@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Applied Science, Faculty of"@en, "Mining Engineering, Keevil Institute of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Erdenebat, Elberel"@en ; dcterms:issued "2017-09-05T15:42:16Z"@en, "2017"@en ; vivo:relatedDegree "Master of Applied Science - MASc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """With extraction of low-grade and high throughput deposits, elimination of tonnes of uneconomic material is highly desired to reduce energy consumption and water usage in the mine/mill production cycle. Even though technologies such as sensor-based sorting has wide application for pre-concentration purposes, effectiveness of sorter systems and key parameters for sortability of a material are still in the developmental stage. Number of factors such as grade variability, mineralogical alteration and ore blending scenarios during material handling will significantly affect contents of a material resulting in unforeseen changes in downstream processes. For these reasons, the ‘ore heterogeneity’ parameter is studied to evaluate sortability of an ore material under varying mine production scenarios. Production data, drillhole data and representative drawpoint samples were provided from the New Afton copper-gold mine located near Kamloops, BC. The New Afton mine utilizes the block caving method for extraction of ore from the copper-gold alkali-porphyry deposit. The distribution heterogeneity (DH) parameter is estimated for the data sets and the quantity of potentially removable material ahead of delivery to mill is studied. The DH is defined by variation of grade of a group of samples that constitute a lot, i.e. a group being an equal tonnage of material drawn from a drawpoint and the lot being the drawpoint. With this approach, the DH is analyzed across drawpoints, vertically within a drawpoint and along drill holes with changing vertical intervals of 0.5m – 10m. The DH values are compared with copper grades and an inverse relationship is found. This finding indicated that sortability of ore material can be defined by a heterogeneity parameter, especially the information can be obtained earlier from drillcore samples. The drillcore information can indicate a measure of heterogeneity and related copper grade of an in-situ material in advance of assay samples or sensor detection where a certain degree of mixing has occurred. Overall, 27% of the sample data from the New Afton historical production record has grades less than 0.4% Cu, which is the current cut-off grade, and it correlates with relatively high heterogeneity and presents an opportunity for sorting and feed grade upgrade."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/62969?expand=metadata"@en ; skos:note " STUDY OF NEW AFTON ORE HETEROGENEITY AND ITS AMENABILITY TO SENSOR BASED ORE SORTING by Elberel Erdenebat B.Sc., National University of Mongolia, 2012 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Mining Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2017 © Elberel Erdenebat, 2017 ii Abstract With extraction of low-grade and high throughput deposits, elimination of tonnes of uneconomic material is highly desired to reduce energy consumption and water usage in the mine/mill production cycle. Even though technologies such as sensor-based sorting has wide application for pre-concentration purposes, effectiveness of sorter systems and key parameters for sortability of a material are still in the developmental stage. Number of factors such as grade variability, mineralogical alteration and ore blending scenarios during material handling will significantly affect contents of a material resulting in unforeseen changes in downstream processes. For these reasons, the ‘ore heterogeneity’ parameter is studied to evaluate sortability of an ore material under varying mine production scenarios. Production data, drillhole data and representative drawpoint samples were provided from the New Afton copper-gold mine located near Kamloops, BC. The New Afton mine utilizes the block caving method for extraction of ore from the copper-gold alkali-porphyry deposit. The distribution heterogeneity (DH) parameter is estimated for the data sets and the quantity of potentially removable material ahead of delivery to mill is studied. The DH is defined by variation of grade of a group of samples that constitute a lot, i.e. a group being an equal tonnage of material drawn from a drawpoint and the lot being the drawpoint. With this approach, the DH is analyzed across drawpoints, vertically within a drawpoint and along drill holes with changing vertical intervals of 0.5m – 10m. The DH values are compared with copper grades and an inverse relationship is found. This finding indicated that sortability of ore material can be defined by a heterogeneity parameter, especially the information can be obtained earlier from drillcore iii samples. The drillcore information can indicate a measure of heterogeneity and related copper grade of an in-situ material in advance of assay samples or sensor detection where a certain degree of mixing has occurred. Overall, 27% of the sample data from the New Afton historical production record has grades less than 0.4% Cu, which is the current cut-off grade, and it correlates with relatively high heterogeneity and presents an opportunity for sorting and feed grade upgrade. iv Lay Summary The objective of this research study was to evaluate the potential of removing waste material in a mine-mill production which will contribute in reducing energy and water usage. To evaluate this possibility of material removal with boundaries by copper grade, a particular parameter that has a correlation with copper grade was needed and thereby ‘ore heterogeneity’ was evaluated. Ore heterogeneity evaluation defines the degree of variability of a mineral or a metal within a given group of samples or a lot and if there is high heterogeneity, the metal/mineral content is widely dispersed throughout the lot and it will be amenable to segregation. In addition, it is found that when the average copper grade decreases, heterogeneity increases and the amount of material that can be eliminated through separation method also increases. v Preface This study is part of a Cave-to-Mill project at UBC NBK Institute of Mining Engineering supported by the New Afton Mine and Natural Science and Engineering Research Council of Canada. The objective of the research is to study ore heterogeneity at the New Afton Mine across numerous dimensions of draw points through historical production data along with experimental data and its application to sensor-based sorting. The approaches developed in this thesis are novel and can be applied to other mining operations. Under the supervision of Dr. Bern Klein, Professor at NBK Institute of Mining Engineering at UBC, I was responsible for the design of test program together with laboratory test works and interpretation of test results. Mr. Stefan Nadolski, PhD candidate at NBK Institute of Mining Engineering at UBC, assisted with the design of test program. vi Table of Contents Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ......................................................................................................................... vi List of Tables ................................................................................................................................ ix List of Figures .................................................................................................................................x List of Symbols ........................................................................................................................... xiii List of Abbreviations ................................................................................................................. xiv Acknowledgements ......................................................................................................................xv Chapter 1: Introduction ................................................................................................................1 1.1 Purpose of the Research .................................................................................................. 1 1.2 Research Methodology ................................................................................................... 2 1.3 New Afton mine .............................................................................................................. 2 1.3.1 Brief Introduction........................................................................................................ 2 1.3.2 Afton Mine Orebody ................................................................................................... 3 1.3.3 Orebody Mineralization .............................................................................................. 4 1.4 Ore Sorting ...................................................................................................................... 6 Chapter 2: Literature Review .......................................................................................................9 2.1 Ore Heterogeneity ......................................................................................................... 10 2.2 Sensor-Based Ore Sorting ............................................................................................. 13 2.3 X-Ray Fluorescence ...................................................................................................... 19 vii 2.4 Electromagnetic Conductivity & Permeability ............................................................. 19 2.4.1 ConductOreTM ........................................................................................................... 20 Chapter 3: Ore Sorting Test .......................................................................................................21 3.1 Sample Preparation ....................................................................................................... 21 3.1.1 Numbering and Labeling .......................................................................................... 21 3.1.2 Lithology Group........................................................................................................ 23 3.1.3 Crushing and Pulverizing .......................................................................................... 26 3.2 X-Ray Fluorescence ...................................................................................................... 28 3.3 Separation by Magnetic Properties ............................................................................... 29 3.3.1 Electromagnetic Spectroscopy .................................................................................. 29 3.3.2 Magnetic Susceptibility ............................................................................................ 32 3.3.3 Magnetic Pull ............................................................................................................ 33 3.3.4 Rare Earth Magnetic Drum Separator....................................................................... 34 Chapter 4: Heterogeneity Analysis.............................................................................................35 4.1 Overview ....................................................................................................................... 35 4.2 Results of Heterogeneity Analysis ................................................................................ 36 4.2.1 Production Data ........................................................................................................ 36 4.2.1.1 Grade Distribution ............................................................................................ 37 4.2.1.2 Distribution Heterogeneity................................................................................ 38 4.2.1.3 Mass Pull – Recovery - %Cu relationship ........................................................ 41 4.2.1.4 Cut-Off grade scenario ...................................................................................... 44 4.2.2 Borehole Data ........................................................................................................... 45 4.2.2.1 Heterogeneity by Intervals ................................................................................ 46 viii 4.2.2.2 Mass Pull – Recovery ....................................................................................... 48 4.2.2.3 Rock samples .................................................................................................... 50 4.2.2.4 Results of Rock and Pulp Analysis ................................................................... 53 Chapter 5: Conclusion .................................................................................................................56 5.1 Conclusion .................................................................................................................... 56 5.2 Future Work .................................................................................................................. 57 Reference List ...............................................................................................................................58 Appendices ....................................................................................................................................60 Appendix A - New Afton Mine Historical Assay Data ............................................................ 60 A.1 Production Data ........................................................................................................ 60 A.2 Borehole Data ........................................................................................................... 61 Appendix B Rock Test Data ..................................................................................................... 63 B.1 X-Ray Fluorescence .................................................................................................. 63 B.2 High-Frequency Electromagnetic Spectroscopy ....................................................... 64 ix List of Tables Table 1-1 Percentage of Ore Type (New Afton Technical Report, 2015) ...................................... 6 Table 2-1 Typical Criteria for Discerning Ore from Waste (Bamber. A., Klein. B., Scoble. M., 2006) ............................................................................................................................................. 15 Table 3-1 List of Draw Points for Samples .................................................................................. 22 Table 3-2 Lithology Group Type .................................................................................................. 24 Table 4-1 Example of Production Data ........................................................................................ 35 Table 4-2 Cut-Off Grade Scenario ................................................................................................ 44 Table 4-3 Example Drill Hole Assay Data ................................................................................... 45 Table 4-4 Data for Drillhole DH across Short Vertical Intervals ................................................. 46 Table 4-5 DH across Drillhole Intervals ....................................................................................... 49 Table 4-6 Rock vs Pulp Copper Grade by Drawpoint .................................................................. 54 x List of Figures Figure 1-1 Mine Area Plan View (New Afton Technical Report, 2015) ........................................ 3 Figure 1-2 Level Plan of the Geology at the 5150m Elevation, New Afton Mine, Kamloops, BC (New Afton Technical Report, 2015) ............................................................................................. 4 Figure 1-3 Cross Section Views of the Lithology, Alteration and Mineralization Models (New Afton Technical Report, 2015) ....................................................................................................... 5 Figure 2-1 Components of a Sortability Study ............................................................................... 9 Figure 2-2 Schematic Illustrating Midscale Heterogeneity (courtesy MineSense Technologies Ltd. 2014) ...................................................................................................................................... 12 Figure 2-3 Grade-recovery curve, CH-10 (Mazhary. A. & Klein. B., 2015) ............................... 13 Figure 2-4 Schematic Representation of the Four Aspects of the Ore-Sorting Process ............... 14 Figure 2-5 Metso Bulk Ore Sorting Concept ................................................................................ 16 Figure 2-6 Overhead Scintillometers (L) used for scanning of U308 bearing trucks (R) at Rossing Uranium, Namibia (Nield, 2002) .................................................................................................. 17 Figure 2-7 MineSense ShovelSenseTM shovel-based bulk batch sorting solution at a BC Cu-Au mine (Bamber et al, 2016) ............................................................................................................ 18 Figure 2-8 Bulk diversion of nickel laterites via XRF analysis to product stockpile (right) and waste stockpile (left) ..................................................................................................................... 18 Figure 3-1 Sample Collection from Muck Pile at Drawpoint B13S ............................................. 21 Figure 3-2 Draw Point Layout of East Cave (courtesy of New Afton mine) ............................... 22 Figure 3-3 Picrite Rock Sample Figure 3-4 Monzonite Rock Sample ..................... 24 xi Figure 3-5 Diorite Rock Sample Figure 3-6 Fault Rock Sample ........................................... 25 Figure 3-7 Sympathetic fields generated by low concentrations of conductive or magnetic mineral (Klein.B & Bamber. A. SME Mineral Processing Handbook, Mineral Sorting, 2017) .. 29 Figure 3-8 ConductOre Unit (courtesy of Minesense Technologies Ltd) .................................... 30 Figure 3-9 Magnetic Susceptibility - Electromagnetic Spectroscopy, Magnitude 100 kHz......... 30 Figure 3-10 EM Magnitude vs %Fe.............................................................................................. 31 Figure 3-11 EM Phase vs %Fe...................................................................................................... 31 Figure 3-12 Kappameter KT-9...................................................................................................... 32 Figure 3-13 Rock Copper Grade - Magnetic Susceptibility (all draw points) .............................. 33 Figure 3-14 Comparison of Fe and Cu against %Magnetics by Weight on Draw Point F33S..... 34 Figure 3-15 Comparison of Fe and Cu against %Magnetics by Weight on Draw Point E30N .... 34 Figure 4-1 Copper Grade Distribution across Draw Points .......................................................... 37 Figure 4-2 Comparison of Distribution Heterogeneity and Copper Grade Average .................... 39 Figure 4-3 Distribution Heterogeneity - Copper Grade ................................................................ 40 Figure 4-4 Mass Pull - Recovery – Grades ................................................................................... 42 Figure 4-5 Distribution Heterogeneity vs Mass Pull - Recovery - %Cu ...................................... 43 Figure 4-6 Mass Pull - Cu Distribution - %Cu ............................................................................. 45 Figure 4-7 Distribution Heterogeneity by Drill Hole Intervals, 1m ............................................. 47 Figure 4-8 Drill Core Samples (Bern.K & Bamber.A., 2017) ...................................................... 48 Figure 4-9 Mass Pull - Recovery - %Cu ....................................................................................... 49 Figure 4-10 Constitutional Heterogeneity – Average %Cu .......................................................... 50 Figure 4-11 Copper Grade Distribution at Drawpoint D11S ........................................................ 51 Figure 4-12 Copper Grade Distribution at Drawpoint G40N ....................................................... 51 xii Figure 4-13 Constitution Heterogeneity vs Mass Pull - Recovery - %Cu (test data) ................... 52 Figure 4-14 Comparison of Copper Grade of Pulp and Rock by Rock ........................................ 53 Figure 4-15 Comparison of Copper Grade of Pulp and Rock by Draw Point Averages .............. 54 xiii List of Symbols Symbol Description ai Particle or group grade aL Group or lot grade Mn Particle or group mass ML Group or lot mass xiv List of Abbreviations CH – Constitutional Heterogeneity DH – Distribution Heterogeneity HFEMS – High Frequency Electromagnetic Spectroscopy P80 – 80% mass passing size XRF – X-Ray Fluorescence xv Acknowledgements I would like to express my deep thankfulness for my supervisor Professor Bern Klein for enlarging my understanding of the subject and guidance throughout the research period. I would also want to thank the project lead Mr. Stefan Nadolski for his guidance as a tutor and a friend throughout the New Afton Cave-to-Mill project. I also want to say thank you for New Afton mine and National Science and Engineering Council (NSERC) for samples and process data and importantly, research grant for this project. I also want to express my appreciation for Minesense Technologies Ltd. for the laboratory access for conducting my test works. Finally, I want to thank my grandparents, parents and family for their endless love and words of support, and financial help throughout my education journey. 1 Chapter 1: Introduction 1.1 Purpose of the Research New Afton mine is an underground block cave operation based on the historic Afton open pit mine that was in operation from 1978 until 1998. The mine is located 15kms northeast of Kamloops, BC and officially commenced operation in 2012. Mineralization of the orebody is divided into three main categories: Hypogene, Mesogene and Supergene. With hypogene being a primary copper sulphide mineralization, mesogene and supergene being secondary mineralization with differing primary copper minerals than hypogene and concentration of native metals, varying degree of mineral variability will be presented in these mineralization types giving satisfactory level of confidence to assess ore heterogeneity. Defining ore heterogeneity and assessing sortability of the ore material based on heterogeneity parameter has several cost benefits in a mine operation. The most important reason is that a significant reduction in tonnage can be achieved with an application of a sorter to eliminate materials at a certain stage of mining. Removal of low-grade material in advance of a concentrator will result in an increased ore grade and tonnage reduction will reduce operational costs associated with material handling and the primary cost benefit will be achieved in a concentrator operation in relation to energy and water usage as well as a reduction in reagent consumption in a flotation process. It is stated in the New Afton Technical Report that processing cost only is estimated to be US$9.2/t in 2017 and approximately 15,000 tonnes of ore is processed daily. If it is assumed that an installation of a sorting machine achieves a removal of 20% of the daily throughput, processing cost reduction will be directly proportional with the decreased tonnage. In addition, the metal content of the concentrator feed grade will be increased and will result in increased metal recovery for per tonne of feed material. With these potential benefits as an objective of this research study, 2 understanding of ore heterogeneity and its amenability to sensor-based sorting will be studied through the three stages stated below: i. Assessment of distribution and constitution heterogeneity of orebody from historical production data and test works on sample from underground operation ii. Analysis of rock and pulverized sample amenability to various types of sensor-based sorting units iii. Evaluation of a measure of potential removable material ahead of the concentrator operation by pre-concentration 1.2 Research Methodology The study involved three separate parts: i. Part One developed an understanding of ore heterogeneity based on historical sampling and analysis of production data as well as drill hole data. ii. Part Two consisted of laboratory test works to assess sensor-based sorting using a range of sensors including X-Ray Fluorescence, High-Frequency Electromagnetic Spectroscopy and Magnetic Susceptibility, and the evaluation of applicability of these methods on the ore material. iii. Part Three involved a constitution heterogeneity analysis on data sets obtained through test works on rock samples from underground operation. 1.3 New Afton mine 1.3.1 Brief Introduction New Gold Inc. constructed a new mine and mill facility at the New Afton deposit and officially started operation in 2012. As stated in the technical report, the New Afton head grades for copper and gold were 0.91% and 0.78 g/t (New Afton Technical Report, 2015). The estimated probable mineral reserves at the New Afton Mine are 42 million tonnes (Mt). Reserves grading 0.84% Cu, 3 0.56 g/t Au, and 2.3 g/t Ag. All the mineral reserves are in the A and B zones of the deposit. There are additional mineral reserves in the C-zone, which is located approximately 550 m below the B-1/B-2 mining horizon. The C-zone has the potential to extend the mine life by approximately four years, five years including ramp-up. Figure 1-1 Mine Area Plan View (New Afton Technical Report, 2015) New Afton mill processes about 15,000tpd producing concentrates with 24% Cu and 20 g/t Au. SAG and ball mill grinding produce flotation feed that is 80% passing 160um and product of tertiary grinding in a cleaning circuit is 80% passing ~30um. After the mill expansion project where a tertiary grinding unit was installed achieving P80 of ~150um or less ahead of rougher flotation, metal recovery in both stages significantly increased with an increased grinding performance. 1.3.2 Afton Mine Orebody “The principal host rock for the New Afton deposit comprises crystalline and polymictic fragmental volcanics belonging to the Triassic Nicola Formation and lesser monolithic intrusive breccias. These rocks have been altered and mineralized by a monzonite intrusive 4 consisting of a fault controlled elongated stock and related dike swarm. The monzonite is generally weakly mineralized to unmineralized and is interpreted as the causative intrusive phase that is less susceptible to the introduction of sulphide mineralization. Its geometry is best described as a narrow elongated stock that remains open at depth and pinches down plunge to the west” (New Afton Technical Report, 2015). Figure 1-2 shows the main lithological units of the Afton ore body. Figure 1-2 Level Plan of the Geology at the 5150m Elevation, New Afton Mine, Kamloops, BC (New Afton Technical Report, 2015) 1.3.3 Orebody Mineralization “The descriptive terms for the three main categories for mineralization are hypogene, mesogene and supergene are used to describe the mineral deposit. Hypogene is understood to be mineralized primarily by chalcopyrite and accessory bornite which forms the core of the deposit and is dominated by biotite alteration. Hypogene is defined as having greater than 1% sulphides or 0.5% sulphides in bornite-dominant zones. Mesogene is described as a secondary hypogene mineralization upon primary sulphide mineralization by tennantite-tetrahedrite with possible bornite and chalcocite. The secondary hypogene mineralization is associated with bleached 5 sericite-dolomite alteration along narrow fault zones and is responsible for the introduction of deleterious elements such as arsenic, antimony and mercury. The distribution of mesogene mineralization is very narrow and discontinuous and commonly restricted to faults. The supergene mineralization is related with the oxidized portion of the deposit, where sulphides are largely replaced by native copper. Supergene domain is defined to have 0.5% or greater native copper” (Geological Setting and Mineralization, New Afton Technical Report, 2015). Figure 1-3 Cross Section Views of the Lithology, Alteration and Mineralization Models (New Afton Technical Report, 2015) “New Afton is interpreted to be an alkali porphyry copper-gold deposit associated with a monzonite stock which has intruded and mineralized Nicola Formation intermediate to mafic volcanic rocks” (Lipske and Wade, 2014). “The mineralization comprises discontinuous copper sulphide stringer veinlets and disseminations, primarily, but not exclusively, confined to the wall Lithology Alteration Mineralization 6 rocks of the monzonite intrusions. The principal host lithology for the mineralization are crystalline and polymictic fragmental rocks and monomictic intrusive breccias, grouped together as BXF (Figure 1-3). In the eastern half of the deposit, the BXF is intruded by a coeval diorite sill, termed DI. Through the central and western portions of the deposit, the mineralization is bounded by an ultramafic picritic flow (PI). The monzonite (MO) bodies encompass both dike swarms and a tabular stock with a near-vertical dip, open at depth, and narrowing down-plunge towards the west. It is interpreted as a causative intrusive phase for mineralization, but is itself only weakly mineralized” (New Afton Technical Report, 2015). Table 1-1 below shows the distribution of mineralization types in life of mine plan with hypogene mineralization comprising 54% of the total feed. Rock Type %Estimated for Testing % in LOM Plan Hypogene 50 52 Mesogene 40 to 50 44 Supergene 5 to 10 4 Table 1-1 Percentage of Ore Type (New Afton Technical Report, 2015) 1.4 Ore Sorting Within the scope of this study, availability of test work units and timeline allowed for tests on ~450 rock samples. Three different types of sensor-based ore sorting methods were tested. X-Ray Fluorescence and High-Frequency Electromagnetic Spectroscopy tests have been done at a laboratory of Minesense Technologies Limited based in Vancouver, BC and Magnetic Susceptibility test work using hand-held Kappameter KT-9 unit at Coal and Mineral Processing laboratory of the University of British Columbia. 7 Out of the three sensor-based methods, the majority of data analysis has been conducted using results of X-Ray Fluorescence test work and the results from the other two methods were evaluated for potential applicability on the New Afton Mine orebody. Orebody heterogeneity analysis is conducted from production data and x-ray fluorescence test work results to evaluate the possibility of installation of a sensor-based sorting machine to discriminate ore and waste material. In addition, Cave-to-Mill study (Nadolski.S et al., 2015) indicated that heterogeneity parameter can be used as a constraint for pre-concentration technologies to be installed in material handling systems of a mine. More specifically, there are more opportunities for installation of sensor-based pre-concentration in cave mines along draw points, feeders and storage bins. The benefit of operational cost is estimated to be higher as the installation of sensor-based sorting machine can be set earlier in mine production cycle. Ore heterogeneity and sensor-based sorting is a single unit process of Cave-to-Mill concept that is at the meeting point of mine operation and mill operation, and will eliminate a mass of waste material which constitutes a significant part of contribution in overall benefit of the Cave-to-Mill concept. The purpose of application of Cave-to-Mill concept on caving mines is to develop an integrated operation of mine and mill that can simultaneously respond to operational change resulted from ore characteristics alteration. Constitution heterogeneity analysis is based on average copper values acquired through XRF test works and it was also used on the measurement of pulverized samples for comparison with particle surface readings. Similar procedures were followed in the HFEMS and Magnetic Susceptibility test works. 8 Data results from HFEMS and Magnetic Susceptibility test works are analyzed to find a correlation with a metal grade or other element or mineral that could be used as an indicator parameter for pre-concentration. 9 Chapter 2: Literature Review Chapter 2 will introduce ore heterogeneity and its importance to sensor-based sorting. Current status of sensor-based sorting applications in mining will also be reviewed, specifically, in relation to sorter systems involving the sensors such as x-ray fluorescence, electromagnetic conductivity and magnetic susceptibility as these methods have been utilized in this study. The literature review will provide an overall introduction of the concepts of sensor-based sorting systems and its relationship with heterogeneity. Figure 2-1 below introduces general stages for a successful sorter performance and the necessary components on the applicability of sorter systems. Four steps are needed for the evaluation of sortability of a given ore material and it starts with heterogeneity analysis. Once the heterogeneity of a material is assessed, sensor performance can be tested on the material. From the combination of heterogeneity parameter and the response from sensor against the heterogeneity, an analysis of an effectiveness of the sorter can be made. At the last stage, a study of economic benefits of the sorter installation compared against a potential amount of material that can be removed can be assessed for the feasibility of a sorter installation. Figure 2-1 Stages of Sortability Study (Klein. B. & Bamber. A., Course Notes) SortabilityHeterogeneity AnalysisSensor EvaluationSorting AnalysisFeasibility10 2.1 Ore Heterogeneity Gy (1976) defined that heterogeneity contributes to sampling error and therefore heterogeneity must be carefully considered and estimated when sampling takes place. Any fraction of a batch of homogenous material has the same composition as the batch itself thereby the sampling of a homogenous material is an exact selection process, whatever the conditions of sampling. On the other hand, the fractions that can be extracted from a batch of heterogeneous material do not usually have the same composition as the batch itself. The sampling of a heterogeneous material is, therefore, a random selection process, generating sampling errors (Gy. P., 1976). Sampling error, random distribution of a material in other words, is therefore caused by heterogeneity of the material. Heterogeneity is divided into two categories: constitution and distribution. While the sconstitution heterogeneity (CH) deals with a variation of a single fragment of a batch against a group, the distribution heterogeneity (DH) accounts for the spatial distribution, which compares a group against a lot. When decreasing the group size (in principle all the way down to each group consisting of just one fragment), the DH asymptotically becomes equal to the CH, which is also the case if a lot could be transformed into a perfectly mixed and blended sample that shows no grouping and segregation – thus spatial variability – in the whole lot. (Robben. C., 2013, Mazhary et al, 2015). CH will not be affected by blending or segregation but crushing and grinding a fragment, and reduction in particle size will change it as the particle is broken down to finer fragments with varying weights and contents. In contrast, crushing and grinding of a lot will not affect the DH as the weight and content of the group will remain same with only a change in the number of fragments within the group. However, blending and segregation will cause a change in DH as the 11 overall content and weight of the group changes. Blending tends to decrease spatial heterogeneity while segregation tends to increase it. Calculation of CH and DH of a material is done by the Equation 1 below with a difference in weight and grade of the material. 𝐻𝑒𝑡𝑒𝑟𝑜𝑔𝑒𝑛𝑒𝑖𝑡𝑦 =∑(𝑎𝑖 − 𝑎𝐿)2𝑎𝐿2∗𝑀𝑛2𝑀𝐿2 Equation 1 Heterogeneity ai – grade of a fragment (group in distribution heterogeneity) aL – average grade of a group (lot in distribution heterogeneity) Mn – mass of a fragment (group in distribution heterogeneity) ML –total mass of a group (lot in distribution heterogeneity). “Heterogeneity is defined by the variability of an individual fragment or a group compared to a larger group or a lot. In mining systems, heterogeneity is generally expressed at the block modeling stage (eg. 5 m x 5 m scale) and is used to distinguish between ore and waste rock, as well as at the beneficiation stage at size ranges where mineral liberation occurs (e.g. 100 um scale). Sorting is applied at length scales intermediate to these two extremes and it is therefore important to understand how heterogeneity is expressed between these two size ranges” (Figure 2-2) (Klein. B. & Bamber. A., 2017). 12 Figure 2-2 Heterogeneity Scale in Mine and Mill Environment (Klein. B. & Bamber. A., Course Notes) “Much can be learned about heterogeneity and therefore the potential for sorting from knowledge of lithology within the orebody and their mineralogy. Preliminary assessment should include a review of reports that describe features of the orebody that would allow sorting (e.g. multiple adjacent lithologies with different metallurgical response). Descriptions of distinctive differences between ore and waste rock lithology and related mineral associations can provide insights into heterogeneity as well as into physical and chemical characteristics that can be exploited as the basis for sensor discrimination” (Klein. B. & Bamber. A., 2017). The parameter ore heterogeneity then can be compared against parameters such as mass pull and recovery. Relative amenability to sorting can be assessed by examining the yield/grade/recovery characteristics indicated by the curve in Figure 2-3 (Mazhary. A. & Klein. B., 2015). >10m >1000T 1m 10cm <1cm <0.1kg Mine PlanningGrinding / Flotation Today Mining Ignores Mid-Scale Heterogeneity Length Scale 13 Figure 2-3 Grade-recovery curve, CH-10 (Mazhary. A. & Klein. B., 2015) Figure 2-3 is from a paper published in 2015 on assessment of amenability of an ore to sensor-based sorting conducted on samples from porphyry copper-molybdenum deposit. It illustrates that 90% of the copper can be recovered with a mass pull of only ~30% with recovered concentrate copper grade being ~1% Cu. Given the heterogeneity parameter of 10, it can be said that 70% of the feed mass can be removed before process plant. 2.2 Sensor-Based Ore Sorting Rock sorting, in general, has long been a method of separation for mines and the methodology was gradually advanced through time with technological innovation and a constant search for mining companies to increase cost efficiency. The removal of non-valuable rock material ahead of downstream processing results in a significant reduction in energy consumption and the associated costs as well as material handling requirements and costs. Due to these benefits, sensor-based sorting has found wide application in mass mining operations and as high-grade 14 orebody are gradually becoming depleted, there is a greater need to sort low-grade material and to increase metal content of the ore reporting to the concentrator. Sensor-based sorting comprises of four main steps as shown in Figure 2-4 and includes Rock Presentation, Sensing, Electronic Processing and Rock Separation. Figure 2-4 Schematic Representation of the Four Aspects of the Ore-Sorting Process (modified from Salter and Wyatt, 1991) More complex or simpler stages will be found depending on the complexity of feed material characteristics, ore mineralogy and capability and capacity of the sensing technology. For instance, the application of optical sorting is primarily for diamond recovery and other gem stone processing and is based on exploiting the physical properties of a color of the material while application of magnetic sensors will be commonly found in iron ore deposit or orebodies that contain magnetic content as a deleterious mineral. 15 Table 2-1 below shows the application of physical properties of rocks and the corresponding technologies that can be used to sense the property. Property Sensor Technology Color/Refectivity Optical/Colorimetric Sorting Conductivity Conductivity Sorter Fluorescence X-Ray / Laser / UV Sorting Radioactivity Radiometric Sorting / Gamma Neutron Activation Magnetism / Paramagnetism Magnetic Separators Temperature Differential Microwave Heating & Sensing Density Gravity Separation / Dense Media Separation Size Scalping / Screening Table 2-1 Typical Criteria for Discerning Ore from Waste (Bamber. A., Klein. B., Scoble. M., 2006) Sorting systems is divided into two main groups primarily depending on mine operation type, particle sorting and bulk sorting. Particle sorting or rock sorting is preferred over bulk sorting in high-grade and low throughput operation while bulk sorting is an ideal solution for low-grade, high throughput operations where sorter capacity is a principal factor. For particle sorting, the material must be prepared through a series of crushing, and screening stages to prepare narrowly-bounded size fractions, typically 3:1 top size:bottom size range. Often washing via sprays or wet screening is needed to clean the particle surfaces for accurate sensing. The sorting systems require feeders that produce a monolayer of spaced particles so that each particle can be sensed and sorted individually. 16 “While the throughput of current particle sorters at less than a few hundred tonnes per hour generally prohibits their deployment in high throughput mine applications, bulk sorting solutions overcome throughput limitations by incorporating sensor systems into the material handling solution of the mine. While less selective, and delivering generally lower yield results than rock sorting, bulk sorting has very low-cost intensity, and where heterogeneity in the ore is present at a relevant length scale, is very effective at delivering increased value. An increasingly wide range of sensors that integrate into conveying equipment, haulage equipment, and more recently into loading assets e.g. mining shovels is now available to support bulk sorting options where these did not previously exist” (Klein & Bamber, 2017). “By removing coarse barren material, pre-concentration has the potential to significantly reduce the amount of material that requires downstream processing. If conducted close to the mining face, it can potentially reduce ore transport requirements by ejecting barren gangue and transporting less ore to the processing plant. Pre-concentration effectively upgrades the plant feed, fewer tonnes of ore are treated in the processing plant per tonne of product, thus reducing the costs, energy and water consumption per tonne of product” (Duffy et al., 2015). Figure 2-5 Metso Bulk Ore Sorting Concept (Duffy et al, 2015). 17 Bulk ore sorting process consists of 5 stages: ore preparation, ore presentation, material detection, data processing and sorting. Depending on the sensor type, detection capability, throughput capacity and penetration capability, ore material should be washed, classified by size and presented with a certain level of thickness. For example, the detection accuracy of x-ray fluorescence sensor would be better if a material is presented in a thin, single layer. However, an advantage will be given to sensing methods that have the capability to penetrate through the material and the material presentation can be in a bulk method such as shovel-based or like the trucks shown in Figure 2-6. A truck loaded with ore material will pass through the overhead scintillometers and scan response with the grade of a material will be transmitted to the control and a decision can be made if the material is ore, waste or if it can be stockpiled. Figure 2-6 Overhead Scintillometers (L) used for scanning of U308 bearing trucks (R) at Rossing Uranium, Namibia (Nield, 2002) Another option is shown in Figure 2-7 where a shovel equipped with sensors, X-ray detector, will scan the scooped material in real-time and the grade of the material can be informed to the operator. 18 Figure 2-7 MineSense ShovelSenseTM shovel-based bulk batch sorting solution at a BC Cu-Au mine (Bamber et al, 2016) “Sensors installed on belt conveyors with integrated diverters can sense the conveyed material, supporting sorting decisions at a higher resolution than the truck or shovel scales, in the tonnes or possibly hundreds of tonnes in range (Figure 2-8). Probably the best cited example of this type of sorting is the application of Laser-Induced Fluorescence (LIF) in the bulk sorting of high phosphorous Fe from low phosphorous Fe at Kiruna” (Kruukka & Briocher, 2002). Figure 2-8 Bulk diversion of nickel laterites via XRF analysis to product stockpile (right) and waste stockpile (left) (retrieved from www.minesense.com/products/beltsense.html) 19 2.3 X-Ray Fluorescence X-ray Fluorescence is a non-destructive analysis technique capable of measuring the surface chemical composition of particles. Individual particles are irradiated, the resulting emissions detected, analyzed and evaluated based on a selected and calibrated algorithm (Rule CM., RJ. Fouchee RJ. & Swart. WCE., 2015). It provides a list of elements detected as each element emits at a unique wavelength during fluorescence, and an indication of the relative amount of the element is inferred based on the relative intensity of the observed value corresponding to that wavelength (Holler, Skoog, & Crouch, 2007). “When X-ray strikes a mineral, the atoms absorb x-rays and electrons are ejected from the inner shells, creating vacancies. These vacancies create an unstable form of the atom, and the atom returns to its stable condition by transferring electrons from the outer shells to the inner shells, and in the process, it emits x-rays of characteristic energy related to the difference between the binding energies of the corresponding shells. The x-rays emitted from this process are called x-rays fluorescence or XRF. Emitted x-rays are classified by a spectrometer designed to recognize fluorescence of specific elements in the periodic table” (Bern. K. & Bamber. A.S., 2017). X-Ray Fluorescence units commonly used today normally detect emissions for over 30 elements. 2.4 Electromagnetic Conductivity & Permeability Electromagnetic (EM) methods are mostly used in base metals application utilizing its magnetic properties to acquire distinctive measure against other mineral combinations present in a material. Electromagnetic coils are set up in a unit and generate an electromagnetic field and a sample specimen is placed within the set-up and a disruption in the field caused by the specimen is measured. This measurement is recorded in magnitude (kHz) and phase (kHz), and an erratic deviation of these parameters will indicate if the material should be treated as valuable or 20 uneconomic. Processing time for a high-speed electromagnetic sensor is usually within a few millisecond range after a material is presented under an array of sensors. 2.4.1 ConductOreTM “ConductOreTM is a High-Frequency Electromagnetic Spectroscopy sensing device developed by MineSenseTM and it consists of an electromagnetic coil which measures magnetic and electromagnetic field above the sensor and can detect materials (rocks) which are either conductive or magnetic. A conductive or magnetic rock positioned near a sensor will disrupt the magnetic field produced by the sensor. The sensor has a capability to measure this disruption and report it” (Dirks, 2014). 21 Chapter 3: Ore Sorting Test As described in Chapter 2, X-ray fluorescence, electromagnetic conductivity, magnetic susceptibility and magnetic separation tests were run on samples from various draw points at the New Afton mine. Magnetic and electromagnetic tests were conducted to evaluate possible correlations between copper grade and any metal bearing mineral or another mineral that show direct, opposite or a certain trend in relation with metals. X-ray fluorescence was the sensor that was tested most in this study to define a copper content of a rock particle or homogenized pulp sample for an evaluation of heterogeneity of a material. 3.1 Sample Preparation 3.1.1 Numbering and Labeling A total of 22 buckets of rock samples from 11 draw points from various areas of the mine were collected and subjected to test work. Two buckets of samples were collected from each draw point and each bucket contained approximately 40-50 kg of samples consisting of fines and intact rocks. The sampling procedure involved taking one full bucket of sample from mid-section of the muck pile and another full bucket from across bottom-section of draw point pile. A few scoops of fines were added to each bucket. Figure 3-1 Sample Collection from Muck Pile at Drawpoint B13S (taken during site sampling) 22 Figure 3-2 displays the East Cave of the underground operation to show the overall view of the draw point locations from which samples were collected. Figure 3-2 Draw Point Layout of East Cave (courtesy of New Afton mine) Table 3-1 displays the number of samples collected for sensor-based sorting test work. Draw points Weight, kg ( >50mm ) Weight, kg ( <50>25mm ) Weight, kg ( <25mm ) Total, kg B13S 38 6.4 14.01 58.3 G40N 38 3.1 9.14 50 E30N 43 3.6 2.4 49.2 F33S 26 4.7 11.5 41.8 E23N 42 3.6 9.02 54.6 E15N 31 3.2 4.8 39.1 E13S 41 5.6 1.001 47.8 D38N 21 4.4 12.8 38.06 D11S 42 4.5 0.9 47.2 D11N 31 5.9 7.2 44.3 D07S 39 5.4 0.9 45.2 Table 3-1 List of Draw Points for Samples 23 Samples were weighed and sized to create three size classes namely coarse (>50mm), medium size (<50>25mm) and fines (<25mm). The coarse and medium size rocks were washed with water to produce clean surfaces, numbered and the basic physical properties such as weight, density, color, and shape were recorded in a spreadsheet. The fine fraction was also weighted and saved for drum magnetic separation test. After labeling and numbering each sample, rock samples were separated into 6 lithological group types within each draw point by visual analysis and with a guidance of New Afton mine core logging manual. These lithology group types include picrite, diorite, monzonite, fault rock, and carbonates. 3.1.2 Lithology Group Rock samples collected from eleven draw points of New Afton Mine operation was classified by their lithology type as described in the previous section. This section presents the lithology types and descriptions of them drawn from the technical report on New Afton mine. In addition, draw point rock samples classified by the technical report guidance are presented in Figure 3-3 to Figure 3-6 by each lithology type. Table 3-2 below illustrates the texture, mineral percentage within lithology groups as well as alteration characteristics of these rock groups. Picrite (PI) Picrite(PI) Diorite (DI) MO(BXFX) Monzonite(volcanic fragmental breccia) Fault rock Carbonates Texture Outer Propylitic (PO), porphyry, alteration obscured(AO) Outer Propylitic (PO), Alteration obscured (AO) Outer Propylitic (PO) Outer Propylitic (PO) Obscured, round shape Alteration obscured (AO) Minerals Chlorite, k feldspar <10%, biotite Chlorite, 50%>k-spar>20%, biotite k-spar<10% k-spar>50%, biotite, calcite Alteration obscured (AO) Dolomite veins, magnetite 24 Picrite (PI) Picrite(PI) Diorite (DI) MO(BXFX) Monzonite(volcanic fragmental breccia) Fault rock Carbonates Alteration Biotite dominant potassic (KB) or Inner Propylitic (PI) or Calcic-Potassic Biotite dominant potassic (KB) or Inner Propylitic (PI) or calcic-potassic Potassium feldspar dominant potassic (KK) or Biotite dominant potassic (KB), phyllic Potassium feldspar dominant potassic (KK) Fault CA Alteration Minerals Biotite looks black sheen on surface, chlorite, k-spar Chlorite, 50%>k-spar>20%, biotite Calcite, biotite K-spar>50%, biotite, calcite Dolomite or ankerite Table 3-2 Lithology Group Type “Picrite (PO) is dark bluish-green to black and strongly magnetic. It is composed of fine to coarse grained, subhedral to euhedral chlorite and serpentine altered mafic phenocrysts mingled at times with magnetite –altered olivine crystals within a moderately to strongly serpentine +/- chlorite +/- magnetite altered groundmass. The texture ranges from massive to fine grained, medium to coarse grained, and porphyritic to autoclastic. Contacts are commonly sheared with chlorite or dolomite +/- calcite. The rock mass has a high magnetic susceptibility, and its dark color and soft talc feel are diagnostic. The picrite is distinctly less competent than other rock types and this impacts ground conditions within the mine” Figure 3-3 Picrite Rock Sample Figure 3-4 Monzonite Rock Sample 25 “Monzonite (MO) is light to dark orange-pink, locally green and fine to medium grained. Textures vary from porphyritic to fine grained equigranular and trachytic. It is primarily composed of subhedral to euhedral K-feldspar, plagioclase, biotite and hornblende +/- fine-grained disseminated magnetite. It is variably altered to pervasive or patchy K-feldspar and patchy to vein controlled epidote, magnetite +/- actinolite. Jigsaw crackle and fault breccias, containing specularite, magnetite +/- chlorite are common along margins within the main mineralized zone. Diorite (DI) is grey-green, fine to medium grained, with an equigranular ‘salt and pepper’ or seriate texture. Phenocrysts comprise subhedral to euhedral plagioclase +/- biotite +/- pyroxene, with medium to dark green, fine to coarse grained mafic xenoliths. Poikilitic biotite is diagnostic, but challenging to recognize when the diorite is moderately to strongly altered. Magnetite commonly occurs as disseminations or as aggregates, replacing mafic crystals and filling fractures and massive veins with epidote +/- actinolite +/- apatite selvages” Figure 3-5 Diorite Rock Sample Figure 3-6 Fault Rock Sample 26 “Volcanic Fragmental Breccia (BXF) is a group of rock comprised of polymictic fragmental volcanic breccia, monomictic volcanic breccia and crystalline volcanic rocks. These rocks are grouped together because of their complex inter-relationships and intensity of mineralization. Polymictic fragmental volcanic breccia – Comprising poorly sorted, variably colored, massive, porphyritic and trachytic angular to sub-rounded, lapilli to block sized clasts of felsic through ultramafic composition. Clast rock types are commonly porphyritic diorite, andesitic flows, mafic volcanics, picrite and aphyric volcanics within ash to coarse-grained crystal dominated matrix. Monomictic volcanic breccia – Contains well sorted crystal-rich clasts of diorite or volcanic breccia with sub-angular fragments. Commonly found on margins of intrusive bodies, particularly diorite. Intense hydrothermal alteration of fragments and matrix is common. Crystalline volcanic rocks – Crystal tuffs and andesite flows dominated by very fine to fine to medium grained subhedral to anhedral, broken or embayed phenocrysts of plagioclase +/- pyroxene +/- hornblende. Contains less than five per cent by volume of coarse ash to lapilli sized lithic fragments within a variably altered fine grained to ash matrix” (New Afton mine Technical Report, 2015). 3.1.3 Crushing and Pulverizing After the numbering and labeling step, the intact rocks were tested for surficial X-ray fluorescence response. This section will present the process of crushing and grinding procedures that were followed by a collection of surface x-ray fluorescence data. Intact rock samples from each size fraction (>50mm; <50>25mm) were chosen based on copper grade obtained from surficial XRF measurements. Selected rock samples were representative of 27 highest to lowest copper grades and lithology group types for the evaluation of potential inter-relationship between rock-mineral types with a metal grade. Coarse fraction (>50mm) rocks were crushed through three stages of crushers at the Center for Coal and Mineral Processing laboratory at University of British Columbia. Crushing stages included jaw crushing followed by large and small cone crushing. The final product of small cone crushing generated material that was 80% passing 8mm. The middle size fraction (<50>25mm) was crushed through stages of large and small cone crusher and the same P80=8mm product was achieved. With the purpose of saving time due to a large amount of individual rock samples, an additional stage of crushing is avoided and the 80% passing 8mm product was pulverized. A total of 20 rocks from each draw point (10 from each size fraction) has been selected for crushing and pulverization. Coarse size fraction rocks normally yielded 600 – 2000 grams of sample and the middle fraction rocks yielded about 600 grams. All samples were pulverized for 5 minutes with 400g in each pulverization cycle. Feed rock with a P80=8mm was pulverized down to a P80=400um and the product was split to obtain representative amount for XRF measurements. For the coarse size fraction, around 300g was obtained through cone splitter and for the middle size fraction, pulverized sample was entirely taken instead of representative amount. With representative sample amount of ~300 grams from each size fraction rock with P80=400um after 2-3 stages of crushing and 1 stage of pulverizing, samples were pulverized for 5 more minutes to obtain homogenized pulp sample. The optimum weight of 300g was estimated by using the Pierre Gy sample size calculation method. 28 3.2 X-Ray Fluorescence As was mentioned in the Sample Preparation section, rock samples were tested for surficial XRF measurements and pulp measurements after pulverization. Differentiation between size fractions, draw point, average copper grade and lithology group type was maintained. Test work was conducted using a hand-held XRF unit located at Minesense Technologies Ltd. Rock samples were exposed to XRF on 4 surfaces for 20 seconds on each surface and the raw data generated spectrum peaks for 30 elements. The raw data represented readings on 4 surfaces and a combined average of the 4 readings. The average value was used in all calculations. The copper reading was used for estimation of grades of elements such as sulfur, calcium, aluminum plus others are taken into consideration when an interrelation between metal grade and an element or mineral is sought. Comparison between rock surface reading and homogenized pulp sample is made to evaluate the accuracy of surface readings and results yielded a Pearson correlation value of 0.95. After the completion of all sensor-based testing, rock samples were crushed and pulverized as defined in the previous section for homogenized pulp test work. Approximately 300g of pulp was taken as a representative sample from coarse fraction rocks and medium fraction rock was taken as its weight was in the range of ~ 50-300g. Coarse fraction pulp was exposed on 8 spots, two on each side of the bag, for duration of 40 seconds on each spot while medium size fraction is shot on 4 surfaces for the same 40 second scanning periods. 29 3.3 Separation by Magnetic Properties Using the same procedures that were used for XRF test work, rock samples were subjected to characterization of magnetic properties to detect the metal content in relation to the magnetic properties. 3.3.1 Electromagnetic Spectroscopy Many economic minerals of interest are metallic and therefore either conductive, magnetic or paramagnetic. These properties can be measured using electromagnetic field devices, an established principle in mining exploration as well as sorting. In principle, passing a current through a coiled conductor generates an electromagnetic field. Conductive and magnetic minerals present distort the field and alter the coil’s electrical properties proportional to the amount of material present (Figure 3-7) (Bamber & Houlahan, 2010). Figure 3-7 Field generated from low conductive or magnetic content (Klein. B. & Bamber. A., Course Notes) The ConductOre test unit (shown in Figure 3-8) at Minesense Technologies was used for measurement of high frequency electromagnetic spectroscopy of the rock samples. Intact rock 30 sample was placed in the unit with exposures on four sides with a duration of 20 seconds on each side. The magnitude and phase are recorded in 100-1400 kHz range for each rock. Figure 3-8 ConductOre Unit (courtesy of Minesense Technologies Ltd) Figure 3-9 illustrates the relationship between data obtained from ConductOre measurement at 100kHz magnitude against magnetic susceptibility measurement acquired from KT-9 Kappameter which is described in the next section. High-frequency electromagnetic spectroscopy and magnetic susceptibility shows a direct correlation and it can be seen in the next section that the magnetic susceptibility is related to the copper grade. Figure 3-9 Magnetic Susceptibility - Electromagnetic Spectroscopy, Magnitude 100 kHz The HFEMS results were also compared to the copper and iron contents. There was no relationship found between the magnitude or phase with copper content. However, as for iron, a 020040060080010000 20 40 60 80 100 120 140HFEMS, magnitude 100kHzMagnetic SusceptibilityMagnetic Susceptibility - Electromagnetic Spectroscopy, magnitude 100kHz31 noticeable increase in counts was observed in the low kHz region with above 5 per cent iron content in both the +50mm and -50+25mm size fractions. Figure 3-10 EM Magnitude vs %Fe Below magnitude level of 500kHz, an increase in count number is observed in parallel with higher iron grade, which is shown in Figure 3-10 above. Figure 3-11 EM Phase vs %Fe A similar trend (shown in Figure 3-11) was observed for the Phase vs %Fe responses such that the count number rises with increase in iron grade below 300kHz. -1000100200300400500-100 100 300 500 700 900 1100 1300 1500Magnitude, kHz%Fe vs Magnitude (>50mm)-500501001500 200 400 600 800 1000 1200 1400 1600Phase, kHz%Fe vs Magnitude (<50>25mm)32 3.3.2 Magnetic Susceptibility The Kappameter KT-9 was used for measuring the magnetic susceptibility of coarse and medium sized rock fractions. “The KT-9 uses a 10kHz LC oscillator and an inductive coil to measure the susceptibility. The frequency of the oscillator is measured in free space and then measured again when the coil is placed on the material for which the susceptibility is required. This frequency difference is directly proportional to the materials susceptibility” (KT-9 Kappameter, User’s Guide, 1997). The KT-9 shown in Figure 3-12 was configured on pin-mode with rock sample configuration of 12 cm diameter and the unit to be placed on the right-hand side during measurements. Figure 3-12 Kappameter KT-9 Unlike the observations from the electromagnetic measurements, results of magnetic susceptibility measurements showed a correlation with copper grade. In Figure 3-13, higher copper grades are primarily in the range of magnetic susceptibility where the range is below 1 units. 33 Figure 3-13 Rock Copper Grade - Magnetic Susceptibility (all draw points) 3.3.3 Magnetic Pull Rare earth magnetic drum was set-up to measure magnetic pull of the rocks. Weights of the rocks were previously measured and the rock was placed under the rare earth magnetic drum on 4 sides to measure the pull subtracted from the original weight. The average of the magnetic pull from the 4 sides of the rock was used for calculation of per cent magnetic in a rock. Weight of the rock is measured in grams and weight pull by a rare earth magnetic drum is obtained and the per cent of this weight in the rock weight gave a magnetic percentage. A specific relationship for this magnetic per cent with copper grade or any other element or mineral was sought. Figure 3-14 displays a comparison of rock magnetic content against the measured iron and copper grade from sample tests of draw point F33S. Iron grade is generally constant around 5% Fe over 0-40% magnetic content while slightly higher grade of copper can be observed with 01234560.01 0.1 1 10 100 1000%CuMagnetic Susceptibility%Cu Rock - Magnetic Susceptibility - 11 draw points34 increasing magnetic content. Figure 3-14 Comparison of Fe and Cu against %Magnetics by Weight on Draw Point F33S 3.3.4 Rare Earth Magnetic Drum Separator Samples of fine fractions (<25mm) from a few of draw points (E30N, G40N, B13S) were collected and each sample was tested for magnetic content. The average magnetic per cent of these three draw points ranged 7-15%. Figure 3-15 illustrates the relationship between Fe and Cu content againts magnetic content on draw point E30N. It can be observed that there is no sharp change in copper grade as magnetic percentage increases and iron grade stays 3.8% Fe with a few ups and downs over the spectrum of varying magnetic content. Figure 3-15 Comparison of Fe and Cu against %Magnetics by Weight on Draw Point E30N 01234567890 5 10 15 20 25 30 35 40 45%Fe, %Cu%Magnetics by weight%Magnetics vs Fe and Cu ( draw point F33S )Cu averageFe average00.511.522.533.544.50% 5% 10% 15% 20% 25% 30% 35% 40%%Fe, %Cu%Magnetic by weight%Magnetic vs Fe and Cu ( draw point E30N )Fe average Cu average35 Chapter 4: Heterogeneity Analysis The main outcomes of heterogeneity analysis study will be presented in this chapter. An overview will provide general insight into the structural organization of results and an overall explanation of how analysis was carried out is described. Individual sections will focus on data explanation and implications of the analysis results. Production data and borehole data were analyzed to understand the distribution heterogeneity (DH) in terms of different spatial dimensions. In addition, samples from underground draw points were collected and tested for elemental content to provide data for constitution analysis. Eleven draw points were selected from varying areas of the underground operation and 2 LHD buckets of coarse rocks and fines were collected from each draw point. 4.1 Overview Each of the three data sets were analyzed separately and were referred to as: i. Production Data ii. Borehole Data iii. Rock Samples Production Data represents historical assay data from each draw point and included sample date, draw point number and assay values for Cu, Au, Ag, Pd, As and Hg for the period from 2011 until 2015. Date Sampled Heading Sample Numbers Cu Per Au Ppm Ag Ppm Pd Ppm As ppm Hg Ppm Hg Ppb 2012-06-16 B06S 2008685 0.02 0.03 0.5 0.01 30.2 0.31 2012-06-19 B06S 2008711 0.02 0.04 0.9 0.01 30.1 0.21 2012-06-27 B06S 2008806 0.006 0.072 0.6 0.006 15.6 0.15 Table 4-1 Example of Production Data 36 Borehole Data from the drilling program included borehole number, location data, sample interval lengths and assay values. The Rock Samples were collected by collecting rocks from selected draw points and the individual rocks were analyzed by XRF as was described in Chapter 3. A total of two buckets from each draw point was collected with one bucket of sample being taken from across mid-section and the other bucket filled from the bottom of the pile. In addition, a few scoops were taken for fine materials. The heterogeneity analysis was conducted across different dimensions in mine production to evaluate the effects of blending and mixing scenarios. In addition, the heterogeneity parameter was compared against mine parameters to assess the connection between the parameters and to formulate an understanding that would inform the potential for ore sorting. The average copper grade, mass pull and recovery analysis in conjunction with heterogeneity analysis were carried out. 4.2 Results of Heterogeneity Analysis 4.2.1 Production Data Production data was provided from the New Afton mine who compiled sample assay records for the period of 2011-2015 from each draw point. “The draw point samples were collected based upon pre-determined sample frequencies. The samples were assayed for copper, silver, gold, palladium, mercury, arsenic, and other metals. Duplicate samples were collected for one in ten of the samples and Certified Reference Material (CRM) was used as a confirmation of the copper and gold analyses. The samples were collected by production personnel with planning and oversight from the mine planning group. The samples were taken to a local offsite laboratory and results were recorded in the geology database. One sample was collected for every 40 to 60 37 buckets mucked material in the central portion of the cave and one sample was taken for every 12 to 20 buckets mucked at the perimeter of the cave. The production sample frequency was reviewed quarterly and updated locally when ore-waste boundaries were crossed” (New Afton Technical Report, 2015). 4.2.1.1 Grade Distribution Production Data contained assay data for approximately 273 draw points sampled over a period of 2011-2015. Figure 4-1 illustrates the distribution of copper grade across these draw points in 0.5% Cu increments. If the cut-off grade is assumed to be 0.3 %Cu, nearly 23 per cent of the material had a grade that was below cut-off and could be eliminated through sorting before any processing takes place. According to the technical report on New Afton mine in 2015, the cut-off grade was estimated to be 0.4 %Cu and in that case, the amount of removable material would increase to 27 per cent. Average grade of the sorter concentrate would be 1.098 %Cu and 1.155 %Cu, respectively for 0.3 and 0.4 per cent cut-off grades. Figure 4-1 Copper Grade Distribution across Draw Points 0%10%20%30%40%50%0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 2.5 3 3.5 MoreCount in percentageCu Grade, %%Cu Distribution in 0.5 Increments38 Removal of waste material comprising 27 per cent of concentrator feed would result in proportionate reduction energy consumption in the concentrator as well as decrease in material handling cost as well. 4.2.1.2 Distribution Heterogeneity With the amount of assay sample data available, there are many options to evaluate the spatial heterogeneity were available. For the purpose of this study, the distribution heterogeneity was used in horizontal and vertical directions of the mass volume. The spatial heterogeneity along a vertical line referred to an analysis of assay data for the borehole data which will be discussed in the next section. For this section, assay sample data is analyzed across draw points on the same data basis, thus representing a horizontal analysis. The figure 4-2 shows the result of DH calculation in comparison with an average copper grade on a weekly basis in the first quarter of 2014. When the DH value is calculated on a weekly basis, individual copper values sampled on a certain day within that week is taken as a group and an average copper grade of that week is taken as the lot grade. To illustrate the horizontal heterogeneity, draw points from any location in the mine operation that was sampled during the same week represents the grade variability across those draw points with zero input in the vertical axis of the draw point. 39 Figure 4-2 Comparison of Distribution Heterogeneity and Copper Grade Average From a comparison of the DH and average copper grades, the two parameters have an inverse correlation. It can be explained that when variation in copper grade within a lot is high, DH will also be high. The reason is that a high variation in copper grade means a larger difference between the grade of a group and the grade of a lot, and that results in an increase in DH value. To assess the interrelationship between DH and average copper grade without preference of spatial classification, the following plot (Figure 4-3) was generated. For the calculation of this plot, the sample assays for each date was taken as a group grade and the lot grade was represented by average of the draw point. Therefore, the plot shows the comparison of DH and the average copper grade for each of the 273 draw points. 0.00.20.40.60.81.01.21.4WEEK 1WEEK 2WEEK 3WEEK 4WEEK 1WEEK 2WEEK 3WEEK 4WEEK 1WEEK 2WEEK 3WEEK 4JAN,2014 FEB,2014 MAR,2014%CuCu grade averages DH40 Figure 4-3 Distribution Heterogeneity - Copper Grade The DH values higher than 1 are very densely populated in an area where an average copper grade is below ~0.35 %Cu and DH normalizes below 1 for grades higher than 0.5 %Cu. Results from these two plots are a proof that when the average grade of that lot decreases, the spatial heterogeneity increases regardless of the amount of the volume taken for the lot. For instance, in Figure 4-2, lot copper grade was taken as an average of the week whereas, in calculation of Figure 4-3, the lot representing the copper grade was the average from a draw point. Each individual sample is assumed to have an equal mass weight of about 400 tonnes and the total weight is calculated by a multiplication of 400 tonnes by the number of samples grouped together depending on a spatial volume being considered. The following is an example calculation of DH using Formula-1: %Cu for a certain sample date ai=0.3% Average %Cu for the draw point aL=0.45% Mass weight for the sample date Mn=400t 0.000010.001000.1000010.000000.0 0.5 1.0 1.5 2.0 2.5DH%CuDistribution Heterogeneity vs Average %Cu (draw point)41 Total mass weight across the draw point ML=400t * 404=161,600t (404 samples taken for this draw point across 2011-2015). 𝐷𝐻 =∑(𝑎𝑖2 − 𝑎𝐿2)𝑎𝐿2∗𝑀𝑛2𝑀𝐿2 =∑(0.3 − 0.45)20.452∗4002(400 ∗ 404)2 The sum of squared difference between each individual sample data and average grade must be calculated. 4.2.1.3 Mass Pull – Recovery - %Cu relationship After defining the relationship between heterogeneity and copper grade, the analysis focused on process benefits of the application of sensor-based sorting. It was found in the previous section that if variation in the copper grade of the material is high, the volume of waste material present in the lot is increased and the material becomes more amenable to sensor-based sorting. The analysis of the data provides information about the effect of sensor-based sorting on grades reporting to the concentrator. Production Data is used for the estimation of mass pull, recovery and copper grade relationship and an amount of material that can be removed for a grade upgrade of feed material to process plant. The relationship between mass pull, recovery, and copper grade is essentially a result of a concentration stage same as that of the mine or mill mass pull and recovery with the only difference that sensor-based sorter being is used as a pre-concentration unit. Production Data from the 273 draw points was used in this calculation with an addition equal percentage weight for each sample date for the mass weight portion of the formula. The blue lines in the graph represent the cumulative concentrate grade while the gray lines represent the cumulative tailing grade. For instance, in Figure 4-4, draw points E23S and D07N are selected for mass pull – 42 recovery – DH relationship. When 50% of the mass is recovered by a sorter, copper distribution within that mass is 87% and 65%, respectively and the DH values are 3.9 and 0.13. Figure 4-4 Mass Pull - Recovery – Grades This is a clear indication that with high DH value, the material will be amenable to sorting. If 40 per cent of the feed material can be removed before processing or even possibly material handling stage with a metal recovery of more than 85%, a significant amount of energy and material transportation cost can be reduced. The relationship between mass pull and DH is closely analyzed on production data of eleven draw points that are same as samples taken from underground. Figure 4-5 shows a similar result to those presented in Figure 4-4 only with a detailed look at a few draw points. However, the DH was measured for each draw point and it can be seen at the right side of the plot with a noticeably higher value than the rest. An interesting finding is that these high DH values refer to higher 01234567891001020304050607080901000 20 40 60 80 100%CuRecovery, %Mass Pull, %Mass Pull - Recovery - %CuRec./Dist. cum grade cum grade tailsDH=0.13 (D07N)DH=3.9 (E23S)43 recovery graphs on the plot and support the same conclusion made earlier that lower copper grades tend to give higher sorter recovery due to its high variation in the copper content. For example, for draw point E23N the dark blue line on the graph has a DH of 2.58, which gives the highest recovery up until 60% per cent mass pull and joins the other two lines which also have a high DH. Figure 4-5 Distribution Heterogeneity vs Mass Pull - Recovery - %Cu Continuing with draw point E23N, only 20 per cent of the mass pull yields a recovery of 80% of the copper with concentrate grade being 0.6 %Cu and rejected 80% of the material grading 0.08 %Cu. Understandably, the high recovery of metal with minimal mass pull is extremely dependent on the draw point whether it is in the center of caving activity or on the perimeter area with a high portion of low-grade material. However, as previously mentioned, these eleven draw points were sampled from the central mining area of the block cave. 01234567891001020304050607080901000 20 40 60 80 100%CuRecovery, %Mass Pull, %Distribution Heterogeneity - Mass Pull - Recovery - %CuB13S DH=0.78D07S DH=0.18D11N DH=0.15E23N DH=2.58D38N DH=0.89E13S DH=0.33G40N DH=0.33E30N DH=0.23F33S DH=0.23D11S DH=0.17E15N DH=0.16DH=2.58DH=0.1644 4.2.1.4 Cut-Off grade scenario Table 4-2 presents results for several cutoff grade scenarios (from 0.2 to 1% Cu) in which concentrate mass pull, copper distribution and copper grade in the concentrate are calculated. CutOff %Cu Wt% Ore Wt% Waste Cu Dist. %Ore Cu Dist. %Waste Cum. Ore Grade Cum. Tails Grade 1.0 35.9 64.1 56.6 43.4 1.28 0.55 0.9 42.9 57.1 64.7 35.3 1.23 0.50 0.8 50.5 49.5 72.8 27.2 1.17 0.45 0.7 59.7 40.3 81.1 18.9 1.10 0.38 0.6 67.0 33.0 86.9 13.1 1.05 0.32 0.5 74.0 26.0 91.6 8.4 1.00 0.27 0.4 80.2 19.8 95.0 5.0 0.96 0.21 0.3 85.7 14.3 97.3 2.7 0.92 0.16 0.2 90.8 9.2 98.9 1.1 0.88 0.10 Table 4-2 Cut-Off Grade Scenario As stated in New Afton mine 2015 technical report released, the cut-off grade was estimated to be 0.4 %Cu (marked by red) and that gives 80 per cent mass pull with 95% of copper recovery and the head grade to the concentrate of 0.96 %Cu. The results from Table 4-2 are graphically presented in Figure 4-6 which shows the trend that the product grade decrease with increase in copper distribution. 45 Figure 4-6 Mass Pull - Cu Distribution - %Cu 4.2.2 Borehole Data “Drilling on the New Afton deposit has been conducted in a series of programs over a period spanning 2000 to the present. By the end of 2014, a total of 263,714 m of drilling had been completed in 644 holes” (New Afton Technical Report. 2015). Out of the 644 holes, assay data was obtained for 490 of the boreholes with complete survey and collar information as well as drill lengths and intervals. The table 4-3 displays the assay data information with hole-id, drill length, interval and copper value. The interval lengths varied from 1 – 8 metres with most of it being 2 or 3m length. holeid sample from to int cu_ppm cu_per AF01-22 21149 234 240 6 1200 0 AF01-22 21130 329 335 6 1100 0 AF01-22 21131 335 341 6 1300 0 AF01-22 21132 341 347 6 1700 0 AF01-22 21133 347 353 6 1000 0 Table 4-3 Example Drill Hole Assay Data 0.00.81.62.43.24.00204060801001.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2%CuMass Pull -Copper Distribution, %Cut Off gradesMass Pull - Cu Distribution - %Cu Wt% OreWt% WasteCu Dist. %OreCu Dist. %WasteCum. OreGradeCum. TailsGrade46 4.2.2.1 Heterogeneity by Intervals Due to the complexity of dipping angle and drill elevation for the vertical heterogeneity analysis, a few representative drill holes with a dipping angle of 90 degrees were randomly selected. Vertical heterogeneity analysis has been conducted in a significantly simpler way than horizontal heterogeneity analysis for an elementary comparison of DH against average copper grade to observe if an inverse relationship still stands. Intervals, m Drilhole 0.5 1 1.5 2 3 5 8 10 %Cu AF03-78 0.63 0.63 0.60 0.63 0.56 0.51 0.49 0.47 1.38 AF02-60 1.09 1.09 1.03 1.08 0.96 0.90 0.82 0.82 0.83 AF01-24 1.26 1.26 1.26 1.20 1.26 1.02 0.92 0.71 0.46 AF01-25 1.87 1.87 1.77 1.77 1.67 1.56 1.52 1.39 0.46 AF03-80 1.94 1.94 1.74 1.95 1.55 1.22 1.05 0.98 0.32 Table 4-4 Data for Drillhole DH across Short Vertical Intervals Table 4-4 displays the boreholes selected for analysis and the calculated average grade and DH for interval length of 0.5m-10m. The DH and average copper grade were calculated for 8 different interval lengths from 0.5m to 10m to observe the variation in spatial heterogeneity. Interval lengths in Borehole Data mostly started with 2m or 3m lengths and with a few random lengths of 7m or 8m or 13m in rare instances. Due to this variability in interval lengths, the interval is broken down to 0.5m to evaluate the heterogeneity with more accuracy and even distribution of intervals. Five graph lines below in Figure 4-7 represents the DH for five drill holes across varying interval lengths and the decrease in DH was observed when interval length is increased. This relationship can mean that the average copper grade increases when wider intervals of material are 47 considered for spatial heterogeneity since DH and copper grade has an inverse correlation. Therefore, lower interval lengths would mean greater amenability to sensor-based sorting as it has higher DH value. Figure 4-7 Distribution Heterogeneity by Drill Hole Intervals, 1m If a smaller vertical interval is selected considering the tonnage with given width and length of a draw point, heterogeneity of the material will be higher and the material will be more amenable to sorting and therefore would result in grade upgrade in the mill feed. For instance, if a draw point is 20m by 20m and the vertical interval is chosen as 1m with a known copper grade from drill core that pass through the section, heterogeneity of that volume of material can be evaluated with given weights and grades, and the certain degree of confidence can be obtained on the sortability of the material. In addition, DH of drill core AF01-25 stays consistent above 1.5 over the range of intervals in contrast to the other four drill cores and this type of information emphasizes the difference in sortability of independent drill cores. 0.00.51.01.52.02.50 1 2 3 4 5 6 7 8 9 10 11DHDrill hole interval, mDH by drill hole intervalsAF03-78 AF02-60 AF01-24 AF01-25 AF03-8048 Early stage assessment can come from visual observations of rock or drill core. Samples include a distribution of valuable and non-valuable constituents, a qualitative assessment of sortability is possible and can justify further testing. When examining the drill core, features that are relevant to sorting are the size of mineralogical features and contacts between valuable and non-valuable constituents. Examination of drill core, particularly for early stage projects, reveals the potential for ore sorting. Figure 4-8 shows sharp contacts between ore and waste. Figure 4-8 Drill Core Samples (Bern. K. & Bamber. A., Course Notes) “It also shows that sections of ore and waste have significant extents, which demonstrates heterogeneity over relatively large sections along the core (5 to 25 cm). Fragmentation should ,therefore, produce particles that have significant differences in mineralogical and physical properties that are sufficiently heterogeneous to allow sorting” (Bern. K. & Bamber. A.S., 2017). 4.2.2.2 Mass Pull – Recovery The mass pull-recovery-grade curves are plotted in Figure 4-9. These curves were created using the drill core lengths to weight the grade distribution. For example, drill hole interval lengths 49 varied from 3 m to 30 m with corresponding copper grades. A 6-m length was weighted to represent two times the mass as a 3-m interval. This graph could be created using a range of interval lengths to indicate how the relations change with interval length. Interval, m Drill hole 3 6 12 18 30 cu_per AF01-24 1.3 1.1 0.6 0.5 0.4 0.5 AF01-25 2.3 2.6 1.3 1.2 1.1 0.5 AF02-60 0.8 0.7 0.8 0.6 0.4 0.8 AF03-78 0.6 0.5 0.4 0.4 0.3 1.4 AF03-80 1.9 1.4 1.5 0.7 0.5 0.3 Table 4-5 DH across Drillhole Intervals If we look at the two highest DH value at interval 6m from Table 4-5, drill hole AF01-25 is DH=2.6 and drill hole AF03-80 has a value of DH=1.4. Both drill holes fall on the highest recovery line in the graph below. It again proves the point that sorting would have a better recovery when a material heterogeneity level is high. For instance, when the mass pull is at 50%, drill hole AF01-25 has a copper recovery of 95% and drill hole AF03-80 has a recovery of 90% compared against drill hole AF03-78, which would have a recovery of 83% copper with DH of 0.5. Figure 4-9 Mass Pull - Recovery - %Cu 02468101214161801020304050607080901000 10 20 30 40 50 60 70 80 90 100Cu grade, %Recovery, %Mass Pull, %Mass Pull - Recovery - Cu GradeAF01-24 AF01-25 AF02-60 AF03-78 AF03-80 AF01-24 AF01-25 AF02-60 AF03-78 AF03-80DH=1.67DH=0.5650 In addition, as it is known that high DH indicates low average copper grade where drill hole AF01-25 is Cu=0.5% and AF03-80 is Cu=0.3%, it proves that low-grade ores are more amenable to sensor-based sorting. 4.2.2.3 Rock samples The third component of the study is based on a data acquired through XRF test work on rock samples from draw points. Since copper values were measured by XRF for each rock, the assessment is for particle sorting rather than bulk sorting. For the constitution heterogeneity analysis (for particle sorting), approximately 40 rocks from 2 size classes (+50mm; -50+25 mm) were selected within each draw point. Figure 4-10 below shows the constitutional versus average copper grade for the eleven different draw points. The figure shows the same relationship as was found for the horizontal distribution heterogeneity analysis from the production draw points. Figure 4-10 Constitutional Heterogeneity – Average %Cu From comparison of the graphs, the CH and copper grade are inversely related for most of the draw points except for a few draw points – B13S and D11N. 0.00.51.01.52.0051015B13S D11N D11S D38N D7S E13S E15N E23N E30N F33S G40N%CuCHDrawpointConstitution Heterogeneity - average %Cu CH Ave. %Cu51 To have a closer look at the copper grade distribution at some of the draw points, D11S and G40N are selected for copper grade distribution plot. Figure 4-11 Copper Grade Distribution at Drawpoint D11S Figure 4-12 Copper Grade Distribution at Draw point G40N 0%10%20%30%40%50%60%70%80%90%100%024681012Frequency%Cu%Cu Distribution at D11SCH=0.6%Cu=0.790%10%20%30%40%50%60%70%80%90%100%05101520253000.20.40.60.8 11.21.41.61.8 22.22.42.62.8 3Frequency%Cu%Cu Distribution at G40NCH=14.252 It can be observed that 31% of the mass is below 0.4 %Cu cut-off grade for draw point D11S with CH value of 0.6 whereas 92% of the mass is below the cut-off grade for G40N where CH is equal to 14.2. The difference in CH value in these two draw points directly relates with the amount of material that is below the cut-off grade. The boxes on the right-hand side of the Figure 4-13 indicate CH with the highest values and the corresponding recoveries on these three draw points, G40N, E23N and F33S, by the marked solid lines, it also displays the highest values – specially within the mass pull range of 9 - 35%. For example, draw point G40N with CH=14.2, 20% of the mass contains 90% of the copper distribution and the product copper grade is 0.67 %Cu. Comparing these numbers against draw points with lower CH values, for instance D07S with CH=1.6, draw point D07S shows a recovery of 58% with a copper grade of 1.7 %Cu at 20% mass pull. The product copper grade can be different due to the varying average grade of draw points. The primary distinction is that copper distribution in the product is 90% compared to 56% at same mass pull of 20% for G40N and D07S, respectively. Figure 4-13 Constitution Heterogeneity vs Mass Pull - Recovery - %Cu (test data) 01234567891001020304050607080901000 20 40 60 80 100%CuRecovery, %Mass Pull, %Constitution Heterogeneity - Mass Pull - Recovery - %CuG40N CH=14.2E23N CH=8.6E30N CH=7F33S CH=5.7D38N CH=4.9B13S CH=2.8D11N CH=2E13S CH=1.8D07S CH=1.6E15N CH=0.9D11S CH=0.6CH=0.9CH=14.253 4.2.2.4 Results of Rock and Pulp Analysis Figure 4-14 presents a comparison of copper values of rock samples and pulverized samples. Approximately 20 rocks, comprised of 10 rocks from each size fraction, that are representative of lithology group (1-6) and copper grade (max=6.1%) were selectively chosen for this assessment. Comparison of the plotted copper values indicates direct correlation. Each dot presents the copper value of rock and pulp, and a total of ~210 rock is plotted on this graph. Figure 4-14 Comparison of Copper Grade of Pulp and Rock by Rock In the Figure 4-14 above, y-intercept (pulverized rock) yields a value of 0.2019 when x-intercept (rock sample) is equal to 0. This can be interpreted as that the copper readings on rock surface can yield 0%Cu while pulverized sample has more evenly distributed copper content and has a higher chance of yielding a copper reading. To estimate the correlation of copper grades between rock samples and pulps within a draw point, one representative copper grade on each draw point is calculated for both rock samples y = 1.1161x + 0.2019R² = 0.7932012345670 1 2 3 4 5 6Pulverized Sample, %CuRock Sample, %CuPulp vs Rock - Copper (rock by rock) 54 and pulp. Comparison of this result is plotted in the Figure 4-15 below indicating R-squared value of ~0.9. Figure 4-15 Comparison of Copper Grade of Pulp and Rock by Draw Point Averages The comparison presented in Figure 4-15 is based on samples from 11 draw points. These draw points were selected from various areas of the underground operation covering several rock types. Averaged copper percentage value from rock surface readings as well as pulp readings for each of these draw points can be found in Table 4-6 below. Draw point Rock Pulp B13S 0.61 0.92 D7S 1.41 1.41 D11N 0.34 0.51 D11S 0.82 1.16 D38N 0.52 0.59 E13S 0.45 0.73 E15N 1.35 1.82 E30N 0.42 0.44 E23N 0.17 0.39 F33S 0.44 0.79 G40N 1.27 1.81 Pearson Correlation 0.95 Table 4-6 Rock vs Pulp Copper Grade by Draw point y = 1.1304x + 0.1596R² = 0.89820.00.40.81.21.62.00.0 0.4 0.8 1.2 1.6 2.0Pulp Average by Drawpoint, %CuRock Average by Drawpoint, %CuRock vs Pulp ( rock, pulp averages ) 55 As it can be seen from the data table, XRF copper readings on rock surfaces and homogenized pulps were almost identical and in some cases, (for instance, draw point D7S) they were equal. This result illustrates the accuracy of XRF unit on surface readings and the surface readings can be safely used for bulk sorting. 56 Chapter 5: Conclusion 5.1 Conclusion The application of sensor-based ore sorting for the New Afton mine will upgrade ore material grade to the mill feed based on mass pull – recovery analysis results conducted on production assay data as well as XRF test work results on rock samples. Compiled copper grade data from 273 draw points show that approximately 27% of the material is under 0.4 %Cu cut-off grade. Based on information that nearly one-third of the material grades lower than the cut-off grade, heterogeneity value is assessed and a relationship was found that these batch of the low-grade material relates with a high heterogeneity value. Across three different types of data set, inverse relationship between copper grade and heterogeneity value was consistent. Therefore, it is understood that if this parameter is defined early through drill core sample information, an ideal blending scenario can be developed for efficient bulk sorting. Based on the potential amount of material that could be eliminated from process cycle, XRF sorter can have the following cost advantages for New Afton Mine: i. Significant reduction in energy, water and reagent consumption in the concentrator due to decline in tonnage ii. Increased ore grade in the material shipped to the mill iii. Reduction in costs associated with material handling following the decrease in tonnage Based on the heterogeneity analyses across three different types of data sets, it is proven that heterogeneity value inversely correlates with copper grade, and it can be used as an indicator for sortability of the material. As for the suitability of sensor, XRF sorter can be utilized for separation of material since correlation of ~0.8 is observed between surface readings and 57 pulverized samples for copper content. However, there has not been a reliable relationship of copper grade or any other element with electromagnetic or magnetic properties assessed by sorters in this study. In the flow sheet of New Afton mill, a magnetic separator is present on the SAG mill product and applied for removal of magnetic particles generated from pebble crusher operation, and is not applied for sorting of any other mineral or particles. 5.2 Future Work Bulk sorting /spatial heterogeneity/ analysis has been made on production data assay values and resulted in a potential removal of significant amount of material. However, XRF sorter performance on bulk material on a conveyor belt still needs to be evaluated with factors such as conveyor speed belt, the intensity of the sorter and most importantly capacity of the sorter. In addition, the performance of another option of sorter that is in underground, shovel based sorter, also needs to be evaluated. If shovel-based sorter proves to be more efficient than conveyor belt sorter, cost reduction will be significant due to an elimination of unnecessary material handling to the surface. For constitution heterogeneity analysis /particle sorting/, only two size fractions (>50mm; <50>25mm) were considered and sorter feed material size variation will have significant influence on sorter efficiency. 58 Reference List Bamber, A.S., & Houlahan, D.J. (2010). Development, Testing and Applications of an Induction-Balance Sensor for Low Grade Base Metal Ores. Proceedings from Sensorgestützte Sortierung 2010, Aachen, Germany. Bamber, A. How, P., McDevitt, C.a., Munoz-Paniagua, D., Dirks, M. and LeRoss, J., (2016), Development and Testing of Real-Time Shovel-Based Mineral Sensing Systems for the Enhanced Recovery of Mined Material, Proceedings 7th Annual Sensor Based Sorting Conference, Aachen Germany, March 2016 Bamber. A., Klein. B. & Scoble. M. (2006). Integrated Underground Mining and Processing of Massive Sulphide Ores. Bergen. R., Rennie. D. & Scott. K. (2015). Technical Report on the New Afton Project, British Columbia, Canada. Dirks. M. (2014). Sensing and Sorting Ore Using a Relational Influence Diagram. Duffy. K., Valery. W., Jankovic. A. & Holtham. P. (2015). Integrating Bulk Ore Sorting into a Mining Operation to Maximise Profitability. Gy. P. (1976). The Sampling of Particulate Materials – A General Theory. Klein. B. & Bamber.A. (2017). MINE508 Course Notes, Integrated Mining and Processing Systems Sensor-based Sorting. 59 Kruukka A, Briocher, H.F., 2002, At Kiruna Mineral Processing Starts Underground – Bulk Sorting by LIF, CIM Bulletin, Vol 96, No. 1066, December 2002 Lipske, J., and Wade, D. (2014): Geological Model of the New Afton Copper and Gold Deposit, British Columbia, internal report to New Gold Inc., 53 p. 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, Canada, 387-400. Nadolski. S., Klein. B., Elmo. D. & Scoble. M. (2015) Cave-to-Mill: a Mine-to-Mill approach for block cave mines Nield. T., (2002), Beyond the Forked Twig, Geoscience Volume 12, No. 12, December 2002 Robben. C. (2013). Characteristic of Sensor-based Sorting Technology and Implementation in Mining. Rule. C., Fouchee. & Swart. W. (2015). Proof of Mine Ore Upgrading – Proof of Concept Plant for XRF Ore Sorting. Vatcha. M. (1996). Grade Distribution at the Whistle Mine (Sudbury, Ontario) with Applications to Ore Sorting. 60 Appendices Appendix A - New Afton Mine Historical Assay Data A.1 Production Data Date Sampled Draw point Sample Numbers CuPer 11/05/2012 D05N 2006551 12.15 03/03/2012 F08N 2007451 8.26 15/05/2013 D16S 3010009 6.78 16/02/2013 F14S 2016431 6.64 23/05/2013 E12N 3010300 6.48 19/10/2013 D18S 3017939 5.93 18/01/2013 F15N 2015468 5.89 13/01/2014 E33N 3022558 5.746 28/12/2014 G40N 3043314 5.71 25/01/2014 E09S 3023195 5.555 02/11/2014 E21S 3039401 5.32 04/04/2015 D16N 3050427 5.145 05/12/2012 D12S 2013843 5.09 29/06/2013 D16S 3012080 5 10/12/2013 D17S 3020733 4.985 14/11/2015 D18N 3066635 4.985 28/01/2014 D18S 3023349 4.9375 08/09/2013 E13S 3015681 4.844 20/04/2013 F10N 2018623 4.82 21/01/2013 D08N 2015495 4.79 Example production data with copper grades sorted from largest to smallest. Due to the amount and confidentiality reasons, only an example is attached. 61 A.2 Borehole Data 10m interval data as an example. holeid int cu_per wt% cum. Wt% cum. Grade cum.tail Rec./Dist. AF03-78 10 2.87 2.33 2.3 2.87 1.35 4.9 AF03-78 10 2.78 2.33 4.7 2.83 1.29 9.7 AF03-78 10 2.78 2.33 7.0 2.81 1.22 14.5 AF03-78 10 2.73 2.33 9.3 2.79 1.16 19.2 AF03-78 10 2.72 2.33 11.6 2.78 1.09 23.8 AF03-78 10 2.63 2.33 14.0 2.75 1.03 28.4 AF03-78 10 2.56 2.33 16.3 2.72 0.97 32.8 AF03-78 10 2.53 2.33 18.6 2.70 0.91 37.1 AF03-78 10 2.40 2.33 20.9 2.67 0.85 41.2 AF03-78 10 2.39 2.33 23.3 2.64 0.80 45.3 AF03-78 10 2.17 2.33 25.6 2.60 0.74 49.1 AF03-78 10 2.14 2.33 27.9 2.56 0.69 52.7 AF03-78 10 2.10 2.33 30.2 2.52 0.64 56.3 AF03-78 10 2.02 2.33 32.6 2.49 0.59 59.8 AF03-78 10 1.98 2.33 34.9 2.45 0.54 63.2 AF03-78 10 1.93 2.33 37.2 2.42 0.50 66.5 AF03-78 10 1.68 2.33 39.5 2.38 0.45 69.4 AF03-78 10 1.43 2.33 41.9 2.32 0.41 71.9 AF03-78 10 1.39 2.33 44.2 2.27 0.38 74.3 AF03-78 10 1.28 2.33 46.5 2.22 0.35 76.5 AF03-78 10 1.24 2.33 48.8 2.18 0.32 78.6 AF03-78 10 1.23 2.33 51.2 2.13 0.29 80.7 AF03-78 10 1.21 2.33 53.5 2.09 0.26 82.8 AF03-78 10 1.14 2.33 55.8 2.05 0.23 84.7 AF03-78 10 1.08 2.33 58.1 2.02 0.21 86.6 AF03-78 10 1.04 2.33 60.5 1.98 0.18 88.4 AF03-78 10 0.95 2.33 62.8 1.94 0.16 90.0 AF03-78 10 0.73 2.33 65.1 1.90 0.14 91.3 AF03-78 10 0.68 2.33 67.4 1.85 0.12 92.4 AF03-78 10 0.65 2.33 69.8 1.81 0.10 93.6 AF03-78 10 0.62 2.33 72.1 1.78 0.09 94.6 AF03-78 10 0.61 2.33 74.4 1.74 0.07 95.7 AF03-78 10 0.55 2.33 76.7 1.70 0.06 96.6 AF03-78 10 0.54 2.33 79.1 1.67 0.05 97.5 AF03-78 10 0.33 2.33 81.4 1.63 0.03 98.1 AF03-78 10 0.23 2.33 83.7 1.59 0.03 98.5 62 holeid int cu_per wt% cum. Wt% cum. Grade cum.tail Rec./Dist. AF03-78 10 0.21 2.33 86.0 1.55 0.02 98.9 AF03-78 10 0.21 2.33 88.4 1.52 0.02 99.2 AF03-78 10 0.17 2.33 90.7 1.48 0.01 99.5 AF03-78 10 0.08 2.33 93.0 1.45 0.01 99.7 AF03-78 10 0.08 2.33 95.3 1.42 0.00 99.8 AF03-78 10 0.07 2.33 97.7 1.38 0.00 99.9 AF03-78 10 0.05 2.33 100.0 1.35 0.00 100.0 DH calculation for 5 drill holes and related %Cu Interval, m Drilhole 0.5 1 1.5 2 3 5 8 10 %Cu AF03-78 0.63 0.63 0.60 0.63 0.56 0.51 0.49 0.47 1.38 AF02-60 1.09 1.09 1.03 1.08 0.96 0.90 0.82 0.82 0.83 AF01-24 1.26 1.26 1.26 1.20 1.26 1.02 0.92 0.71 0.46 AF01-25 1.87 1.87 1.77 1.77 1.67 1.56 1.52 1.39 0.46 AF03-80 1.94 1.94 1.74 1.95 1.55 1.22 1.05 0.98 0.32 63 Appendix B Rock Test Data B.1 X-Ray Fluorescence Sample ID Weight Cu average wt% cum wt% cum grade cum tails grade rec/dist. B13S(Nov, 2015), B, 36 124 2.98 0.29 0.29 2.98 0.33 2.67 B13S(Nov, 2015), A, 38 128 2.80 0.30 0.60 2.89 0.32 5.26 B13S(Nov, 2015), B, 19 108 2.64 0.26 0.85 2.82 0.31 7.33 B13S(Nov, 2015), A, 47 582 2.48 1.38 2.23 2.61 0.31 17.74 B13S(Nov, 2015), B, 52 52 2.28 0.12 2.35 2.59 0.28 18.60 B13S(Nov, 2015), B, 54 30 1.88 0.07 2.42 2.57 0.27 19.00 B13S(Nov, 2015), B, 8 72 1.57 0.17 2.59 2.50 0.27 19.82 B13S(Nov, 2015), B, 48 30 1.43 0.07 2.66 2.47 0.27 20.13 B13S(Nov, 2015), A, 43 320 1.42 0.76 3.42 2.24 0.27 23.41 B13S(Nov, 2015), A, 29 1252 1.29 2.96 6.39 1.80 0.26 35.05 B13S(Nov, 2015), B, 17 72 1.21 0.17 6.56 1.78 0.23 35.68 B13S(Nov, 2015), B, 14 82 1.19 0.19 6.75 1.77 0.23 36.38 64 B.2 High-Frequency Electromagnetic Spectroscopy Magnitude, kHz Batch Size Rock # Drawpoint Date 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 New Afton (+50 mm) A50 E30N 02/12/2016 44 9 2 1 1 1 -1 -2 -2 -4 -3 -6 -8 -8 New Afton (+50 mm) A42 E30N 02/12/2016 0 -2 -2 -1 -1 -3 -4 -5 -5 -8 -9 -13 -16 -19 New Afton (+50 mm) A49 E30N 02/12/2016 293 109 50 30 22 17 12 8 7 3 2 -2 -3 -2 New Afton (+50 mm) A17 E30N 02/12/2016 -2 -3 -4 -2 -2 -4 -4 -5 -5 -8 -8 -12 -15 -19 New Afton (+50 mm) A30 E30N 02/12/2016 -1 -2 -3 -1 -2 -4 -3 -4 -4 -6 -7 -11 -13 -16 New Afton (+50 mm) A36 E30N 02/12/2016 28 4 -2 -2 -2 -2 -4 -5 -5 -8 -9 -13 -15 -18 New Afton (+50 mm) A55 E30N 02/12/2016 128 36 12 6 4 2 1 -1 -2 -5 -6 -10 -12 -14 New Afton (+50 mm) A62 E30N 02/12/2016 256 90 39 23 17 12 9 5 3 -1 -2 -6 -8 -7 Phase, kHz Batch Frac Size Rock # Drawpoint Date 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 New Afton (+50 mm) A50 E30N 02/12/2016 22 14 9 6 4 4 2 2 2 -1 1 1 1 0 New Afton (+50 mm) A42 E30N 02/12/2016 5 1 0 -1 -2 -1 -3 -1 -2 -3 -1 -2 -2 -3 New Afton (+50 mm) A49 E30N 02/12/2016 53 45 36 28 22 18 13 11 9 6 6 4 4 1 New Afton (+50 mm) A17 E30N 02/12/2016 4 1 0 -1 -2 -1 -2 -2 -2 -4 -2 -2 -2 -2 New Afton (+50 mm) A30 E30N 02/12/2016 4 1 0 0 -2 -1 -2 -1 -2 -4 -1 -2 -2 -3 New Afton (+50 mm) A36 E30N 02/12/2016 20 11 7 5 3 2 0 0 0 -2 1 -2 0 -1 New Afton (+50 mm) A55 E30N 02/12/2016 42 29 21 15 11 9 6 6 3 1 3 1 1 0 New Afton (+50 mm) A62 E30N 02/12/2016 53 43 33 26 20 16 11 10 8 5 6 4 2 1 "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2017-09"@en ; edm:isShownAt "10.14288/1.0355403"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Mining Engineering"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "Attribution-NonCommercial-NoDerivatives 4.0 International"@* ; ns0:rightsURI "http://creativecommons.org/licenses/by-nc-nd/4.0/"@* ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Study of New Afton ore heterogeneity and its amenability to sensor based ore sorting"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/62969"@en .