nov, 2015 BOND BALL GRINDABILITY TEST D7S Drawpoint sample (Nov 2015)1252.9 g Aperture Test Sieve : 106µm358.0 g Percent Undersize : 9.7%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1252.9 200 247.7 121.4 126.3 0.632 247.7 529 478.9 24.0 454.9 0.863 478.9 362 395.7 46.4 349.3 0.964 395.7 331 364.7 38.3 326.4 0.985 364.7 328 362.4 35.3 327.1 1.006 362.4 323 360.6 35.1 325.5 1.017 360.6 321 358.6 34.9 323.7 1.01BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 1.00 gP = 80% Passing size of test product. 80 µmF = 80% Passing size of test feed. 2764 µm18.0 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 43.60 9.08 90.9 - - -7 Mesh 2800 47.60 9.92 81.0 - - -8 Mesh 2360 59.20 12.33 68.7 - - -9 Mesh 2000 41.00 8.54 60.1 - - -10 Mesh 1700 43.90 9.15 51.0 - - -12 Mesh 1400 35.30 7.35 43.6 - - -14 Mesh 1180 24.30 5.06 38.6 - - -20 Mesh 850 36.90 7.69 30.9 - - -28 Mesh 600 28.80 6.00 24.9 - - -35 Mesh 425 21.70 4.52 20.4 - - -48 Mesh 300 16.50 3.44 16.9 - - -65 Mesh 212 13.70 2.85 14.1 - - -100 Mesh 150 11.90 2.48 11.6 - - -150 Mesh 106 9.10 1.90 9.7 2.40 2.22 97.8170 Mesh 90 - - - 10.70 9.89 87.9200 Mesh 75 - - - 12.60 11.65 76.2270 Mesh 53 - - - 14.40 13.31 62.9325 Mesh 45 - - - 7.00 6.47 56.5400 Mesh 38 - - - 5.50 5.08 51.4TOTAL 480.0 100.00 ** 108.2 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2764µm K80 = 80µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS226 Table F.5 Bond ball mill work index test results for drawpoint sample D11N collected Nov, 2015 BOND BALL GRINDABILITY TEST D11N Drawpoint sample (Nov 2015)1216 g Aperture Test Sieve : 106µm347.4 g Percent Undersize : 9.0%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1216.0 200 210.5 109.4 101.1 0.512 210.5 650 545.2 18.9 526.3 0.813 545.2 368 404.9 49.0 355.9 0.974 404.9 322 354.9 36.4 318.5 0.995 354.9 319 348.5 31.9 316.6 0.996 348.5 318 343.0 31.3 311.7 0.98BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.99 gP = 80% Passing size of test product. 80 µmF = 80% Passing size of test feed. 2645 µm18.4 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 33.50 6.54 93.5 - - -7 Mesh 2800 47.00 9.17 84.3 - - -8 Mesh 2360 62.80 12.25 72.0 - - -9 Mesh 2000 49.10 9.58 62.5 - - -10 Mesh 1700 48.60 9.48 53.0 - - -12 Mesh 1400 42.20 8.23 44.7 - - -14 Mesh 1180 30.00 5.85 38.9 - - -20 Mesh 850 45.60 8.90 30.0 - - -28 Mesh 600 34.20 6.67 23.3 - - -35 Mesh 425 24.00 4.68 18.6 - - -48 Mesh 300 16.80 3.28 15.4 - - -65 Mesh 212 13.20 2.58 12.8 - - -100 Mesh 150 11.00 2.15 10.6 - - -150 Mesh 106 8.40 1.64 9.0 3.50 3.48 96.5170 Mesh 90 - - - 9.00 8.94 87.6200 Mesh 75 - - - 11.70 11.62 76.0270 Mesh 53 - - - 13.90 13.80 62.2325 Mesh 45 - - - 6.30 6.26 55.9400 Mesh 38 - - - 5.10 5.06 50.8TOTAL 512.5 100.00 ** 100.7 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2645µm K80 = 80µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS227 Table F.6 Bond ball mill work index test results for drawpoint sample D11S collected Nov, 2015 BOND BALL GRINDABILITY TEST D11S Drawpoint sample (Nov 2015)1274.7 g Aperture Test Sieve : 106µm364.2 g Percent Undersize : 10.2%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1274.7 200 242.9 129.6 113.3 0.572 242.9 599 487.4 24.7 462.7 0.773 487.4 407 408.8 49.5 359.3 0.884 408.8 366 375.0 41.6 333.4 0.915 375.0 358 366.5 38.1 328.4 0.926 366.5 356 368.4 37.3 331.1 0.937 368.4 352 367.2 37.4 329.8 0.948 367.2 348 362.1 37.3 324.8 0.93BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.93 gP = 80% Passing size of test product. 81 µmF = 80% Passing size of test feed. 2701 µm19.3 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 38.60 7.96 92.0 - - -7 Mesh 2800 45.10 9.30 82.7 - - -8 Mesh 2360 59.50 12.27 70.5 - - -9 Mesh 2000 41.50 8.56 61.9 - - -10 Mesh 1700 40.60 8.37 53.5 - - -12 Mesh 1400 35.40 7.30 46.2 - - -14 Mesh 1180 24.70 5.09 41.2 - - -20 Mesh 850 38.50 7.94 33.2 - - -28 Mesh 600 31.30 6.45 26.8 - - -35 Mesh 425 23.90 4.93 21.8 - - -48 Mesh 300 17.90 3.69 18.1 - - -65 Mesh 212 15.10 3.11 15.0 - - -100 Mesh 150 13.20 2.72 12.3 - - -150 Mesh 106 10.40 2.14 10.2 3.50 3.28 96.7170 Mesh 90 - - - 10.00 9.37 87.3200 Mesh 75 - - - 12.80 12.00 75.4270 Mesh 53 - - - 14.30 13.40 61.9325 Mesh 45 - - - 7.10 6.65 55.3400 Mesh 38 - - - 5.70 5.34 50.0TOTAL 485.0 100.00 ** 106.7 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2701µm K80 = 81µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)228 Table F.7 Bond ball mill work index test results for drawpoint sample D38N collected Nov, 2015 BOND BALL GRINDABILITY TEST D38N Drawpoint sample (Nov 2015)1282 g Aperture Test Sieve : 106µm366.3 g Percent Undersize : 11.2%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1282.0 200 254.5 144.0 110.5 0.552 254.5 611 536.1 28.6 507.5 0.833 536.1 369 416.1 60.2 355.9 0.974 416.1 331 371.2 46.7 324.5 0.985 371.2 331 365.9 41.7 324.2 0.986 365.9 332 368.5 41.1 327.4 0.99BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.98 gP = 80% Passing size of test product. 81 µmF = 80% Passing size of test feed. 2606 µm18.6 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 30.10 6.23 93.8 - - -7 Mesh 2800 40.60 8.40 85.4 - - -8 Mesh 2360 59.10 12.22 73.2 - - -9 Mesh 2000 38.60 7.98 65.2 - - -10 Mesh 1700 44.20 9.14 56.0 - - -12 Mesh 1400 36.80 7.61 48.4 - - -14 Mesh 1180 25.90 5.36 43.1 - - -20 Mesh 850 40.00 8.27 34.8 - - -28 Mesh 600 31.80 6.58 28.2 - - -35 Mesh 425 23.90 4.94 23.3 - - -48 Mesh 300 18.20 3.76 19.5 - - -65 Mesh 212 15.40 3.19 16.3 - - -100 Mesh 150 13.80 2.85 13.5 - - -150 Mesh 106 10.80 2.23 11.2 2.60 2.39 97.6170 Mesh 90 - - - 11.00 10.09 87.5200 Mesh 75 - - - 13.20 12.11 75.4270 Mesh 53 - - - 14.90 13.67 61.7325 Mesh 45 - - - 6.60 6.06 55.7400 Mesh 38 - - - 5.30 4.86 50.8TOTAL 483.5 100.00 ** 109.0 100.00 ** Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2606µm K80 = 81µm229 Table F.8 Bond ball mill work index test results for drawpoint sample E13S collected Nov, 2015 1311 g Aperture Test Sieve : 106µm374.6 g Percent Undersize : 11.5%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1311.0 200 255.3 150.3 105.0 0.522 255.3 658 551.5 29.3 522.2 0.793 551.5 392 407.6 63.2 344.4 0.884 407.6 373 390.6 46.7 343.9 0.925 390.6 358 371.7 44.8 326.9 0.916 371.7 364 379.9 42.6 337.3 0.93BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.92 gP = 80% Passing size of test product. 78 µmF = 80% Passing size of test feed. 2544 µm19.3 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 24.50 5.11 94.9 - - -7 Mesh 2800 38.90 8.11 86.8 - - -8 Mesh 2360 56.20 11.72 75.1 - - -9 Mesh 2000 43.90 9.15 65.9 - - -10 Mesh 1700 38.80 8.09 57.8 - - -12 Mesh 1400 35.30 7.36 50.5 - - -14 Mesh 1180 25.30 5.27 45.2 - - -20 Mesh 850 38.30 7.98 37.2 - - -28 Mesh 600 33.60 7.00 30.2 - - -35 Mesh 425 26.30 5.48 24.7 - - -48 Mesh 300 19.60 4.09 20.6 - - -65 Mesh 212 17.00 3.54 17.1 - - -100 Mesh 150 15.20 3.17 13.9 - - -150 Mesh 106 11.80 2.46 11.5 1.90 1.97 98.0170 Mesh 90 - - - 8.60 8.92 89.1200 Mesh 75 - - - 11.40 11.83 77.3270 Mesh 53 - - - 13.70 14.21 63.1325 Mesh 45 - - - 6.30 6.54 56.5400 Mesh 38 - - - 5.30 5.50 51.0TOTAL 479.7 100.00 ** 96.4 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2544µm K80 = 78µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS230 Table F.9 Bond ball mill work index test results for drawpoint sample E15N collected Nov, 2015 BOND BALL GRINDABILITY TEST E15N Drawpoint sample (Nov 2015)1268.7 g Aperture Test Sieve : 106µm362.5 g Percent Undersize : 10.7%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1268.7 200 258.2 135.8 122.4 0.612 258.2 547 470.0 27.6 442.4 0.813 470.0 386 405.1 50.3 354.8 0.924 405.1 347 360.7 43.4 317.3 0.915 360.7 355 372.0 38.6 333.4 0.946 372.0 343 355.4 39.8 315.6 0.92BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.92 gP = 80% Passing size of test product. 80 µmF = 80% Passing size of test feed. 2690 µm19.4 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 38.40 8.06 91.9 - - -7 Mesh 2800 42.90 9.01 82.9 - - -8 Mesh 2360 56.30 11.82 71.1 - - -9 Mesh 2000 39.30 8.25 62.9 - - -10 Mesh 1700 39.50 8.29 54.6 - - -12 Mesh 1400 34.10 7.16 47.4 - - -14 Mesh 1180 23.90 5.02 42.4 - - -20 Mesh 850 38.40 8.06 34.3 - - -28 Mesh 600 30.90 6.49 27.9 - - -35 Mesh 425 23.80 5.00 22.9 - - -48 Mesh 300 18.20 3.82 19.0 - - -65 Mesh 212 15.30 3.21 15.8 - - -100 Mesh 150 13.60 2.85 13.0 - - -150 Mesh 106 10.80 2.27 10.7 3.70 3.36 96.6170 Mesh 90 - - - 10.00 9.08 87.6200 Mesh 75 - - - 13.10 11.90 75.7270 Mesh 53 - - - 15.10 13.71 61.9325 Mesh 45 - - - 7.00 6.36 55.6400 Mesh 38 - - - 6.00 5.45 50.1TOTAL 476.4 100.00 ** 110.1 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2690µm K80 = 80µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL WORK INDEX (Wi)BOND BALL SCREEN ANALYSIS231 Table F.10 Bond ball mill work index test results for drawpoint sample E23N collected Nov, 2015 BOND BALL GRINDABILITY TEST E23N Drawpoint sample (Nov 2015)1277 g Aperture Test Sieve : 106µm364.9 g Percent Undersize : 9.3%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1277.0 200 267.5 118.3 149.2 0.752 267.5 456 423.2 24.8 398.4 0.873 423.2 373 401.5 39.2 362.3 0.974 401.5 337 379.0 37.2 341.8 1.015 379.0 325 366.7 35.1 331.6 1.026 366.7 324 365.2 34.0 331.2 1.027 365.2 324 366.5 33.8 332.7 1.03BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 1.02 gP = 80% Passing size of test product. 81 µmF = 80% Passing size of test feed. 2779 µm17.9 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 53.40 10.00 90.0 - - -7 Mesh 2800 50.30 9.42 80.6 - - -8 Mesh 2360 64.80 12.13 68.5 - - -9 Mesh 2000 44.70 8.37 60.1 - - -10 Mesh 1700 44.60 8.35 51.7 - - -12 Mesh 1400 38.40 7.19 44.6 - - -14 Mesh 1180 26.70 5.00 39.6 - - -20 Mesh 850 40.90 7.66 31.9 - - -28 Mesh 600 32.30 6.05 25.9 - - -35 Mesh 425 24.70 4.62 21.2 - - -48 Mesh 300 19.30 3.61 17.6 - - -65 Mesh 212 16.80 3.14 14.5 - - -100 Mesh 150 15.30 2.86 11.6 - - -150 Mesh 106 12.50 2.34 9.3 4.70 3.85 96.2170 Mesh 90 - - - 10.80 8.84 87.3200 Mesh 75 - - - 15.60 12.77 74.5270 Mesh 53 - - - 18.50 15.14 59.4325 Mesh 45 - - - 8.30 6.79 52.6400 Mesh 38 - - - 6.70 5.48 47.1TOTAL 534.2 100.00 ** 122.2 100.00 ** Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2779µm K80 = 81µm232 Table F.11 Bond ball mill work index test results for drawpoint sample E30N collected Nov, 2015 BOND BALL GRINDABILITY TEST E30N Drawpoint sample (Nov 2015)1267.5 g Aperture Test Sieve : 106µm362.1 g Percent Undersize : 8.2%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1267.5 200 209.4 103.4 106.0 0.532 209.4 651 485.5 17.1 468.4 0.723 485.5 448 421.1 39.6 381.5 0.854 421.1 385 344.4 34.3 310.1 0.815 344.4 415 371.7 28.1 343.6 0.836 371.7 401 366.0 30.3 335.7 0.847 366.0 397 355.7 29.9 325.9 0.82BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.83 gP = 80% Passing size of test product. 78 µmF = 80% Passing size of test feed. 2831 µm20.8 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 47.80 9.80 90.2 - - -7 Mesh 2800 52.80 10.82 79.4 - - -8 Mesh 2360 64.70 13.26 66.1 - - -9 Mesh 2000 46.00 9.43 56.7 - - -10 Mesh 1700 42.80 8.77 47.9 - - -12 Mesh 1400 36.60 7.50 40.4 - - -14 Mesh 1180 24.60 5.04 35.4 - - -20 Mesh 850 37.30 7.64 27.7 - - -28 Mesh 600 28.50 5.84 21.9 - - -35 Mesh 425 20.80 4.26 17.6 - - -48 Mesh 300 15.10 3.09 14.5 - - -65 Mesh 212 12.30 2.52 12.0 - - -100 Mesh 150 10.60 2.17 9.9 - - -150 Mesh 106 8.30 1.70 8.2 3.40 3.41 96.6170 Mesh 90 - - - 7.90 7.92 88.7200 Mesh 75 - - - 11.20 11.22 77.5270 Mesh 53 - - - 13.50 13.53 63.9325 Mesh 45 - - - 6.40 6.41 57.5400 Mesh 38 - - - 5.20 5.21 52.3TOTAL 488.0 100.00 ** 99.8 100.00 ** Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2831µm K80 = 78µm233 Table F.12 Bond ball mill work index test results for drawpoint sample F33S collected Nov, 2015 BOND BALL GRINDABILITY TEST F33S Drawpoint sample (Nov 2015)1241.6 g Aperture Test Sieve : 106µm354.7 g Percent Undersize : 14.5%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1241.6 200 266.6 180.2 86.4 0.432 266.6 732 628.0 38.7 589.3 0.813 628.0 327 404.2 91.1 313.1 0.964 404.2 309 359.1 58.7 300.4 0.975 359.1 312 350.9 52.1 298.8 0.966 350.9 317 353.7 50.9 302.8 0.96BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.96 gP = 80% Passing size of test product. 74 µmF = 80% Passing size of test feed. 2667 µm17.9 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 36.30 7.31 92.7 - - -7 Mesh 2800 44.50 8.96 83.7 - - -8 Mesh 2360 61.80 12.44 71.3 - - -9 Mesh 2000 42.50 8.55 62.7 - - -10 Mesh 1700 40.50 8.15 54.6 - - -12 Mesh 1400 34.60 6.96 47.6 - - -14 Mesh 1180 22.90 4.61 43.0 - - -20 Mesh 850 36.20 7.29 35.7 - - -28 Mesh 600 28.90 5.82 29.9 - - -35 Mesh 425 22.30 4.49 25.4 - - -48 Mesh 300 17.20 3.46 22.0 - - -65 Mesh 212 14.60 2.94 19.0 - - -100 Mesh 150 12.70 2.56 16.5 - - -150 Mesh 106 9.70 1.95 14.5 2.50 2.58 97.4170 Mesh 90 - - - 7.10 7.33 90.1200 Mesh 75 - - - 9.50 9.81 80.3270 Mesh 53 - - - 10.70 11.05 69.2325 Mesh 45 - - - 4.90 5.06 64.2400 Mesh 38 - - - 3.90 4.03 60.1TOTAL 496.8 100.00 ** 96.8 100.00 **Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2667µm K80 = 74µm Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)234 Table F.13 Bond ball mill work index test results for drawpoint sample G40N collected Nov, 2015 BOND BALL GRINDABILITY TEST G40N Drawpoint sample (Nov 2015)1244.4 g Aperture Test Sieve : 106µm355.5 g Percent Undersize : 10.5%Weight of Number of Weight of UndersizeNew Feed Revolutions Product Feed Net Product Net/Rev1 1244.4 200 221.4 130.5 90.9 0.452 221.4 731 564.5 23.2 541.3 0.743 564.5 400 394.9 59.2 335.7 0.844 394.9 375 367.3 41.4 325.9 0.875 367.3 364 353.9 38.5 315.4 0.876 353.9 368 352.1 37.1 315.0 0.86BOND WORK INDEX FORMULA = / ^ ^ /√ /√ Pi = Sieve Size Tested. 106 µmGbp = Net undersize produced per revolution of mill. 0.86 gP = 80% Passing size of test product. 78 µmF = 80% Passing size of test feed. 2690 µm20.1 kw-hr/tonneWeight (g) Weight Cumulative Weight (g) Weight Cumulativemesh µm Retained % Retained % Passing Retained % Retained % Passing6 Mesh 3360 37.80 7.85 92.1 - - -7 Mesh 2800 44.00 9.14 83.0 - - -8 Mesh 2360 58.10 12.07 70.9 - - -9 Mesh 2000 42.40 8.81 62.1 - - -10 Mesh 1700 40.70 8.45 53.7 - - -12 Mesh 1400 35.70 7.41 46.3 - - -14 Mesh 1180 25.10 5.21 41.1 - - -20 Mesh 850 39.10 8.12 32.9 - - -28 Mesh 600 30.40 6.31 26.6 - - -35 Mesh 425 22.90 4.76 21.9 - - -48 Mesh 300 17.40 3.61 18.3 - - -65 Mesh 212 14.50 3.01 15.2 - - -100 Mesh 150 12.90 2.68 12.6 - - -150 Mesh 106 10.00 2.08 10.5 2.60 2.42 97.6170 Mesh 90 - - - 9.00 8.38 89.2200 Mesh 75 - - - 12.20 11.36 77.8270 Mesh 53 - - - 14.60 13.59 64.2325 Mesh 45 - - - 7.00 6.52 57.7400 Mesh 38 - - - 5.40 5.03 52.7TOTAL 481.5 100.00 ** 107.4 100.00 ** Weight of 700 ml Sample :1/3.5 of Sample Weight :CycleBOND BALL SCREEN ANALYSISBOND BALL WORK INDEX (Wi)Particle SizeFeed to Cycle 1 Equilibrium Cycle UndersizeK80 = 2690µm K80 = 78µm235 Drawpoint Logs Table G.1 Logs of rock types reporting to drawpoints Date Collected Units 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12Drawpoint E23N B13S D7S D11N D11S D38NEstimated HOD [m] 99 182 203 251 238 58Dominant rock type at drawpoint Monzonite Phyllic Hypogene Hypogene Hypogene SupergeneF100 [mm] 1,015 318 532 858 722 568F80 [mm] 390 98 213 256 381 121F50 [mm] 123 70 93 93 196 71IS50 Mean [MPa] 5.6 3.8 5.4 5.7 5.1 2.2Dropweight Axb 39.19 46.10 39.47 38.43 32.25 61.16ta 0.28 0.44 0.40 0.39 0.41 0.92DWi [kWh/m3] 6.91 5.82 6.86 6.91 8.31 4.45SG 31.5 x 26.5 2.72 2.69 2.72 2.67 2.70 2.72Average SG (+50 mm) 2.72 2.71 2.74 2.70 2.74 2.72Emin [kWh/t] 0.09 0.05 0.05 0.08 0.07 -0.07Bond Ball Mill Work Index [kWh/t] 17.93 18.77 17.99 18.38 19.29 18.55Lithology/ Alteration, average proportions% Group 1, Biotite >50 &/or Group 2* (picrite) [%] 22 32 46 36 46 46% Group 3, 50>K-spar>20 [%] 37 33 20 25 50 23% Group 5, K-spar >50 [%] 41 19 27 36 4 7% Group 6, Fault rock [%] 0 15 2 0 1 9% Group 7, Carbonates [%] 0 1 5 3 1 1% Group 8, Oxide [%] 0 0 0 0 0 14% Total Carbonates & Fault (Group 6 & Group 7) [%] 0 16 7 3 1 10'+50 mm % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 23 37 31 36 52 42'+50 mm % Group 3, 50>K-spar>20 [%] 32 27 26 19 41 23'+50 mm % Group 5, K-spar >50 [%] 45 8 35 42 5 6'+50 mm % Group 6, Fault rock [%] 0 25 3 0 2 13'+50 mm % Group 7, Carbonates [%] 0 2 5 3 0 0'+50 mm % Group 8, Oxide [%] 17'-50,+25 mm, % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 19 29 51 36 46 44'-50,+25 mm, % Group 3, 50>K-spar>20 [%] 42 35 16 20 49 29'-50,+25 mm, % Group 5, K-spar >50 [%] 38 17 27 42 4 8'-50,+25 mm, % Group 6, Fault rock [%] 0 19 4 0 0 5'-50,+25 mm, % Group 7, Carbonates [%] 0 0 1 2 0 0'-50,+25 mm, % Group 8, Oxide [%] 0 0 0 0 0 14-25,+ 12.5 mm, % % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 24 29 57 36 38 54-25,+ 12.5 mm, % % Group 3, 50>K-spar>20 [%] 37 37 17 35 59 18-25,+ 12.5 mm, % % Group 5, K-spar >50 [%] 40 33 18 25 4 7-25,+ 12.5 mm, % % Group 6, Fault rock [%] 0 0 0 0 0 8-25,+ 12.5 mm, % % Group 7, Carbonates [%] 0 2 8 4 2 3-25,+ 12.5 mm, % % Group 8, Oxide [%] 0 0 0 0 0 10Average shape for both size fractions) 2.72 2.07 2.28 2.00 2.33 1.99Shape +50 mm (1 =Round, 3 = Angular) 2.76 2.08 2.11 1.89 2.07 1.88Shape -50,+25 mm (1 =Round, 3 = Angular) 2.68 2.06 2.45 2.12 2.60 2.11236 Table G.2 Logs of rock types reporting to drawpoints * Note: Sample from drawpoint F33S was the only sample that contained noticeable picrite Date Collected Units 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12Drawpoint E13S E15N E30N F33S G40NEstimated HOD [m] 237 229 112 98 64Dominant rock type at drawpoint Hypogene Hypogene Hypogene Picrite 2nd. HypogeneF100 [mm] 990 707 709 495 494F80 [mm] 397 303 285 138 163F50 [mm] 190 135 122 73 86IS50 Mean [MPa] 5.3 5.8 6.3 3.2 3.6Dropweight Axb 34.83 33.50 38.80 67.46 40.89ta 0.36 0.27 0.38 0.66 0.50DWi [kWh/m3] 7.79 8.08 7.08 3.91 6.47SG 31.5 x 26.5 2.73 2.72 2.76 2.64 2.66Average SG (+50 mm) 2.73 2.72 2.73 2.66 2.66Emin [kWh/t] 0.00 0.07 -0.02 -0.03Bond Ball Mill Work Index [kWh/t] 19.28 19.40 20.77 17.88 20.11Lithology/ Alteration, average proportions% Group 1, Biotite >50 &/or Group 2* (picrite) [%] 60 57 51 56 27% Group 3, 50>K-spar>20 [%] 27 34 16 17 30% Group 5, K-spar >50 [%] 11 8 22 18 34% Group 6, Fault rock [%] 0 0 1 3 0% Group 7, Carbonates [%] 2 1 9 6 8% Group 8, Oxide [%] 0 0 0 0 0% Total Carbonates & Fault (Group 6 & Group 7) [%] 2 1 10 9 8'+50 mm % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 50 54 58 64 23'+50 mm % Group 3, 50>K-spar>20 [%] 38 40 17 21 32'+50 mm % Group 5, K-spar >50 [%] 12 6 18 10 42'+50 mm % Group 6, Fault rock [%] 0 0 3 3 0'+50 mm % Group 7, Carbonates [%] 0 0 3 3 4'+50 mm % Group 8, Oxide [%] 0 0 0 0'-50,+25 mm, % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 70 54 42 45 29'-50,+25 mm, % Group 3, 50>K-spar>20 [%] 26 32 32 15 29'-50,+25 mm, % Group 5, K-spar >50 [%] 4 12 19 25 32'-50,+25 mm, % Group 6, Fault rock [%] 0 0 1 0 0'-50,+25 mm, % Group 7, Carbonates [%] 0 2 6 14 11'-50,+25 mm, % Group 8, Oxide [%] 0 0 0 0-25,+ 12.5 mm, % % Group 1, Biotite >50 &/or Group 2* (picrite) [%] 60 62 53 59 30-25,+ 12.5 mm, % % Group 3, 50>K-spar>20 [%] 17 30 0 15 30-25,+ 12.5 mm, % % Group 5, K-spar >50 [%] 16 7 30 19 30-25,+ 12.5 mm, % % Group 6, Fault rock [%] 0 7 0-25,+ 12.5 mm, % % Group 7, Carbonates [%] 6 2 17 0 10-25,+ 12.5 mm, % % Group 8, Oxide [%] 0 0 0 0Average shape for both size fractions) 2.16 2.33 2.27 1.86 2.01Shape +50 mm (1 =Round, 3 = Angular) 2.08 2.19 2.53 1.62 1.87Shape -50,+25 mm (1 =Round, 3 = Angular) 2.24 2.46 2.01 2.11 2.14237 Figure G.1 Mill feed sample (~2 tonnes) collected March, 2016 and example of rock type groupings (+50 mm) Table G.3 Rock types identified in mill feed sample (collected March, 2016) SizeMass % of Mill Feed SampleWeight[mm] [%] Group #1 Group #3 Group #5 Group #6 Group #7-150, +125* 0.36-125, +100 4.92 50 50-100, +75 8.18 63 9 13 10 4-75, +63 5.10 47 30 17 1 5-63, +50 5.90 29 37 28 3 4Mass Weighted Average (-125, +50 mm) 48 29 15 4 3Lithology / Alteration (% by mass)238 Image-based Size Distributions Table H.1 Drawpoint size distributions from image-based measurements and sieving Date inspected Units 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12 2015, Nov 12Drawpoint E23N B13S D7S D11N D11S D38N E13S E15N E30N F33S G40NEstimated HOD [m] 99 182 203 251 238 58 237 229 112 98 64F100 (bold = measured with tape) [mm] 1015 318 532 858 722 568 990 707 709 495 494F80 [mm] 389.6 98.4 213.0 256.4 381.3 120.7 397.1 302.8 284.7 138.0 163.4F50 [mm] 122.5 69.7 93.2 93.2 196.2 70.8 189.8 134.6 122.1 73.5 86.4Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 100 1001000 92 100 100 100 100 100 100 100 100 100 100820 92 100 100 97 96 100 98 100 100 100 100800 92 100 100 97 96 100 97 100 100 100 100700 92 100 100 95 94 100 96 100 100 100 100600 86 100 100 92 91 99 90 95 99 99 98500 84 100 100 89 88 97 85 90 98 99 97400 81 100 97 89 82 97 80 90 91 99 97300 76 98 90 83 73 96 68 80 83 95 93250 70 96 84 79 63 95 61 74 74 91 89200 66 95 78 74 51 92 52 65 73 91 87150 57 90 68 67 34 88 41 56 61 83 77100 44 81 52 53 17 75 24 37 42 70 57Sieved -25 mm drawpoint samplesSize (mm)25 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.019 83.6 86.2 88.3 78.4 65.8 56.4 70.0 55.9 86.4 87.812.5 65.3 68.9 68.2 55.8 37.6 20.0 36.3 30.5 63.3 68.05.6 38.3 46.5 38.5 29.5 8.5 6.6 10.7 13.9 34.5 37.32 19.8 29.1 20.7 16.7 4.4 3.5 5.4 7.5 18.8 19.20.5 9.2 15.1 10.8 7.1 2.8 1.7 3.6 4.0 8.2 9.0% Passing from Wipfrag% Passing -239 Table H.2 Drawpoint size distributions from image-based measurements Table H.3 Drawpoint size distributions from image-based measurements Date inspected Units 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21Drawpoint E23S C36N C37N C39N D37N E11S E19N E30N E32N E38NEstimated HOD [m] 190 107 115 64 167 289 272 168 179 143F100 (bold = measured with tape) [mm] 160 280 280 520 490 300 580 310 270 210F80 [mm] 142.5 143.5 139.1 228.3 149.9 146.6 212.0 215.7 171.0 140.6F50 [mm] 124.7 123.5 107.2 126.8 123.9 122.1 129.5 122.6 128.7 110.8Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 1001000 100 100 100 100 100 100 100 100 100 100820 100 100 100 100 100 100 100 100 100 100800 100 100 100 100 100 100 100 100 100 100700 100 100 100 96 100 100 96 100 100 100600 100 100 100 96 96 100 96 96 100 100500 100 100 100 91 93 100 89 96 100 100400 100 100 100 88 90 100 87 93 100 100300 100 98 100 83 88 98 83 86 93 100250 100 96 100 82 87 94 81 83 90 99200 97 94 95 77 84 90 80 79 85 95150 93 90 90 73 80 84 71 74 76 90100 8 15 43 23 22 23 19 30 15 39% PassingDate inspected Units 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2016, Nov 21 2014, Mar 27 2014, Mar 27Drawpoint E39S E40S E43N E43S G33S G34S G40N G40S B9S C7SEstimated HOD [m] 127 23 38 33 60 59 64 52 75 145F100 (bold = measured with tape) [mm] 390 570 430 480 390 180 520 270 405 594F80 [mm] 174.6 149.9 205.9 188.1 147.4 139.8 140.3 142.2 133.8 236.1F50 [mm] 126.7 126.7 128.1 122.6 119.3 117.8 110.3 116.3 71.5 94.7Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 1001000 100 100 100 100 100 100 100 100 100 100820 100 100 100 100 100 100 100 100 100 100800 100 100 100 100 100 100 100 100 100 100700 100 100 100 100 100 100 100 100 100 100600 100 93 100 100 100 100 100 100 100 100500 100 93 97 100 96 100 100 97 100 94400 96 93 94 99 96 98 95 97 99 90300 91 91 89 93 94 97 95 97 98 88250 87 86 85 91 91 97 95 96 95 82200 84 85 79 82 89 96 95 94 91 74150 76 80 69 73 83 94 90 89 84 64100 20 15 26 31 29 26 40 31 72 52% Passing240 Table H.4 Drawpoint size distributions from image-based measurements Table H.5 Drawpoint size distributions from image-based measurements Date inspected Units 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27 2014, Mar 27Drawpoint C8N E9S E10S G35N E10N - Berm E12S - Berm E31N - Berm (OK) E31S - Blast (OK) E33N - (OK) E34S - BermEstimated HOD [m] 95 155 158 34 157 145 35 40 26 25F100 (bold = measured with tape) [mm] 367 896 820 706 251 261 552 645 235 527F80 [mm] 224 347 257 172 80 91 79 507 80 261F50 [mm] 148.9 128.8 74.8 73.5 67.5 68.8 68.1 334.4 68.4 75.6Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 1001000 100 100 100 100 100 100 100 100 100 100820 100 96 100 100 100 100 100 100 100 100800 100 95 95 100 100 100 100 100 100 100700 100 95 90 96 100 100 100 100 100 100600 100 92 90 96 100 100 100 96 100 100500 100 89 90 96 100 100 95 79 100 93400 100 84 86 96 100 100 95 66 100 87300 95 77 83 92 100 100 95 42 100 83250 88 71 80 89 98 99 95 29 100 79200 73 61 76 85 95 95 95 17 98 74150 51 53 69 76 92 90 93 8 92 69100 17 46 63 66 85 81 86 2 84 62% PassingDate inspected Units 2014, Mar 27 2014, Mar 27 2012, Jan 13 2013, Nov 4 2013, Nov 4 2012, Oct 31 2012, Apr 19 2013, Nov 4 2013, Nov 4 2013, Nov 4Drawpoint F19S - Berm (OK) F21S - Berm C7S F8N F10N F10N F10N C5S C8S C10SEstimated HOD [m] 114 98 65 116 130 70 33 129 123 120F100 (bold = measured with tape) [mm] 432 315 476 921 728 649 511 997 775 652F80 [mm] 126 184 181 430 350 278 293 589 354 131F50 [mm] 71.0 129.3 91.9 209.9 201.2 85.7 128.8 196.5 158.7 69.9Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 1001000 100 100 100 100 100 100 100 100 100 100820 100 100 100 96 100 100 100 81 100 100800 100 100 100 96 100 100 100 81 100 100700 100 100 100 93 97 100 100 81 92 100600 100 100 100 90 97 97 100 81 92 98500 100 100 100 84 94 90 98 72 88 95400 96 100 96 78 88 87 91 70 85 94300 93 92 94 64 72 81 81 62 74 89250 91 92 90 57 62 79 74 57 69 88200 88 89 84 48 50 74 68 51 60 86150 84 62 73 40 40 68 57 43 48 83100 76 33 54 28 29 55 40 34 29 76% Passing241 Table H.6 Drawpoint size distributions from image-based measurements Table H.7 Drawpoint size distributions from image-based measurements Date inspected Units 2013, Nov 4 2013, Nov 4 2012, Feb 24 2012, Dec 11 2012, Oct 17 2012, Mar 22 2012, Jun 27 2012, Apr 19 2012, June 27 2012, May 3Drawpoint B11S B12S C5S C6S C7S D7S D11S E10N F10S G11N Estimated HOD [m] 52 50 43 72 54 32 24 20 42 10F100 (bold = measured with tape) [mm] 384 626 765 741 715 1029 568 581 662 537F80 [mm] 104 256 391 346 397 527 155 140 269 197F50 [mm] 69.4 74.7 170.6 86.7 196.0 252.3 75.0 74.6 96.2 90.1Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 100 100 1001000 100 100 100 100 100 90 100 100 100 100820 100 100 100 100 100 90 100 100 100 100800 100 100 100 100 100 90 100 100 100 100700 100 100 93 96 94 90 100 100 100 100600 100 97 91 96 94 82 100 100 96 100500 100 93 88 93 94 79 98 97 93 97400 100 89 81 83 80 69 98 95 90 94300 97 85 69 77 69 60 95 94 82 89250 97 79 62 74 60 50 91 92 79 86200 93 76 55 67 51 42 86 91 71 81150 87 70 46 61 40 30 79 83 64 70100 79 64 35 53 29 18 66 67 51 54% PassingDate inspected Units 2012, Dec 11 2012, Nov 16 2012, Oct 31 2012, Dec 11 2013, Jan 24 2012, Nov 16 2012, Oct 31 2012, Dec 11 2012, Dec 11 2012, Oct 17Drawpoint B9S B9S B9S B11S F19S B11S B11S B12S C5S C5SEstimated HOD [m] 17 15 13 14 21 11 9 13 79 73F100 (bold = measured with tape) [mm] 419 2046 566 1028.0 971 574 310 307 492 651F80 [mm] 270 950 238 275 796 481 275 142 267 421F50 [mm] 139.2 139.8 136.5 134.7 415.2 309.4 131.0 125.8 140.7 251.3Image-based measurementSize (mm)2000 100 98 100 100 100 100 100 100 100 1001500 100 81 100 100 100 100 100 100 100 1001200 100 81 100 100 100 100 100 100 100 1001000 100 81 100 95 100 100 100 100 100 100820 100 78 100 93 83 100 100 100 100 100800 100 77 100 92 80 100 100 100 100 100700 100 77 100 92 76 100 100 100 100 100600 100 77 100 90 70 100 100 100 100 96500 100 77 98 87 63 84 100 100 100 85400 98 77 96 87 48 65 100 100 93 79300 85 72 88 81 26 48 86 99 83 57250 77 69 81 79 14 36 74 98 79 50200 70 65 76 76 7 20 71 96 71 41150 62 63 67 71 3 7 62 95 60 32100 6 0 3 2 0 2 30 2 6 4% Passing242 Table H.8 Drawpoint size distributions from image-based measurements Date inspected Units 2012, July 11 2012, June 27 2012, May 31 2012, May 3 2012, Oct 17 2012, Feb 02 2012, Dec 11 2012, Oct 17Drawpoint C5S C5S C5S C5S E9S F8N C8N B9SEstimated HOD [m] 63 61 54 50 61 13 32 12F100 (bold = measured with tape) [mm] 637 846 929 799 431 330 384 1210F80 [mm] 599 486 703 344 272 78 195 494F50 [mm] 384.8 138.6 144.8 139.7 180.1 67.7 135.1 278.6Image-based measurementSize (mm)2000 100 100 100 100 100 100 100 1001500 100 100 100 100 100 100 100 1001200 100 100 100 100 100 100 100 901000 100 100 100 100 100 100 100 90820 100 97 88 100 100 100 100 90800 100 94 88 100 100 100 100 90700 100 90 80 97 100 100 100 90600 80 87 80 95 100 100 100 87500 58 80 80 91 100 100 100 81400 52 77 69 85 94 100 100 72300 39 72 67 76 85 98 90 54250 35 70 64 72 76 96 86 44200 29 67 61 67 59 96 81 31150 21 64 56 63 37 93 71 15100 8 2 2 0 7 87 2 3% Passing243 Mill Survey Results for Survey Carried Out March 3rd, 2015 (survey used for model validation) Table I.1 Summary of samples taken March 3rd, 2015 Samples collected on: March 3rd, 2015 12:45 to 1:15 AM (30 minute span)SAG Mill Feed taken at ~2AM just prior to mill shut-downGross Weight Packaging weight Net Sample Weight - Wet Net Sample Weight - Dry % Solids % Moisture[kg] [kg] [kg] [kg]SAG Mill Feed 667.8 30.8 637.0 620.2 97.4 2.6Cyclone Underflow #1/2 35.3 1.3 34.0 27.7 81.3 18.7Cyclone Underflow #2/2 36.2 1.3 34.9 28.7 82.3 17.7Cyclone Underflow (Total) 71.5 2.6 68.9 56.4 81.8 18.2SAG Mill Screen Over-Size 53.4 2.3 51.1 50.5 98.8 1.2Cyclone Overflow #1/2 20.4 1.3 19.1 7.1 37.1 62.9Cyclone Overflow #2/2 21.4 1.3 20.1 7.7 38.2 61.8Cyclone Overflow (Total) 41.8 2.6 39.2 14.7 37.6 62.4244 Table I.2 Summary of samples taken March 3rd, 2015 * Note: mill F80 feed size was finer than expected. Potentially due to size segregation in mill feed stockpile Survey Date: March 3rd, 2015. 1 to 3 AMSpecific Energy Power[kWh/mt] [kW]SAG mill main motor 6.64 4,324Pebble circuit conveyors S1-S4 0.16 101Pebble crusher 0.35 228SAG mill screen 0.03 20Ball mill main motor 8.37 5,446Slurry pump motor 0.78 508Total: 16.32 10,628Solids throughput rate% Moisture (measured, wt/wt)% Solids (measured, wt/wt)F80 F50[mtph] [%] [%] [µm] [µm]SAG Mill Feed (S1) 651 2.6 97.4 34,583SAG Mill Screen Over-Size 151 1.2 98.8 42,748 25,837Ball Mill Cyclone Underflow 18.2 81.8 4,004 1,004Ball Mill Cyclone Overflow 651 62.4 37.6 226 78ItemStream:245 Table I.3 Size distributions for samples taken March 3rd, 2015 Stream: Pebble Crusher Feed Cyclone Underflow Cyclone OverflowCumulative % Passing Cumulative % Passing Cumulative % Passing Cumulative % Passing[mm] [µm] [%] [%] [%] [%]150 150,000 100.0 100.0 100.0 100.0125 125,000 99.4 100.0 100.0 100.0100 100,000 98.1 100.0 100.0 100.075 75,000 95.8 98.5 100.0 100.050 50,000 88.2 87.9 100.0 100.037.5 37,500 82.2 74.0 100.0 100.025 25,000 71.7 47.8 100.0 100.019 19,000 64.5 33.1 100.0 100.013.2 13,200 52.1 15.3 100.0 100.09.5 9,500 45.1 1.4 99.6 100.06.3 6,300 36.4 0.1 88.4 100.04.75 4,750 31.6 0.0 82.2 100.03.35 3,350 26.3 0.0 77.7 100.02.63 2,630 21.3 0.0 71.6 100.01.7 1,700 18.0 0.0 61.7 100.01.4 1,400 16.4 0.0 57.8 100.01 1,000 12.9 0.0 49.9 100.00.71 710 10.2 0.0 41.2 100.00.5 500 8.3 0 30.1 99.10.355 355 6.2 0 18.9 93.90.25 250 4.8 0 11.9 83.70.18 180 4.3 0 8.0 72.50.15 150 3.9 0 6.8 66.50.125 125 3.6 0 5.9 61.30.09 90 3.1 0 4.6 53.20.063 63 2.7 0 3.7 45.50.045 45 2.3 0 3.0 39.1-0.045 -45Total:P80 [mm] 34.58 42.75 4.00 0.226P50 [mm] 12.03 26.05 1.00 0.08SizeSAG Mill Feed246 Figure I.1 Size distributions for samples taken March 3rd, 2015 247 Mill Survey Results for Survey Carried Out March 16th, 2016 (survey used for model calibration) Table J.1 Summary of samples taken March 16th, 2016 Stream Time & Date Gross WeightPackaging weightNet Sample Weight - WetNet Sample Weight - Dry % Solids % Moisture[kg] [kg] [kg] [kg]SAG Mill Feed (S1) 2016-03-16 1039 2,237.0 0.0 2,237.0 2,156.0 96.4 3.6Pebble Crusher Feed 2016-03-16 1405 335.5 12.0 323.5 319.1 98.6 1.4SAG Mill Discharge #1/2 2016-03-16 1240 - 1340 30.9 1.1 29.8 0.0 100.0SAG Mill Discharge #2/2 2016-03-16 1240 - 1340 33.1 1.1 32.0 0.0 100.0SAG Mill Discharge Total 64.0 2.2 61.8 49.1 79.5 20.5Ball Mill Discharge 2016-03-16 1240 - 1340 30.9 1.1 29.8 20.1 67.4 32.6BM Cyclone Underflow 2016-03-16 1240 - 1340 37.3 1.1 36.2 26.8 74.1 25.9BM Cyclone Overflow 2016-03-16 1240 - 1340 17.7 1.1 16.6 6.1 36.6 63.4VertiMill Product 2016-03-16 1240 - 1340 34 1.1 32.9 22.3 67.8 32.2VertiMill Cyclone Underflow 2016-03-16 1240 - 1340 30.4 1.1 29.3 19.9 67.9 32.1VertiMill Cyclone Overflow 2016-03-16 1240 - 1340 22 1.1 20.9 7.4 35.2 64.8248 Table J.2 Size distributions of samples taken March 16th, 2016 Sample: SAG Mill Feed *SAG Mill Discharge - from mass balance SAG Mill Undersize - from mass balanceCumulative % Passing Cumulative % Passing Cumulative % Passing[mm] [µm] [%] [%]1,000 1,000,000 100 100 100849 848,528 100.00 100 100600 600,000 100.00 100 100424 424,264 100.00 100 100300 300,000 100.00 100 100212 212,132 100.00 100 100150 150,000 100.00 100 100125 125,000 99.64 100 100100 100,000 94.71 100 10075 75,000 86.54 100 10063 63,000 81.43 99 10053 53,000 76.98 99 10045 45,000 72.85 98 10037.5 37,500 68.55 97 10026.5 26,500 59.11 93 10019 19,000 51.82 89 10013.2 13,200 43.60 84 1009.5 9,500 36.68 81 996.3 6,300 29.40 75 944.75 4,750 26.06 69 873.35 3,350 22.32 64 802.63 2,630 20.00 60 761.7 1,700 16.23 53 661.4 1,400 14.82 50 631 1,000 12.66 44 560.71 710 11.11 40 500.5 500 10.01 35 440.355 355 8.94 31 390.25 250 8.00 27 340.18 180 7.16 24 300.15 150 6.72 22 280.125 125 6.31 21 260.09 90 5.62 18 220.063 63 5.04 16 20P80 [mm] 59.7 8.82 3.31P50 [mm] 17.6 1.42 0.72Particle Size249 Table J.3 Size distributions of samples taken March 16th, 2016 Sample: Pebbles Ball Mill Discharge Ball Mill Cyclone Underflow Ball Mill Cyclone OverflowCumulative % Passing Cumulative % Passing Cumulative % Passing Cumulative % Passing[mm] [µm] [%] [%] [%] [%]1,000 1,000,000 100 100 100 100849 848,528 100 100 100 100600 600,000 100 100 100 100424 424,264 100 100 100 100300 300,000 100 100 100 100212 212,132 100 100 100 100150 150,000 100 100 100 100125 125,000 100 100 100 100100 100,000 100 100 100 10075 75,000 99 100 100 10063 63,000 96 100 100 10053 53,000 94 100 100 10045 45,000 89 100 100 10037.5 37,500 83 100 100 10026.5 26,500 66 100 100 10019 19,000 46 100 100 10013.2 13,200 24 100 100 1009.5 9,500 10 100 100 1006.3 6,300 1 99 96 1004.75 4,750 1 96 91 1003.35 3,350 0 94 86 1002.63 2,630 0 91 81 1001.7 1,700 0 85 72 1001.4 1,400 0 82 67 1001 1,000 0 75 58 1000.71 710 0 66 47 990.5 500 0 53 33 950.355 355 0 41 23 870.25 250 0 33 16 760.18 180 0 26 12 650.15 150 0 24 11 600.125 125 0 21 10 550.09 90 0 18 8 470.063 63 0 16 7 41P80 [mm] 35.27 1.256 2.505 0.289P50 [mm] 20.46 0.459 0.792 0.102Particle Size250 Table J.4 Size distributions of samples taken March 16th, 2016 Sample: Vertimill Product Vertimill Cyclone Underflow Vertimill Cyclone OverflowCumulative % Passing Cumulative % Passing Cumulative % Passing[mm] [µm] [%] [%] [%]1,000 1,000,000 100 100 100849 848,528 100 100 100600 600,000 100 100 100424 424,264 100 100 100300 300,000 100 100 100212 212,132 100 100 100150 150,000 100 100 100125 125,000 100 100 100100 100,000 100 100 10075 75,000 100 100 10063 63,000 100 100 10053 53,000 100 100 10045 45,000 100 100 10037.5 37,500 100 100 10026.5 26,500 100 100 10019 19,000 100 100 10013.2 13,200 100 100 1009.5 9,500 100 100 1006.3 6,300 100 100 1004.75 4,750 100 100 1003.35 3,350 100 100 1002.63 2,630 100 100 1001.7 1,700 100 100 1001.4 1,400 100 100 1001 1,000 99 99 1000.71 710 99 98 1000.5 500 95 92 1000.355 355 84 78 990.25 250 66 55 950.18 180 47 35 870.15 150 39 28 800.125 125 33 23 730.09 90 26 18 620.063 63 0 0 0P80 [mm] 0.328 0.378 0.150P50 [mm] 0.191 0.234 -Particle Size251 Figure J.1 Size distributions for the SAG mill circuit (March, 2016) Figure J.2 Size distributions for the SAG mill circuit (March, 2016) 252 Calculation of SAG Mill Screen Undersize The SAG mill screen undersize stream could not be sampled directly at New Afton. However, the size distribution could be calculated from either of two mass balances: (1) mass balance around the SAG product screen, and (2) mass balance around the ball mill cyclone. Both mass balances were used to determine the final SAG screen undersize value. Figure J.3 Mass balances relating to SAG mill screen undersize calculation Calculation of the two mass balances in Excel provided an average of 3,289 microns (3,115 for method 1 and 3,460 for method 2). In JKSimMet, the mass balance feature was used by weighting each measured value according to the confidence in the measurement method. Lower weighting was assigned to the SAG and ball mill discharge samples than the other streams (eg. cyclone overflow & underflow) which could be cut properly with samplers. Based on the JKSimMet balance calculation, a screen undersize P80 value of 3,311 microns was determined and nominated as the ball mill fresh feed size 80% passing size. 253 Mass balance #1: Overall SAG mill discharge (SDIS) size distribution was calculated based on: 1) SAG mill discharge sampled with sampler (49 kg of solids). The sampler opening was about 150 mm in diam. 2) Screen oversize (319 kg of solids) diverted into a bin for collection 3) Dry throughput of fresh feed to mill (percent solids from moisture level of collected+2 tonne sample and wet throughput from belt scale on the mill feed conveyor) 4) Dry throughput of SAG mill screen oversize (wet mtph from belt scale on recirc. conveyor and % solids from moisture level of collected 324 kg wet sample) For +13.2 mm (effective aperture size of screen) size fractions, the pebble size distribution was used and weighted by tph from belt scale on the recirculation conveyor. Note: the SAG mill screen uses slotted apertures. Since there was no +13.2 mm material in the cyclone underflow and there was a steep drop at fractions below 13.2 mm in the pebble size distribution, a value of 13.2 mm was nominated as the effective square aperture size For -13.2 mm size fractions, the sampler size distribution was used and weighted by Fresh Feed tph + oversize tph. In the next section, capitals refer to solids throughput and lower case refers to percentage passing sieve size x. The following is now available to find the size of the SAG mill screen undersize, us(x): Table J.5 Outline of measured and calculated streams Stream Solids MTPH % passing of sieve size x Fresh Feed FF ff(x) SAG Mill Discharge SDIS sdis(x) Screen Oversize OS os(x) Screen Undersize US (equals FF) 254 Mass Balance Equation #1 around screen: Figure J.4 Mass balance calculation #1 for SAG mill screen undersize IN = SDIS.sdis(x) = OUT = US.us(x) + OS.os(x), (J.1) Solving for the unknown % passing size x for undersize, us(x): us(x) = (SDIS.sdis(x) -OS.os(x)) / US note: US = mill fresh feed tph (J.2) Mass balance #2, Around cyclone The following measurements were referenced: 1) Ball mill discharge sampled with sampler (20 kg solids) and sieved 2) Ball mill cyclone overflow and underflow samples (27 and 6 kg solids weight, respectively) 3) Mill solids fresh feed rate, FF (as in previous calc) 4) Cyclone feed tph measured online (nuclear density gauge and vol. flow rate). 255 The following is now available to find the size of the SAG mill screen undersize us(x): Stream Solids MTPH % passing of sieve size x Fresh feed FF ff(x) Ball Mill Discharge BDIS bdis(x) Cyclone Feed CF (from DCS) Cyclone Underflow CU (from CF – FF) cu(x) Cyclone Overflow CO (equals FF) co(x) Screen Undersize US (equals FF) Figure J.5 Mass balance calculation #2 for SAG mill screen undersize Mass Balance Equation #2 around cyclone: IN = CF.cf(x) = OUT = OF.of(x) + UF.uf(x) (J.3) Since, CF.cf(x) = BDIS.bdis(x) + US.us(x) (J.4) BDIS.bdis(x) + US.us(x) = OF.of(x) + UF.uf(x) (J.5) Therefore, us(x) = (OF.of(x) + UF.uf(x)-BDIS.bdis(x))/US (J.6) 256 Resulting size distributions for the two separate calculations are shown in Figure J.6. Figure J.6 Size distributions SAG mill screen undersize using calculations methods #1 and #2 257 Sample Preparation Flowsheet for PGNAA Testing Figure K.1 Sample preparation procedure for sulfide samples (applied to eight of nine samples) 258 Figure K.2 Sample preparation procedure for oxide sample (applied to one of nine samples)