UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The role of hydrology, geochemistry and microbiology in flow and solute transport through highly heterogeneous,… Blackmore, Sharon 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_september_blackmore_sharon.pdf [ 6.66MB ]
Metadata
JSON: 24-1.0166611.json
JSON-LD: 24-1.0166611-ld.json
RDF/XML (Pretty): 24-1.0166611-rdf.xml
RDF/JSON: 24-1.0166611-rdf.json
Turtle: 24-1.0166611-turtle.txt
N-Triples: 24-1.0166611-rdf-ntriples.txt
Original Record: 24-1.0166611-source.json
Full Text
24-1.0166611-fulltext.txt
Citation
24-1.0166611.ris

Full Text

  THE ROLE OF HYDROLOGY, GEOCHEMISTRY AND MICROBIOLOGY IN FLOW AND SOLUTE TRANSPORT THROUGH HIGHLY HETEROGENEOUS, UNSATURATED WASTE ROCK AT VARIOUS TEST SCALES by Sharon Blackmore  B.Sc.(Hons.), The University of Western Ontario, 2003 M.Sc.(Geology), The University of Western Ontario, 2005  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Geological Sciences) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2015 © Sharon Blackmore, 2015 ii  ABSTRACT Drainage chemistries from unsaturated waste rock are affected by a number of hydrological, geochemical and microbiological processes.  These processes are generally coupled and reliable prediction of solute loads is a difficult task for most mine sites, making it challenging to provide accurate estimates of future water treatment requirements.  An assessment of preferential and matrix flow, geochemical controls and microbially-enhanced weathering is beneficial to provide an improved understanding of the dominant drainage controls.   A multi-scale waste rock study was implemented at the Antamina mine (Peru) to assess drainage controls using 10-m high pile experiments, 1-m field barrels and 0.8-m laboratory columns.  Observed drainage chemistries show a strong seasonal pattern in response to changes in infiltration rates, with increasing concentrations during the dry season and decreasing concentrations during the wet season.  These seasonal fluctuations are less pronounced for finer-grained material, likely due to lower proportions of preferential flow, indicating that hydrological processes provide a key control on observed drainage chemistries.  Microbiological analyses show iron-oxidizing neutrophiles are ubiquitous to all rock types, whereas proportions of acidophiles are strongly influenced by lithology.  Drainage waters are more acidic and contain higher metal loads with increasing proportions of acidophilic microbes.  Breakthrough curves of an externally-applied bromide tracer show that infiltration migrates mostly through matrix materials, with a minor proportion following preferential flow paths.  Long tails indicate a portion of mass enters extremely slow matrix flow paths and/or immobile domains.  Chloride, originating from blasting residues, was used as an internal tracer.  Mean residence times from chloride breakthroughs are longer than bromide values for the same spatially-specific region, suggesting a slow release of chloride from low permeability matrix material.   Flow and solute transport processes were successfully modeled using a dual-porosity mobile-immobile approach in HYDRUS1D.   This research provides an improved understanding of the governing hydrologic and geochemical processes relevant to Antamina waste rock with implications for the large-scale waste rock dumps at site.  Other aspects of this research, such as using blasting residues as an internally-applied tracer and using a iii  dual-domain approach to model flow and solute transport, will be of value to other mines with similar conditions.   iv  PREFACE This dissertation is presented in five body chapters, which includes recently peer-reviewed, journal-published research in Chapter 4.  Chapters 2 and 3 provide a synthesis of the construction, instrumentation of experiments conducted at the research site and complementary, smaller scale experiments performed in the university’s laboratory.  It is anticipated that research presented in Chapter 5 and 6 will be submitted to a peer-reviewed scientific journal in the near future.  Notably, research presented in Chapter 5 is a continuation of a published conference paper, which was wholly written by myself with advice and guidance from the co-authors.   A version of Chapter 4, entitled Comparison of unsaturated flow and solute transport through waste rock at two experimental scales using temporal moments and numerical modeling was co-authored by Leslie Smith, K. Ulrich Mayer, and Roger D. Beckie, underwent peer review and was published in 2014 in Journal of Contaminant Hydrology (Volume 171).  (Blackmore et al., 2014). I was the lead investigator in the work described in Chapter 4 and was responsible for all major areas of concept formation, data collection/analysis and the majority of the manuscript composition. Research presented in Chapter 5, is built upon a paper entitled The Influence of Spatial Heterogeneity and Material Characteristics on Fluid Flow Patterns through Two Unsaturated, Mixed Waste-rock Piles was co-authored by B. Speidel, A. Critchley, Roger D. Beckie, K. Ulrich Mayer, and Leslie Smith.  This paper underwent peer review and was published in 2012 in the Proceedings of the 9th International Conference on Acid Rock Drainage. (Blackmore et al., 2012) and can be found in Appendix D.   This research is part of a collaboration of students and researchers from the University of British Columbia, the University of Alberta, Teck Metals Limited’s Applied Research and Technology Group and Compania Minera Antamina S.A.  I was largely responsible for construction and instrumentation of the two waste rock piles included in this thesis work (Pile 4 and 5) and the accompanying field barrels located at the research site.  I was solely responsible for processing and analyzing the data from these experiments unless otherwise noted: soil water characteristic curves generated by Speidel (2011) and Critchley (2011); pyrosequencing results and figures generated by Melanie Scofield.  Myself, Holly Peterson and Michael Gupton, along with advice from my research advisors (i.e., Dr.’s Roger Beckie, K. v  Ulrich Mayer and Leslie Smith), designed and implemented a tracer study with the assistance of numerous persons associated with the mine site staff.  This work is described in Chapters 2, 4 and 5.  My research advisors provided feedback and guidance on the analytical approach and interpretation of data obtained from the research site and associated laboratory experiments, which are described in this dissertation.  I was wholly responsible for design, construction and instrumentation of the two laboratory-based column experiments used in this thesis work.  I was the sole person responsible for processing and analyzing the data from these experiments, with guidance on scientific interpretation from my research advisors.  This work is described in Chapters 3, 4 and 6 and my research advisors and colleagues provided feedback on methodologies I developed for this dissertation, as well as scientific interpretation.    vi  TABLE OF CONTENTS  ABSTRACT ................................................................................................................................................... ii PREFACE .................................................................................................................................................... iv TABLE OF CONTENTS ............................................................................................................................... vi LIST OF TABLES ........................................................................................................................................ xiii LIST OF FIGURES ...................................................................................................................................... xvi ACKNOWLEDGEMENTS ........................................................................................................................... xxi CHAPTER 1: INTRODUCTION .................................................................................................................... 1 1.1. Problem Description ...................................................................................................................... 1 1.2. Thesis Research Site .................................................................................................................... 4 1.2.1. Site Characteristics ............................................................................................................... 4 1.2.2. Experimental Waste Rock Pile Program ............................................................................... 4 1.3. Literature Review of Complementary Waste Rock Studies .......................................................... 5 1.3.1. Cluff Lake Mine ..................................................................................................................... 5 1.3.2. Diavik Mine ............................................................................................................................ 8 1.3.3. Aitik Mine ............................................................................................................................. 11 1.4. Recommendations for the Antamina Study ................................................................................ 13 1.4.1. Preferential Flow versus Matrix Flow .................................................................................. 13 1.4.2. Particle Size and Scale-dependence .................................................................................. 15 1.4.3. Microbial Activity.................................................................................................................. 17 1.5. Thesis Scope and Organization .................................................................................................. 18 1.6. Tables .......................................................................................................................................... 20 vii  1.7. Figures ........................................................................................................................................ 21 CHAPTER 2: METHODOLOGY AND CONCEPTUAL MODEL DEVELOPMENT – EXPERIMENTAL WASTE ROCK PILES ................................................................................................................................. 25 2.1. Introduction .................................................................................................................................. 25 2.2. Antamina Waste Rock Classification .......................................................................................... 26 2.2.1. Lithology .............................................................................................................................. 26 2.2.2. Sulphides and Metal Content .............................................................................................. 27 2.2.3. Classification Criteria .......................................................................................................... 28 2.3. Experimental Waste Rock Piles .................................................................................................. 28 2.3.1. Construction and Instrumentation ....................................................................................... 29 2.3.2. Waste Rock Physical Characteristics .................................................................................. 30 2.3.3. Waste Rock Geochemical Characteristics .......................................................................... 34 2.3.4. Water Balance Recording and Calibration .......................................................................... 36 2.3.5. Aqueous Sampling .............................................................................................................. 37 2.3.6. Tracer Test .......................................................................................................................... 38 2.4. Observations ............................................................................................................................... 40 2.4.1. Precipitation ......................................................................................................................... 40 2.4.2. Flow ..................................................................................................................................... 40 2.4.3. Evaporation ......................................................................................................................... 41 2.4.4. Tracer Results ..................................................................................................................... 42 2.4.5. Aqueous Chemistry ............................................................................................................. 45 2.5. Conceptual Models for the Large-scale ...................................................................................... 48 2.5.1. Pile 4 ................................................................................................................................... 48 viii  2.5.2. Pile 5 ................................................................................................................................... 50 2.6. Summary ..................................................................................................................................... 51 2.7. Tables .......................................................................................................................................... 52 2.8. Figures ........................................................................................................................................ 63 CHAPTER 3: METHODOLOGY AND CONCEPTUAL MODEL DEVELOPMENT – FIELD BARRELS AND COLUMN EXPERIMENTS .......................................................................................................................... 81 3.1. Introduction .................................................................................................................................. 81 3.2. Small-scale Waste Rock Experiments ........................................................................................ 82 3.2.1. Construction and Instrumentation ....................................................................................... 82 3.2.2. Waste Rock Physical Characteristics .................................................................................. 86 3.2.3. Waste Rock Geochemical Characterization ....................................................................... 89 3.2.4. Waste Rock Microbiology .................................................................................................... 91 3.2.5. Water Balance and Calibration ........................................................................................... 92 3.2.6. Aqueous Sampling .............................................................................................................. 93 3.2.7. Tracer Test .......................................................................................................................... 94 3.3. Observations ............................................................................................................................... 94 3.3.1. Flow and Evaporation ......................................................................................................... 94 3.3.2. Tracer Results ..................................................................................................................... 95 3.3.3. Aqueous Chemistry ............................................................................................................. 96 3.4. Conceptual Models for the Small-scale .................................................................................... 102 3.4.1. Field Barrels ...................................................................................................................... 102 3.4.2. Laboratory Columns .......................................................................................................... 104 3.5. Summary ................................................................................................................................... 105 ix  3.6. Tables ........................................................................................................................................ 108 3.7. Figures ...................................................................................................................................... 117 CHAPTER 4: COMPARISON OF UNSATURATED FLOW AND SOLUTE TRANSPORT THROUGH WASTE ROCK AT TWO EXPERIMENTAL SCALES USING TEMPORAL MOMENTS AND NUMERICAL MODELING ............................................................................................................................................... 133 4.1. Introduction ................................................................................................................................ 133 4.2. Materials and Methods .............................................................................................................. 136 4.2.1. Site Description ................................................................................................................. 136 4.2.2. Experimental Piles............................................................................................................. 136 4.2.3. Laboratory Columns .......................................................................................................... 138 4.2.4. Parameter Estimation ........................................................................................................ 140 4.2.5. Numerical Simulations ...................................................................................................... 143 4.3. Results ...................................................................................................................................... 149 4.3.1. Measured Flow and Bromide Breakthrough ..................................................................... 149 4.3.2. Temporal Moment Analysis ............................................................................................... 151 4.3.3. Numerical Modeling........................................................................................................... 152 4.4. Discussion ................................................................................................................................. 155 4.4.1. Parameter-estimation Uncertainties and Future Prediction .............................................. 158 4.5. Conclusions ............................................................................................................................... 159 4.6. Tables ........................................................................................................................................ 163 4.7. Figures ...................................................................................................................................... 167 CHAPTER 5: FIELD INVESTIGATION OF PREFERENTIAL AND MATRIX FLOW IN UNSATURATED MINE WASTE ROCK USING MULTIPLE TRACERS .............................................................................. 176 5.1. Introduction ................................................................................................................................ 176 x  5.2. Site Description ......................................................................................................................... 178 5.3. Methodology .............................................................................................................................. 178 5.3.1. Experimental Pile Construction ......................................................................................... 178 5.3.2. Complementary Field Barrel Construction ........................................................................ 180 5.3.3. Material Characteristics ..................................................................................................... 180 5.3.4. Aqueous Sampling ............................................................................................................ 181 5.3.5. Tracer Tests ...................................................................................................................... 181 5.3.6. Parameterization of Flow Properties ................................................................................. 183 5.3.7. Numerical Modeling........................................................................................................... 184 5.4. Results and Discussion ............................................................................................................. 186 5.4.1. Physical Characteristics .................................................................................................... 187 5.4.2. Specific Discharge............................................................................................................. 187 5.4.3. Sulphate Concentrations ................................................................................................... 188 5.4.4. Tracer Breakthrough ......................................................................................................... 189 5.4.5. Temporal Moment Analysis ............................................................................................... 192 5.4.6. Numerical Modeling........................................................................................................... 196 5.5. Conclusions ............................................................................................................................... 200 5.6. Tables ........................................................................................................................................ 203 5.7. Figures ...................................................................................................................................... 208 CHAPTER 6: MICROBIOLOGICAL AND GEOCHEMICAL CONTROLS ON SULPHIDE OXIDATION AND METAL RELEASE FROM UNSATURATED WASTE ROCK ................................................................... 217 6.1. Introduction ................................................................................................................................ 217 6.2. Materials and Methods .............................................................................................................. 219 xi  6.2.1. Origin of Waste Rock ........................................................................................................ 219 6.2.2. Column Design .................................................................................................................. 220 6.2.3. Precipitation and Evaporation ........................................................................................... 221 6.2.4. Mineralogical Characterization .......................................................................................... 221 6.2.5. Microbiological Characterization ....................................................................................... 222 6.2.6. Aqueous Chemistry and Mass Loadings .......................................................................... 223 6.3. Results and Discussion ............................................................................................................. 224 6.3.1. Flow and Water Balance ................................................................................................... 224 6.3.2. Mineralogical Composition ................................................................................................ 225 6.3.3. Microbiological Analyses ................................................................................................... 225 6.3.4. Drainage Chemistry and Controls ..................................................................................... 228 6.3.5. Efficiency of Acid Neutralization ........................................................................................ 232 6.3.6. Effect of Heat Treatment on Drainage Water Chemistry .................................................. 233 6.3.7. Microbial Abundance and Diversity as a Function of Rock Type ...................................... 233 6.4. Conclusions ............................................................................................................................... 234 6.5. Tables ........................................................................................................................................ 237 6.6. Figures ...................................................................................................................................... 241 CHAPTER 7: CONCLUSIONS ................................................................................................................. 250 7.1. Thesis Objectives ...................................................................................................................... 251 7.1.1. Quantification of Water Balance Components of Waste Rock Piles at Antamina ............ 251 7.1.2. Hydrology and Degree of Preferential/Matrix Flow Associated with Waste Rock ............ 252 7.1.3. Flow and Solute Transport Modeling: Uniform Flow Model versus Dual-Domain Model . 253 7.1.4. Blasting Residues as a Resident Tracer ........................................................................... 253 xii  7.1.5. Is it Possible to Replicate the Influence of Large-scale Structures in the Laboratory? ..... 254 7.1.6. Microbial Community Structures in Waste Rock ............................................................... 255 7.1.7. Quantification of Microbially-enhanced Weathering in Waste Rock ................................. 255 7.2. Significance to Other Mine Sites ............................................................................................... 256 7.3. Recommendations for Future Work .......................................................................................... 257 REFERENCES .......................................................................................................................................... 259 APPENDIX A ............................................................................................................................................. 277 APPENDIX B ............................................................................................................................................. 297 APPENDIX C ............................................................................................................................................ 313 APPENDIX D ............................................................................................................................................ 333 APPENDIX E ............................................................................................................................................. 334 APPENDIX F ............................................................................................................................................. 376  xiii  LIST OF TABLES Table 1.1 Comparison of Three Studies using Mesoscale and/or Multi-scale Waste Rock Experiments. . 20 Table 2.1 Acid-base Accounting and Bulk Geochemistry of Major Antamina Waste Rock Types. ............ 52 Table 2.2 Antamina Waste Rock Classification. ......................................................................................... 53 Table 2.3 Waste Rock Classification and Composition of Pile 4. ............................................................... 54 Table 2.4 Waste Rock Classification and Composition of Pile 5. ............................................................... 55 Table 2.5 Bulk Geochemical Characteristics of Pile 4 Waste Rock Material. ............................................. 56 Table 2.6 Bulk Geochemical Characteristics of Pile 5 Waste Rock Material. ............................................. 57 Table 2.7 X-ray Fluorescence (XRF) Whole Rock Geochemistry of Antamina Waste Rock. .................... 58 Table 2.8 Mineralogical Composition of Pile Waste Rock using X-ray Diffraction (XRD) Analysis. ........... 59 Table 2.9 Pile 4 and 5 Calibration Equations and Average Tipping Bucket Volume. ................................. 60 Table 2.10 Pile 4 and 5 Tracer Test Characteristics. .................................................................................. 61 Table 2.11 Lysimeter Contribution (in m3) to Annual Drainage in Pile 4 and 5. ......................................... 62 Table 3.1 Summary of Field Barrel Experiment Characteristics. .............................................................. 108 Table 3.2 Waste Rock used in Laboratory-based Column Experiments. ................................................. 109 Table 3.3 Physical Characteristics of Waste Rock used in Column Experiment. ..................................... 110 Table 3.4 Bulk Geochemical Characteristics of Pile 2 Waste Rock used in Column Experiments. ......... 111 Table 3.5 X-ray Fluorescence (XRF) Whole Rock Geochemistry of Antamina Waste Rock used in Field Barrel and Column Experiments. .............................................................................................................. 112 Table 3.6 Mineralogical Compositions (by XRD) of Waste Rock used in Field Barrel and Column Experiments. ............................................................................................................................................. 113 Table 3.7 Microbial Initial Population Enumeration via the Most Probable Number Technique. .............. 114 Table 3.8 Calculated Daily Precipitation Rate and Duration for Laboratory-based Columns. .................. 115 Table 3.9 Field Barrel versus Experimental Pile Mass Loading Rates (mg/kg/week). ............................. 116 Table 4.1 Antamina Waste Rock Classification (modified from Antamina, 2007). ................................... 163 Table 4.2 Initial Model Simulation Input Parameters. ............................................................................... 164 Table 4.3 Calculation of Effective Flow and Transport Parameters. ........................................................ 165 Table 4.4 Final Flow and Solute Transport Parameters used in Model Simulations. ............................... 166 xiv  Table 5.1 Antamina Waste Rock Classification Scheme (revised from Antamina, 2007). ....................... 203 Table 5.2 Summary of Material Characteristics of Waste Rock Classes (± 1 S.D.). ................................ 204 Table 5.3 Summary of Measured Chloride Concentrations from Field Barrels, Pile 4 and 5. .................. 205 Table 5.4 Temporal Moment Analysis of Field Barrels and Experimental Piles using Bromide and Chloride Tracer Solutes. .......................................................................................................................................... 206 Table 5.5 Hydraulic Properties and Flow Transport Parameters of Pile 4 and 5 Waste Rock Materials. 207 Table 6.1 Antamina Waste Rock Classification (revised from Antamina, 2007). ..................................... 237 Table 6.2 Physical and Hydraulic Parameters of Laboratory Column Waste Rock.................................. 238 Table 6.3 Averaged Mineralogical Composition (weight %) of Waste Rock used in Column Experiments, by X-ray Diffraction with Rietveld Analyses. ............................................................................................. 239 Table 6.4 Microbial Enumerations (as bacteria∙g-1) after Heat Treatment. ............................................... 240 Table A.1 UTM Co-ordinates of Pile 4 Instrumentation Lines. .................................................................. 277 Table A.2 UTM Co-ordinates of Pile 5 Instrumentation Lines. .................................................................. 278 Table A.3 Particle Size Distribution (PSD) Curves Waste Rock Used in Pile 4 and 5. ............................ 279 Table A.4 Single Ring Infiltrometer Measurements of Pile 1 – Class B Waste Rock. .............................. 280 Table A.5 Single Ring Infiltrometer Measurements of Pile 2 – Class A (Intrusive) Waste Rock. ............. 280 Table A.6 Single Ring Infiltrometer Measurements of Pile 3 – Class A (Skarn) Waste Rock. ................. 281 Table A.7 Soil Water Characteristic Curve Results (as Gravimetric Water Content m3m-3). .................... 282 Table A.8 X-ray Fluorescence (XRF) Results. .......................................................................................... 284 Table A.9 Compilation of Raw Data from Tipping Bucket Files. ............................................................... 285 Table A.10 Volumes Recorded from Pile 4 Contributing to CUC Calculation. ......................................... 290 Table A.11 Volumes Recorded from Pile 5 Contributing to CUC Calculation. ......................................... 293 Table B.1 Growth Media Compositions of Three Microbial Species for MPN Enumeration. .................... 297 Table B.2 Comparison of Initial Microbial Mopulations via the Most Probable Number technique from Column 1 and 2. ........................................................................................................................................ 298 Table B.3 Bulk Geochemical Characterization of Waste Rock Excavated from Antamina’s Open Pit and Used in Laboratory Column Experiments. ................................................................................................ 306 Table C.1 Input Values for HYDRUS1D Simulations of Pile 5. ................................................................. 315 xv  Table C.2 Input Values for HYDRUS1D Simulations of Column 1. .......................................................... 322 Table C.3 Input Values for HYDRUS1D Simulations of Column 2. .......................................................... 326 Table C.4 Estimated van Genuchten Soil Parameters (from Carsel and Parish (1988)) related to USDA Soil Types.................................................................................................................................................. 331 Table E.1 Background Bromide Concentrations from Antamina Pile Outflow. ......................................... 334 Table E.2 Meterological Conditions for Hargreaves Evaporation Estimation. .......................................... 335 Table E.3 Pile 4 Upper Boundary Conditions. .......................................................................................... 345 Table E.4 Pile 5 Upper Boundary Conditions. .......................................................................................... 355 Table E.5 Pile 4 Graphical Summary. ....................................................................................................... 365 Table E.6 Pile 5 Graphical Summary. ....................................................................................................... 367 Table F.1 Bulk Geochemical Characterization of Waste Rock Excavated from Antamina’s Open Pit and Used in Laboratory Column Experiments. ................................................................................................ 376 Table F.2 Sequencing Statistics. .............................................................................................................. 378 Table F.3 X-ray Diffraction Analyses from Individual Samples used in Column Waste Rock. ................. 382 xvi  LIST OF FIGURES Figure 1-1. Location of mining projects in Canada. .................................................................................... 21 Figure 1-2. Location of the Antamina Mine. ................................................................................................ 22 Figure 1-3. Map location of three waste rock pile studies; Cluff Lake uranium mine and Diavik diamond mine (A) and Aitik copper mine (B). ............................................................................................................ 23 Figure 1-4. Comparison of four conceptual models for water flow and solute transport. ........................... 24 Figure 2-1. Schematic of experimental waste rock piles, plan view (A) and side view (B). ........................ 63 Figure 2-2. Trenching of Pile 4 instrumentation line 4. ............................................................................... 64 Figure 2-3. Instrumentation line placement (A), sensor installation (B) and covering (C). ......................... 65 Figure 2-4. Particle size distribution of waste rock used in Pile 4 and 5. .................................................... 66 Figure 2-5. Soil water characteristic curves (SWCCs) from Pile 4 and 5 waste rock. ................................ 67 Figure 2-6. Calibration curve from Pile 4 lysimeter C. ................................................................................ 68 Figure 2-7. Comparison of weekly precipitation from Punto B and Yanacancha rain gauges between Sept 14, 2009 and August 15th, 2010. ................................................................................................................. 69 Figure 2-8. Comparison of uranine adsorption in Class A and B waste rock (A and B, respectively). ....... 70 Figure 2-9. Daily precipitation (mm) recorded from research site rain gauge. ........................................... 71 Figure 2-10. Pile 4 (A) and Pile 5 (B) area-normalized daily outflow recorded from pile lysimeters. ......... 72 Figure 2-11. Pile 4 tracer breakthrough curves of bromide (A and B) and uranine (C and D). .................. 73 Figure 2-12. Pile 5 tracer breakthrough curves of bromide (A and B) and uranine (C). ............................. 74 Figure 2-13. Comparison of drainage chemistries from Pile 4 lysimeter outflows. ..................................... 75 Figure 2-14. Comparison of drainage chemistries from Pile 5 lysimeter outflows. ..................................... 76 Figure 2-15. Comparison of As, Fe, Mo, Sb, Se, and Zn concentrations from Pile 4 outflows. ................. 77 Figure 2-16. Comparison of Cd, Cu, Co, Pb, and Ni concentrations from Pile 4 outflows. ........................ 78 Figure 2-17. Comparison As, Fe, Mo, Sb, Se, and Zn concentrations from Pile 5 outflows. ..................... 79 Figure 2-18. Comparison of Cd, Cu, Co, Pb, and Ni concentrations from Pile 5 outflows. ........................ 80 Figure 3-1. Schematic of field barrel construction (A) and photographs of field barrels (B and C). ......... 117 Figure 3-2. Column experiment construction. ........................................................................................... 118 Figure 3-3. Instrumentation of column experiments. ................................................................................ 119 xvii  Figure 3-4. Particle size distribution of waste rock used in field barrels and column experiments. .......... 120 Figure 3-5. Soil water characteristic curves (SWCCs) of Pile 4 and 5 field barrel materials and waste rock used in column experiments. .................................................................................................................... 121 Figure 3-6. Cumulative area-normalized outflow (m3∙m-2) from field barrels. ........................................... 122 Figure 3-7. Cumulative precipitation (dashed lines) and area-normalized outflow (solid lines) from column experiments. .............................................................................................................................................. 123 Figure 3-8. Comparison of tracer breakthrough curves (A) and cumulative mass recovery (B) from column experiments. .............................................................................................................................................. 124 Figure 3-9. Comparison of pH and sulphate concentrations from field barrels. ....................................... 125 Figure 3-10. Comparison of alkalinity and nitrate concentrations from field barrels. ................................ 126 Figure 3-11. Comparison of dissolved As and Zn concentrations from field barrels. ............................... 127 Figure 3-12. Comparison of dissolved Mo and Sb concentrations from field barrels. .............................. 128 Figure 3-13. Comparison of dissolved Cu and Co concentrations from field barrels. .............................. 129 Figure 3-14. pH, sulphate and dissolved concentrations (Mn, Si, Ca, and Na) from column experiments. .................................................................................................................................................................. 130 Figure 3-15. Dissolved metal(loid) concentrations (As, Cd, Mo, Sb, Se, and Zn) from column experiments. .................................................................................................................................................................. 131 Figure 3-16. Dissolved metal(loid) concentrations (Co, Cr, Cu and Ni) from column experiments. ......... 132 Figure 4-1. Large-scale constructed pile (Pile 5) (A) and small-scale laboratory column (B), including complementary photographs of experiments. ........................................................................................... 167 Figure 4-2. Average particle size distribution curves for Class A (A) and C (B) waste rock material used in CPE and column experiments................................................................................................................... 168 Figure 4-3. Cumulative area-normalized outflow (A) and bromide tracer concentration for Pile 5 (B) and laboratory columns (C) versus total experimental time. ............................................................................ 169 Figure 4-4. Cumulative tracer mass recoveries from Pile 5 and column experiments versus normalized outflow. ...................................................................................................................................................... 170 Figure 4-5. Bromide breakthrough curves for Pile 5 (A) and column (B) experiments, in terms of normalized mass flux rate as a function of flow-corrected time (𝝉; Equation 4-1). ................................... 171 xviii  Figure 4-6. Fitted ADE model results with measured normalized mass flux rate. .................................... 172 Figure 4-7. Measured (solid lines) versus simulated (dashed lines) flow using a mobile-immobile model. .................................................................................................................................................................. 173 Figure 4-8. Measured (symbols) versus simulated (lines) solute transport using a dual- MIM model (dashed lines) and a best–fit uniform (dotted lines) modeling approach. ................................................. 174 Figure 4-9. Multi-year simulations (dashed lines) of Column 1 bromide breakthrough. ........................... 175 Figure 5-1. Side-view schematic of Antamina Pile 4 and 5 (A) and field barrels (B). ............................... 208 Figure 5-2. Averaged particle size distribution curves (A) and soil water characteristic curves (B) of waste rock associated with Pile 4 and 5. ............................................................................................................. 209 Figure 5-3. Cumulative discharge from field barrels (A), Pile 4 (B), and Pile 5 (C). ................................. 210 Figure 5-4. Comparison of measured sulphate concentrations from Pile 4 and 5 (basal lysimeter D) to measured daily rainfall amounts. .............................................................................................................. 211 Figure 5-5. Bromide breakthrough curves from Antamina Pile 4 (A) and 5(B) over two years. ............... 212 Figure 5-6. Cumulative chloride recovery from field barrels (A), Pile 4 (B) and Pile 5 (C). Dashed and dotted lines of same colour in A reflect field barrels of Class A (red), Class B (green) and Class C (blue) materials. ................................................................................................................................................... 213 Figure 5-7. Waste rock particle size distribution curves, truncated to field barrel screen size and renormalized to 100% for the largest particle diameter (i.e., d < 10 cm). ................................................. 214 Figure 5-8. Comparison of tracer recovery from Pile 4 (A) and Pile 5 (B) versus cumulative discharge. 215 Figure 5-9. Measured and modeled bromide (A, B) and chloride (C, D) breakthrough curves for Pile 4 – Lysimeter A (blue; LHS) and Pile 5 – Lysimeter B (green; RHS). ............................................................ 216 Figure 6-1. Experimental column set up (A) and conceptual model (B). .................................................. 241 Figure 6-2. Cumulative precipitation (dashed lines) and drainage (solid lines) from Column 1 and 2. .... 242 Figure 6-3. Venn diagram from column pyrotag data. .............................................................................. 243 Figure 6-4. Summary of pyrotag data from columns, as Bray-Curtis hierarchial cluster dendrogram with complete linkage method (top) and indicator species analysis (ISA) bubble plots(bottom). .................... 244 Figure 6-5. Measured pH (A) and cumulative sulphate (B) from Column 1 and 2. ................................... 246 Figure 6-6. Daily mass loading of major cations versus elapsed experimental time for Column 1 and 2. 247 xix  Figure 6-7. Saturation indices (SIs) for select secondary precipitates and primary mineral phases versus elapsed experimental time for Column 1 (dashed lines) and Column 2 (solid lines). ............................... 248 Figure 6-8. Cumulative metal(loid) loads versus time from Column 1 and 2. ........................................... 249 Figure A-1. Comparison of area-normalized outflow from Pile 4 lysimeters versus precipitation (A) and development of correction factors to use as a rainfall proxy during data-gap months (B). ...................... 287 Figure B-1. Schematic of instruments found in Class C material in Column 1 and Column 2. ................ 301 Figure B-2. Schematic of instruments found in Class A-1 material in Column 1 and Column 2. ............. 302 Figure B-3. Schematic of instruments found in Class A-2 material in Column 1 and Column 2. ............. 303 Figure B-4. Cross-view schematic material segregation in Column 1 and Column 2............................... 304 Figure B-5 Measured moisture contents from Column 1 (A) and Column 2 (B). ...................................... 307 Figure B-6. Comparison of dissolved Al and Ca concentrations from field barrels. ................................. 308 Figure B-7. Comparison of dissolved Fe and Si concentrations from field barrels. .................................. 309 Figure B-8. Comparison of dissolved Mn and Ni concentrations from field barrels. ................................. 310 Figure B-9. Comparison of dissolved Pb and Se concentrations from field barrels. ................................ 311 Figure B-10. Comparison of dissolved Al, Fe and K concentrations from Column 1 and 2. .................... 312 Figure C-1. Daily precipitation (mm) amounts at the experimental pile field site (Antamina, Peru). ........ 314 Figure C-2. Soil types (adapted from USDA soil texture triangle; Brown, 2003) relevant to finer-grained (d < 4.75 mm) waste rock fraction. ................................................................................................................ 330 Figure C-3. Standard score statistical results from Pile 5 breakthrough between day-250 to day-650. .. 332 Figure E-1. Gypsum saturation indices (SI) of Pile 4 and 5 outflow concentration. .................................. 369 Figure E-2. Measured chloride concentrations from field barrels associated with Pile 4 and 5. .............. 370 Figure E-3. Measured chloride concentrations from Pile 4 and 5 lysimeter outflow. ................................ 371 Figure E-4. Comparison of in situ water contents at the height of the dry and wet season. .................... 372 Figure E-5. Comparison of modeled versus measured outflow from HYDRUS1D simulations. .............. 373 Figure E-6. Comparison of immobile (A) and mobile (B) bromide concentrations through Pile 4 1D model profile with time. ........................................................................................................................................ 374 Figure E-7. Comparison of immobile (A) and mobile (B) bromide concentrations through Pile 5 1D model profile with time. ........................................................................................................................................ 374 xx  Figure E-8. Comparison of immobile (A) and mobile (B) chloride concentrations through Pile 4 1D model profile with time. ........................................................................................................................................ 375 Figure E-9. Comparison of immobile (A) and mobile (B) chloride concentrations through Pile 5 1D model profile with time. ........................................................................................................................................ 375 Figure F-1. Measured moisture contents from Column 1 (A) and Column 2 (B). ..................................... 381 Figure F-2. Chao1 rarefraction curves from column pyrotag data. ........................................................... 383 Figure F-3. Photograph of the blue secondary mineral precipitated on Class C waste rock surfaces. .... 384    xxi  ACKNOWLEDGEMENTS I would like to first thank Dr. Roger Beckie, Dr. Uli Mayer and Dr. Leslie Smith, who gave me the opportunity to work on this project, learn from so many great people, travel to various places and fine-tune my now barely passable Spanish.   Your support, encouragement and mentorship throughout my Ph.D have been invaluable and I hope to do the same for others in the future.   Thanks are also extended to my external examiner, Dr. Roger Herbert, and my examining committee of Dr. Sue Baldwin, Dr. Erik Eberhardt, and Dr. Barbara Lence for their insightful comments and assistance. This project would not have been possible with funding from the Natural Science and Engineering Research Council of Canada, Teck Metals Limited’s Applied Research and Technology Group and Compañia Minera Antamaina S.A.  Thank you for funding a unique project that has advanced the literature on waste rock hydrology, geochemistry and microbiology.   I was told two memorable phrases about a Ph.D. The first was from my Master’s advisor, Dr. Gordon Southam, when I was contemplating a Ph.D.  He said, “A Ph.D is the marathon of academia.”  After doing a marathon myself, I can say this is very true.  Similar to a marathon, you draw on the support of many people along the way.   My time as a Ph.D student would not have been nearly as successful without the help of the UBC hydrogroup members (in no particular order); Holly Peterson, Dawn Paszkowksi, Kristina Small, Trevor Hirsche, Pablo Urrutia, John Dockrey, Katie Jones, Danny Bay, Maria Lorca, Mehrnoush Javadi, Laura Laurenzi, Natasha Sihota, Nathan Fretz, Jessica Doyle, Andrea Chong, Celedonio Aranda, Charlene Haudpt, Tom Gleeson, Cassanda Koenig, Daniele Pedretti, Kun Jia and Andrew Krentz.  As well, those people who helped remind me that there is a world outside of graduate school and provided many laughs along the way; Morgan Tanner, Nick Ochoski, Wren Bruce, Lyndsay Moore, Jenn Fohring, the East Coast-Vancouver crew, Kristina Brown, Shukling Ng, many people at BGC Engineering and many more friends and family. As well, I received an immense amount of help while conducting the field work side of this thesis.  Specifically, Trevor Hirsche, Pablo Urrutia , Holly Peterson and Michael Gupton were instrumental in pile xxii  construction and conducting the tracer experiment.  Also, the construction and instrumentation of the pile experiment would not have been possible without the help of the following (again, in no particular order): Bartolo Vargas, Fabiola Sifuentes, Bevin Harrison, Edsael Sanchez, Raul Jamanca, Roberto Manrique, Antonio Mendoza, Janeth Visconde, Humberto Valdiva and Dr. Ward Wilson.   And finally, my cheerleaders throughout my Ph.D, particularly in these last few months…which leads me to the second memorable phrase told to me by Dr. Uli Mayer (prior to starting the final Ph.D stretch)…”your life may be terrible for a few months, but then it will all be done”.  Truer words were never spoken.   My sister, Eileen, has stood by me through my incredibly long university career and never wavered in her 100% support throughout all of the highs and the lows.  My parents, Michael and Maureen, who have been steadfast in their belief in me and can now say they have a ‘doctor’ in the family.   Last, but not least in any way, I thank my fiancé Matthew Dawson who should receive the highest degree imaginable for all of his love, support and encouragement.   CHAPTER 1  1  CHAPTER 1: INTRODUCTION 1.1. Problem Description Mines are not native to the environments they exist in and often displace or alter natural structures (e.g., lakes) during their construction and life-of-mine periods.  In the mining process, large amounts of crushed rock are typically excavated in the process of ore extraction and may be stockpiled on the surface.  Stockpiles of waste rock are unsaturated porous structures that can reach 100’s of meters in height.  The exposure of these piles to climatic conditions can accelerate the weathering of sulphide minerals found within the waste rock material.  Specifically, the reaction between oxygen, water and sulphides (or sulphide oxidation; Equation 1-1) can occur throughout the waste rock pile, assuming these components are available, and result in the release of acidity (as H+) and soluble metals.    2𝐹𝑒𝑆2 + 7𝑂2 + 2𝐻2𝑂 → 2𝐹𝑒2+ + 4𝐻+ + 4𝑆𝑂42− (Eq. 1-1) Drainage from these waste rock piles can be acidic and is termed acid rock drainage (ARD).  ARD forms as a result of sulphide oxidation (e.g., pyrite (FeS2), pyrrhotite (Fe1-xS)) reactions outpacing acid-consuming reactions (e.g., carbonate or aluminosilicate dissolution; Equation 1-2 and 1-3, respectively).   𝐶𝑎𝐶𝑂3 + 𝐻+  → 𝐶𝑎2+ + 𝐻𝐶𝑂3− (Eq. 1-2) 𝑁𝑎𝐴𝑙𝑆𝑖3𝑂8 + 4𝐻+ + 4𝐻2𝑂 → 𝑁𝑎+ + 𝐴𝑙3+ + 3𝐻4𝑆𝑖𝑂4(𝑎𝑞) (Eq. 1-3) There exists a large focus in the literature on ARD conditions, relative to research on the generation of pH-neutral drainage or neutral rock drainage (NRD) from mine waste.  The production of NRD may occur in scenarios where the proportion of acid-consuming minerals exceeds that of acid-generating (or sulphide) minerals.  In these scenarios, drainage will likely present circumneutral pH values and may be enriched in metal(loid)s such as antimony, arsenic, cadmium, chromium, lead, molybdenum, selenium, and zinc (Price, 2003).  These metal(loid)s are mobile at neutral pH conditions by forming stable oxyanions (e.g., MoO42-, AsO43-, SeO42-) that do not readily adsorb to solid phases, remain as cations in solution (e.g., Zn2+, Cd2+) or complex with ligands such as chloride (e.g., PbCl+) (Blowes et al., 1995, 1998; Heikkinen and Räisänen, 2008; Nicholson and Rinker, 2000).   CHAPTER 1  2  In Canada, there are over 600 metal, non-metal, coal and oil-sands producing mines (Figure 1-1).  These mines produce large volumes of waste that may create ARD or NRD conditions and therein influence its surrounding environments.   Although the discussion regarding NRD is less common than ARD, the need to enhance the knowledge and information on NRD is coincident with a growing awareness in the mining community that neutral drainage cannot be considered benign. In addition to the larger focus on ARD relative to NRD research, mine tailings have received greater attention relative to waste rock (e.g., Blowes et al., 1991; Dubrosky et al., 1985; Gunsinger et al., 2006; Johnson et al., 2000; Lindsay et al., 2009; Morin et al., 1988; Moncur et al., 2005; Power et al., 2010).  Tailings are defined as gangue (or left over) material following ore processing and may contain residual sulphide contents.   Tailings are commonly fine-grained material with diameters spanning 3 to 4 orders of magnitude (e.g., 10-2 – 10-6 m; USEPA, 1994).  The characteristics of finer grain sizes, with or without sulphide contents, can augment mineral dissolution reaction rates and accelerate the production of ARD or NRD conditions.  Conversely, waste rock is produced from pit blasting, which creates material with a significantly wider range of particle sizes in comparison to tailings.  Waste rock grain diameters can range from clay particles to large boulders (i.e., 10-6 m to greater than 1 m), and the volume of waste rock is generally greater than that of tailings at a mine site.    These two characteristics, a greater spatial and physical heterogeneity of stockpiled waste rock relative to tailings, increases the complexity in replicating field-scale processes at more manageable scales in the laboratory or with smaller experimental sizes.  Many studies have shown laboratory- and field- based weathering rates differ considerably, with the latter presenting weathering rates several orders of magnitude lower than the former (Malmström et al., 2000; Swoboda-Colberg and Drever, 1993; Velbel, 1993).  Interactions between physical and chemical processes in flow systems can be nonlinear and/or coupled and observed responses can be non-intuitive.  In a similar manner, these observations suggest up-scaling, or the process of relating results from smaller scale experiments to larger scale structures, may also exhibit a nonlinear or linear dependence on certain variables and compound the complexity in using laboratory-based experiments to understand field-scale controls on flow and solute transport.    CHAPTER 1  3  In an effort to advance the knowledge base in these three areas (i.e., unsaturated waste rock hydrology, NRD mine wastes and scale-up), this thesis centers on a mine site in Peru that excavates waste rock and stockpiles the material in one of two waste dumps.  The waste rock produced at this mine is generally carbonate-rich and sulphide-poor and contains metals or metalloids with enhanced mobility at neutral pH values (i.e., As, Mo, Zn) and is therefore capable of producing NRD conditions.  Experimental evidence centers around the construction and instrumentation of several 10 m high waste rock piles, as well as smallerscale experiments (i.e., h ≤ 1 m) commonly used by environmental consultants in the field and laboratory.   The application of experiments at multiple scales is relatively common; however mesoscale experiments or those conducted at a scale between mine site structures (e.g., 100’s m in height) and laboratory-based or small-scale kinetic experiments (i.e., ≤1 m) is relatively unique and has been applied at a limited number of sites.  For example, an early application of the mesoscale approach was performed at the Cluff Lake Mine (Saskatchewan, Canada).   The data/results and lessons learned from the Cluff Lake multi-year experiment helped in the development of two other mesoscale projects; the Diavik Waste Rock Project (Northwest Territories, Canada) and the Antamina Waste Rock Project (Antamina, Peru), the latter of which is the research site of this thesis.    Mesoscale projects remain uncommon, which is likely due to the effort required to construct, instrument and conduct these multi-year studies.  More commonly, mine sites study on-site waste rock piles to understand the site-specific geochemical processes governing observed and future drainage chemistries.  The Aitik Mine in Gällivare, Sweden, is an example of a well-researched mine site that has extensively studied an on-site waste rock pile and conducted several smaller scale experiments to quantify the long-term dynamics of copper transport (e.g., Eriksson and Destouni, 1997; Eriksson et al., 1997).    An overview of the thesis research site and the three previously mentioned studies (i.e., Cluff Lake, Diavik and Aitik) is provided in the next sections.  This literature review of research sites that applied a similar mesoscale and/or multi-scale methodology helped to constrain specific questions/objectives to be addressed in this thesis in an effort to continually improve/refine the field of unsaturated waste rock hydrology, NRD and scale-up.   CHAPTER 1  4  1.2. Thesis Research Site The research site is the Antamina mine (Antamina, Peru, South America), which is located in the Ancash Department of north-central Peru at 9o32 S and 77o03 W, approximately 270 km north of Lima, Peru (Figure 1-2).  The mine site is located within the central Andes and the eastern flank of the Cordillera Blanca, at an elevation between 4,100 and 4,700 m above sea level (a.s.l.).   1.2.1. Site Characteristics The Antamina mine site is subject to a distinct ‘wet’ and ‘dry’ season spanning October to April and May to September, respectively.  Mean annual precipitation ranges between 1200 - 1300 mm, with a maximum average daily precipitation amount of 32 mm.  Approximately 80% of the total annual precipitation falls in the wet season as rain.  Mean annual temperatures at Antamina are 5.5oC – 6.0oC.  Weather at Antamina is highly variable and governed by fluctuations within the tropical rain belt. Antamina is a skarn-hosted deposit that formed ~ 9.86 – 10.2 Ma and is described as a small monzogranite porphyry intrusion hosted by Upper Cretaceous carbonate strata within the Maranon Thrust and Fold Belt (Love et al., 2004).  The majority of the deposit geology consists of endo- and exo- skarn, with subordinate breccia bodies’ cross-cutting the skarn and intrusive assemblages (Love et al., 2004).  Five main zones describe the Antamina deposit; 1) endoskarn inner shell; 2) stockwork; 3) breccia; 4) brown garnet exoskarn with Cu-/Mo- ore; and 5) green garnet exoskarn outer shell with Cu-Zn ore (Love et al., 2004).  Ore-hosted sulphides include pyrite, pyrrhotite, chalcopyrite, molybdenite, sphalerite, and galena.  The mine began production of Cu-Ag and Zn concentrates and Pb, Mo and Bi by-products in mid-2001 (Zuzunaga, 2003) and is expected to remain in production until approximately 2024 (Brown et al., 2006). 1.2.2. Experimental Waste Rock Pile Program Five mesoscale waste rock piles, with a footprint of 36 m (l) x 36 m (w) and 10 m height, were built at Antamina between 2006 and 2009.  Each mesoscale pile contains between 19,000 and 25,000 tonnes of waste rock material and were constructed through multiple tips (or tipping phases at 37o) of end-dumped material.  The focus of this thesis is on Pile 4 and 5, which contains mixed waste rock material of two (of CHAPTER 1  5  three) waste rock types excavated at Antamina.  Further details of waste rock pile construction and material types are provided in Chapter 2, whereas  details of the remaining 3 waste rock piles (i.e., Pile 1, 2 and 3) can be found in previous theses and papers (e.g., Bay, 2009; Bay et al., 2009; Corazao Gallegos, 2007; Peterson et al., 2012; Peterson, 2014).   1.3. Literature Review of Complementary Waste Rock Studies Three studies involving mesoscale waste rock piles and/or multi-scale waste rock experiments are discussed in this section.  The first two, the Cluff Lake Mine (Saskatchewan, Canada) and the Diavik Diamond Mine (Northwest Territories, Canada), involve the construction of mesoscale experimental waste rock piles.  The third site, the Aitik mine (near Gällavare, Sweden), investigates leachate from a non-instrumented, mine-produced waste rock pile and several smaller scale experiments.  The following sections review some of the characteristics unique to each site, major research findings and their contribution to available literature on unsaturated waste rock hydrology. 1.3.1. Cluff Lake Mine 1.3.1.1. Site characteristics The Cluff Lake Mine was a uranium mine located in northern Saskatchewan (Canada) or approximately 700 km from Saskatoon, Saskatchewan (58o22′ N, 109o32′ W) (Figure 1-3A).  The mine was open from 1980 to 2002 and was owned and operated by Areva Resources Canada (formerly Cogema Resources).  The mine site is located in a semi-arid climate with air temperature ranging between -40oC to +35oC, with a mean annual temperature of 0oC.  The average annual precipitation at Cluff Lake is 455 mm, of which the majority (305 mm) occurs as rainfall (Nichol, 2002).     1.3.1.2. Pile construction and composition The Cluff Lake waste rock pile had a footprint of 8 m (w) x 8 m (l) and was 5 m in height.  A 5 m height represented the average height of each waste rock lift at Cluff Lake.  Drainage was collected from 16 contiguous lysimeters of PVC geomembrane, each covering a 2 m (w) x 2 m (l) area.  Waste rock used in the experimental pile was mined in 1996 and was comprised of aluminous gneiss and granitoid from the CHAPTER 1  6  Precambrian Earl River and Peter River Gneiss Formations, with an average sulphide content of less than 0.64% (Nichol, 2002).  Application of the grain size classification system by Dawson and Morgenstern (1995) shows this material falls on the boundary of a ‘soil-like’ and ‘rock-like’ porous structure, with 20% of its material passing the 2 mm sieve size. The waste rock pile was instrumented with sensors to monitor water content, matric suction, temperature, water chemistry, pore gas pressure and chemistry.  Tipping buckets monitored outflow rates and volumes and aqueous samples were collected at regular intervals for drainage chemistry (Nichol et al., 2005).  The waste rock pile was fully operational in the fall of 1998; however drainage did not report to the lysimeters until the following year.  This delay was associated with the time required for the waste rock material to return to residual saturation following drying during construction.  As well, the length of the equilibration period was assumed to be dependent on the initial water content of the waste rock and the site-specific climatic conditions (Nichol et al., 2005).   A tracer was applied to the pile one year following construction and breakthrough curves were measured for 2.5 years following application.  Chloride (as LiCl; 2100 mg·L-1 Cl) was selected as the conservative tracer solute and was applied to the surface of the pile as a single rainfall event of 53 mm over 3 hours.    1.3.1.3. Major results Infiltration to the Cluff Lake waste rock pile was estimated to be 50% to 85% of rainfall during high intensity events, with an average of 55% over the long term (Nichol, 2002).  However, net infiltration varied from 30% to 121% of precipitation between the 16 individual lysimeters (i.e., standard deviation = 23%; Nichol, 2002).   These results indicated flow and transport properties varied substantially at spatial scales less than 2 m and individual, 2 m x 2 m lysimeters were poor predictors of net infiltration (Nichol, 2002).   Results of the tracer experiment were used to indicate if unsaturated flow at Cluff Lake is uniform and can be reasonably characterized by the Richard’s equation, which is shown in one-dimension in Equation 1-4 (Richards, 1931).   CHAPTER 1  7  𝜕𝜕𝑡=  𝜕𝜕𝑧[𝐾(𝜃) (𝜕𝜑𝜕𝑧+  1)] (Eq. 1-4) Where; K is the hydraulic conductivity [L∙T-1] and is a function of the volumetric water content (𝜃); [-]); z is the spatial coordinate [L], t is the time [T], and 𝜑 is the pressure head [L]. Uniform water flow typically infiltrates matrix flow paths, where water is held by capillary tension and gravity in fine-grained materials (Nichol et al., 2005).  In contrast, unsaturated flow that is not defined by Equation 1-4 is generally described as preferential flow.   Preferential flow can reflect a continuum of flow paths and flow rates, but is characterized as water flow that is concentrated to an area that is spatially smaller than its area of infiltration.  For example, preferential flow paths can be described as well-connected textures containing larger pores (i.e., macropores) that produce channelized flow and maintains some interaction with the solid phase and waters in the surrounding matrix (Nichol et al., 2005).  Non-capillary flow constitutes an end member of preferential flow and differs qualitatively from other preferential flow paths with respect to the limited interaction between infiltrating fluids and the matrix (Webb et al., 2008).  Instead, non-capillary flow may be conceptualized as flow over clast-supported particles with small amounts of water stored at grain contacts or on the upper surface of larger particles (Elboushi and Davis, 1969; Pruess, 1999). The proportion of matrix and preferential flow is fundamental to the prediction of contaminant transport in unsaturated porous media.    Preferential flow was evidenced from tracer breakthrough curves at 5 m depths within hours of tracer application (Nichol et al., 2005).  Approximately 5% of net infiltration from larger rainfall events was captured at the base of the pile via macropore flow paths.  Non-capillary flow was also noted from tracer results following large rain events, and was estimated to contribute to less than 0.1% of the total outflow volume (Nichol et al., 2005).   The short duration of preferential flow pathways, shown by rapid breakthroughs, was not captured by soil water solution samplers since they sampled a biased fraction of the water present in the matrix (Nichol et al., 2005).  Data derived from basal lysimeters was found to be the most useful and indicative of flow regimes over the entire pile length and represent flux-averaged water outflow chemistry.  CHAPTER 1  8  1.3.2. Diavik Mine 1.3.2.1. Site characteristics The Diavik Diamond Mine is located approximately 300 km north of Yellowknife, Northwest Territories (64o29ʹ N, 110o18 ʹ W; Figure 1-3A) at an elevation of 440 m.a.s.l. and on a 20 km2 island in the Lac de Gras lake (Bailey et al., 2013).  The mine site area can be described as an arctic desert in a region of continuous permafrost.  According to Environment Canada (2010), the region received an average annual precipitation of 280 mm from 1998 to 2007, with ~ 65% of it falling as snow.  The mean annual temperature of the Diavik mine is -10oC (Golder, 2008), with a wide average temperature range of -31oC (January/February) to + 18oC (July) (Environment Canada, 2008).  Diavik is an open pit and underground diamond mine, with a diamondiferous kimberlite ore body hosted in Archean-aged granite and granite pegmatite assemblages.  Waste rock from Diavik is segregated based on its sulphur content.  The highest sulphur contents occur in the biotite schist waste rock (i.e., <0.42 weight % S) and is classified as potentially acid-generating (PAG) material.   In contrast, the granitic host rock contains trace sulphides and a low carbonate mineral content (Bailey et al., 2013; Jambor, 1997; Smith et al., 2013b) and, although the neutralization potential is small, host rock assemblages are considered to be non-acid generating (NAG) material. Unsaturated flow mechanisms in cold environments are typically governed by; temperature, humidity, rainfall, total spring snow-pack, number of days with above freezing ground temperature, and presence/type of permafrost (Indian and Northern Affairs, 1993).  Furthermore, low temperatures in arctic regions results in slower sulphide oxidation kinetics that reduce ARD or NRD production.  However, thermal controls may not be sufficient to completely mitigate potential ARD/NRD effects (Holubec, 2004)    such that the role of free-thaw conditions is an important consideration in the design of waste rock piles and covers.    1.3.2.2. Pile construction and composition Three 15 m high waste rock piles were constructed at Diavik in 2005 - 2006.  Two of the piles are uncovered and have a footprint of 50 m (w) x 60 m (l).  These uncovered piles are composed of NAG CHAPTER 1  9  material and potentially acid-generating (PAG) material.  The third pile contains PAG material and is covered with 1.5 m glacial till and 3 m NAG waste rock layers.  The covered pile has a larger footprint relative to the uncovered piles (i.e., 80 m (w) x 125 m (l)) and the geometry and composition was based on Diavik’s reclamation concept for PAG waste rock.   Both covered and uncovered Diavik experimental piles are underlain by an impermeable HDPE liner (i.e., basal lysimeter) to convey drainage to collection huts for sampling and measurement of physical and chemical parameters.  In addition to the large basal lysimeter, a total of 12 sub-lysimeters are constructed at the base of the pile.  Six sub-lysimeters, with dimensions of 4 m (w) x 4 m (h), and four smaller sub-lysimeters, with dimensions 2 m (w) x 2 m (h), are located near the base of the pile and beneath the crown of the pile.  Two other 2 m (w) x 2 m (h) sub-lysimeters were installed beneath one of the pile batters.  All piles are instrumented with sensors to record pore water chemistry, gas concentrations, moisture content and temperature.  Further details of these additional sensors can be found in Amos et al., (2015), Bailey et al., (2013), Chi et al., (2013), Fretz et al. (2011), Momeyer (2014), Neuner et al., (2013), and Smith et al., (2013a, b).  A distinct active zone layer (or AZL) is often used to describe the upper few meters of waste rock piles that are seasonally influenced by large shifts in air temperatures.  Several 2 m experiments were constructed at Diavik to understand the hydrology and geochemistry specific to this freeze-thaw zone.  Four 2 m (diameter) x 2 m (height) HDPE tanks were installed on a graded bench, with the upper surface representative of the topography of the piles.  As well, these 2 m experiments represent smaller scale analogs of the uncovered waste rock piles.  1.3.2.3. Major results on flow mechanisms Results from time domain reflectometry (TDR) sensors installed in the waste rock piles captured the initial wetting-up of the pile’s matrix fraction and showed a seasonal freeze-thaw cycle occurs yearly at Diavik (Fretz et al., 2012).  For the former, the matrix portion of waste rock was at, or below, residual saturation during pile construction (Neuner, 2009) and early infiltration to the piles contributed to the storage within CHAPTER 1  10  the matrix porosity.  The early contributions to water storage were recorded by TDR sensors and the evolution of moisture contents at depth describes a clear wet-up period of the pile.      The observation of an annual oscillating response of volumetric moisture contents is associated with waste rock undergoing freeze/thaw cycles due to seasonal changes in air temperatures (Fretz et al., 2012).  Specifically, as air temperatures drop below freezing the TDR sensors respond with decreasing moisture content values due to differences in the dielectric properties of water and ice.   Infiltration into the two uncovered piles is variable and based on the amount, timing and magnitude of precipitation.  Specifically, net infiltration is higher following smaller rainfall events that occurred early and late in the rain season, which is likely attributed to lower solar radiation and temperatures and a lower evaporative component (Fretz et al., 2012).  Yearly infiltration to Diavik waste rock piles ranges between 40 – 60% from 2007 - 2011, with the exception of 2009 that reported 15% infiltration.   Flow regimes in unsaturated waste rock at Diavik are dominated by matrix flow or water infiltrating matrix flow paths and controlled by capillary tension.   Preferential flow, where observed, is typically limited to unique scenarios of large magnitude/intensity rainfall events in late summer when evaporative losses are reduced (e.g., 36 mm over 25 hours, August 2008; Fretz et al., 2012; Momeyer, 2014).  A study by Bailey et al. (2013) used blasting residues (i.e., NO2-, NO3- , NH4+, Cl-, ClO4-) as a resident tracer and showed the timing of tracer breakthroughs were correlated with breakthrough estimates using matrix porosities within active zone layers (Bailey et al., 2013).  Overall flow rates during summer or active flow periods ranges between 7 and 10 mm per day, which is equivalent to 1.2 and 1.7 years residence time (corrected for 137 days per year of active flow).    Bailey et al. (2015) conducted a microbiological characterization of waste rock containing 0.014 weight % S and 0.035 weight% S at Diavik.  Microbial enumerations of acidophilic and neutrophilic iron- and sulphur- oxidizing bacteria show acidophiles are more abundant in the higher S% waste rock, which also releases drainage with lower pH values and higher solute concentrations (Bailey et al., 2015).  Neutrophilic bacteria are more abundant in the lower S% waste rock material, which releases neutral-pH drainage and lower solute concentrations (relative to higher S% waste rock).  These results indicate the CHAPTER 1  11  colonization of S- and Fe- oxidizing bacteria occurs at cold temperatures prevalent at Diavik and acidic drainage, likely catalyzed by higher acidophilic bacteria populations, can occur from material with very low sulphur contents (Bailey et al., 2015).   1.3.3. Aitik Mine 1.3.3.1. Location, geology and climate The Aitik mine is a porphyry copper deposit situated approximately 15 km southeast of the town of Gällivare and 100 km north of the Arctic Circle in Sweden (Figure 1-3B).  The mine, currently operated by Boliden Mineral AB, opened in 1968 and is Sweden’s largest copper mine.  The main ore mineral from Aitik is chalcopyrite (CuFeS2) with a mean copper concentration of 0.4% and residual economic gold (0.2 g/t) and silver (3.5 g/t).  Chalcopyrite occurs as disseminated or veined occurrences hosted in a porphyritic diorite, which is mined from an open pit and processed by flotation.   The climate at Aitik is sub-arctic, with an average annual precipitation of 680 mm and average annual infiltration of 500 mm (Axelsson et al., 1992).  Annual mean temperatures are ~ 0oC, with average winter and summer temperatures of -15oC and +15oC (respectively).  Snowmelt occurs within a short time period between May and June, and results in freshet conditions of high flow and possible erosive settings.  Waste rock piles at Aitik are unsaturated (i.e., average volumetric water content = 0.10 m3·m-3; Gibson et al., 1992) and are relatively coarse-grained.    1.3.3.2. Aitik research project Although a waste rock pile was not constructed at the Aitik mine, the majority of the research focuses on drainage from a 200 m (w) x 400 m (l) and 24 m high waste rock pile containing approximately 3 million tonnes of material.  Initially, ARD was not believed to be a principal concern due to the low sulphur percentage (i.e., <0.7% sulphide volume percent) of the waste rock (Lindvall and Eriksson, 2005); however, ARD has developed from approximately 400 ha of waste rock piles.  The drainage chemistry from these diversion ditches surrounding waste rock piles is monitored at regular intervals and commonly used as process waters at the mine’s mill (Eriksson et al., 1997).   CHAPTER 1  12  Seven columns were constructed in the laboratory, with one column at 0.2 m diameter and 1 m in height (C0) and six columns at 0.8 m diameter and 2 m in height (C1 through C6).  Laboratory conditions maintained a temperature of approximately 4oC, for the purpose of mimicking field conditions.  Field lysimeters were installed in Aitik waste rock, approximately 2 m below ground surface and 0.49 m in diameter.  In total, these field systems include 1.5 m of overlying waste rock (minus large boulders) and approximately 40 cm of basal gravels/sands to convey infiltration for leachate collection.  Further description of these field lysimeters and seven laboratory columns can be referenced in Eriksson et al. (1997).  1.3.3.3. Major results Flow rates and major cation concentrations measured from drainage ditches associated with the 3 million tonne waste rock pile generally remain constant throughout the year, with the exception of a 2 – 10x increase during spring freshet periods (Strömberg and Banwart, 1994).  A geochemical model was developed by Strömberg and Banwart (1994, 1995) to characterize leachate at the Aitik mine site.  Modelled leachate predictions were compared to drainage waters from the D1 and D2 ditches to evaluate its accuracy/relevance to site-specific geochemical processes.  The comparison results showed two points:  Weathering rates, normalized to physical surface area and temperature corrected activation energies, are ~ 100x lower in the field relative to laboratory experiments; and,  The overall Fe2+ oxidation rate is ≥ 1000 x faster than abiotic kinetics.   For the latter point, it was previously assumed that low temperatures would depress microbial activities at Aitik (i.e., Jaynes et al., 1984; Strömberg and Banwart, 1994).  However a higher iron oxidation rate suggests the presence of iron-oxidizing bacteria should be considered in future models as microbiological activity has the potential to increase sulphide (pyrite) oxidation by up to 5 orders of magnitude over abiotic conditions in acidic environments (Singer and Stumm, 1970; Southam and Beveridge, 1992).   Oxygen was considered to be the dominant oxidant for sulphide oxidation at Aitik due to high laboratory and field oxygen measurements (i.e., 10 – 20 volume % O2; Bennett et al., 1992) and high field-measured CHAPTER 1  13  oxygen consumption rates.   However, the inconsistency between field and laboratory weathering rates suggests another factor(s) controls these measured values.  A study by Eriksson et al. (1997) was conducted to investigate the occurrence of preferential flow paths in Aitik waste rock to explain lower field-measured weathering rates.  This study involved the seven laboratory columns and four field lysimeters and the application of a tracer test (i.e., bromide (as LiBr) and uranine) to quantify effective flow parameters.  Flow parameters were calculated from breakthrough curve (BTC) results, which are presented on a flow-corrected time scale, and analyzed using temporal moments and advection-dispersion modeling. Results from this study show preferential flow paths are activated in field lysimeters only (Eriksson et al., 1997).  The main difference between these two scales is the complete exclusion of boulders in the laboratory experiments.  Although these large boulders constitute a small proportion of the overall particle size distribution used in field lysimeters, the occurrence of preferential flow in the field tests indicates the presence of larger waste rock particle sizes plays a significant role in the overall flow regimes of unsaturated waste rock from the Aitik mine (Eriksson et al., 1997).   1.4. Recommendations for the Antamina Study Table 1.1 is a summary of the literature review of the three waste rock studies described in the previous section that employed mesoscale test piles and/or multi-scale experiments.  It is acknowledged that the contributions from these studies are considerable and not entirely captured in the results summaries provided in Section 1.3, however, this review and comparison identified three main areas that should and will be addressed with the Antamina project and this thesis work.  These three areas are discussed in the following sections.  1.4.1. Preferential Flow versus Matrix Flow As shown in all three studies, unsaturated flow through waste rock piles includes both matrix and preferential flow paths.  Critical to the identification of preferential flow in these studies was the application of a tracer test.  Therefore a similar technique was applied in this thesis research to elucidate the presence/absence of preferential flow paths in Antamina waste rock.  Although preferential flow was CHAPTER 1  14  minor in the Diavik waste rock project, the use of a resident tracer (i.e., blasting residues) as well as applied tracer events (Bailey et al., 2013; Neuner et al., 2013; Momeyer, 2014), provided an estimate of internal flow regimes and initial weathering rates as a result of blasting.  The Antamina mine uses a similar blasting material as Diavik (i.e., ANFO), and measurement of solutes associated with blasting residues aimed to provide another perspective of unsaturated flow paths in Antamina waste rock.   The presence of some degree of preferential flow also supports the suggestion that classical, uniform models may not adequately simulate flow and solute transport processes in many soils (e.g., Hendrickx and Flury, 2001; Köhne et al., 2006; Pot et al., 2005; Šimůnek and van Genuchten, 2008).  Specifically, flow and solute transport through unsaturated media can be conceptualized using a variety of model approaches (Figure 1-4). The simplest is the uniform model, which assumes the porous medium is a collection of impermeable soil particles separated by pores or fractures through which flow and transport takes place (Šimůnek and van Genuchten, 2008).  Uniform flow is generally described by the Richards equation (i.e., refer to Equation 1-4) and solute transport is characterized using the classical advection-dispersion equation, which is shown in one-dimension in Equation 1-5.   𝜕𝐶𝜕𝑡= 𝐷𝑥𝜕2𝐶𝜕𝑥2− 𝑣𝑥𝜕𝐶𝜕𝑥 (Eq. 1-5) Where C is the concentration [M∙T-1], t is time [T], x is the distance [L], D is the hydrodynamic dispersion coefficient [L2∙T-1], and v is average linear velocity [L∙T-1] The other three models shown in Figure 1-4 increase in complexity from left to right and are examples of dual domain approaches.  In dual domain scenarios, soil particles are assumed to have their own microporosity and the porous domain is divided into mobile and immobile partitions (Šimůnek and van Genuchten, 2008).  The mobile-immobile or MIM approach is conceptualized as uniform water flow, similar to the uniform model and described by the Richards equation, whereas dissolved solutes may move into and out of the immobile domain by molecular diffusion (e.g., van Genuchten and Wierenga, 1976).   CHAPTER 1  15  In a dual-porosity scenario, both water and solute can move between the immobile and mobile domain (Šimůnek et al., 2003).  As well, solute can move into the immobile domain of a dual porosity model by both molecular diffusion and advection with flow (exchanging) waters.  In a dual-permeability model, the immobile and mobile domains are considered to be ‘overlapping’ such that water and solute can move between immobile and mobile domains, as well as between immobile domains.  Further details regarding these model approaches and their governing equations may be found in Šimůnek and van Genuchten (2008) and references therein.  Unsaturated porous media that present preferential flow characteristics may be better simulated using a dual permeability model approach; however a drawback of using increasingly complex models is an increasing level of uncertainty may be derived from estimating a larger number of variables.  In addition to using tracer tests to identify the degree (if any) of preferential flow in Antamina waste rock, this thesis also complemented measured tracer breakthrough curves with the use of analytical solutions and numerical modeling to simulate flow and solute transport.   The use of numerical models to conceptualize and parameterize flow and solute transport can be highly beneficial in deciphering the in situ complexity associated with highly heterogenous, field-scale waste rock piles.     1.4.2. Particle Size and Scale-dependence As was previously stated, grain size distributions of waste rock material typically span six orders of magnitude, from clay- to boulder- sized particles (Chi, 2011; Neuner et al., 2013; Nichol et al., 2005).  The heterogeneous clast composition of waste rock is a function of many factors (i.e., mineralogical assemblage, blasting technique, waste deposition method; Smith and Beckie, 2003).  The final particle size distribution can influence both flow and solute parameters, namely:   Infiltration rates;   Flow rates;   Proportion of preferential to matrix flow paths; and,   Mineral reaction/dissolution rates.   CHAPTER 1  16  Infiltration is dependent on the grain size of surface materials and can be spatially variable.  Furthermore, the rate of infiltration subsequently impacts the flow rate of infiltrating waters and the proportion of preferential to matrix flow paths.  For example, at low infiltration events, water infiltrates finer grained material or matrix flow paths and is held under capillary tension (Nichol, 2002).  At higher infiltration events, water infiltrates coarser grained material and enters larger pores (i.e., macropores) that are hydrologically effective at channeling water (Beven and Germann, 1982; Bews et al., 1997; Birkholzer and Tsang, 1997).   Particle size can indirectly or directly influence mineral reaction rates.  Specifically, particle size indirectly impacts infiltration and flow rates through preferential and/or matrix flow paths whereby low infiltration rates, supporting mostly matrix flow, result in longer water-rock interactions (relative to preferential flow) and therefore higher concentrations (Smith and Beckie, 2003).  The opposite is typically observed in waste rock piles dominated by preferential flow paths, and shorter water-rock interactions result in lower concentrations.   Particle size can also directly influence mineral reactions rates, such that smaller particle sizes are associated with higher weathering rates.  At the Aitik mine, Strӧmberg and Banwart (1999b) found particle diameters less than 0.25 mm contributed to approximately 80% of the sulphide and silicate dissolution.  More recent studies of mine wastes by Sracek et al. (2004) show sulphide weathering rates are higher in more friable material, even if sulphide contents are similar between variably friable materials.         The influence of particle size on observed flow and solute transport suggests it is a critical factor in the success or failure of scale-up studies.  This was demonstrated at the Aitik site as tracer breakthrough curves from field lysimeters signified preferential flow paths were present at this scale and were not observed in the laboratory.  The major difference between the two experiments was the complete exclusion of boulder-sized particles in the laboratory (Strӧmberg and Banwart, 1999a).   A study by Malmström et al. (2000) on Aitik mine weathering rates at three scales (i.e., laboratory column, field lysimeter, field-scale waste rock pile), showed scale-up correction factors for particle size were empirical and site-specific.  Therefore the impact of particle size on scale-up is considered in this thesis research and should be a consideration for most (if not all) mine sites that rely on laboratory-scale experiments.  CHAPTER 1  17  1.4.3. Microbial Activity It is well-recognized in the literature that microbial processes facilitate the formation of ARD/NRD and can increase sulphide weathering rates by several orders of magnitude over abiotic conditions (e.g., Colmer and Hinkle, 1947; Jambor and Blowes, 1994; Olson, 1991; Sand et al., 1995; Schippers et al., 1996; Schippers and Sand, 1999; Singer and Stumm, 1970; Southam and Beveridge, 1992).  In particular, iron-oxidizing bacteria in acidic environments mediate the cycling of ferrous to ferric iron, the latter of which is a more effective oxidant (relative to oxygen) of most sulphide minerals (Luther, 1987).   Acidophilic iron- and sulphur- oxidizing bacteria are considered to be ubiquitous in pyrite-bearing environments (Edwards et al., 1999) and can be found in environments with variable pH values.  Specifically, numerous studies have cultured acidophilic bacteria from neutral pH mine wastes (Blowes et al., 1995; Blowes et al., 2003a; Moncur et al., 2005) and it is proposed that bacteria form microenvironments to armour microbes from unfavourable conditions (Mielke et al., 2003; Nordstrom and Southam, 1997; Southam and Beveridge, 1992).  This may be a possible explanation for the high iron oxidation rates observed at the Aitik mine and recognition that lower temperatures do not inherently equate to depressed microbial activities (Strömberg and Banwart, 1994).  Unsaturated waste rock piles may provide essential physical and chemical requirements for microbial growth (Harrison, 1984).  Indeed capillary borders on mineral surfaces, found in partially drained piles, provide water for bacterial growth and a protection against drying (Mielke et al., 2003).  Microbial enumerations of waste rock by Bailey et al. (2015) support the presence of bacteria in unsaturated waste rock environments, in spite of the cold temperatures and low sulphur contents at Diavik.  The presence of microbes on Antamina waste rock surfaces was confirmed by microbial investigations by Dockrey et al. (2014), which revealed the presence of acidophilic microenvironments on a significantly weathered iron sulfide grain.  A few studies have investigated the diversity of microbes in neutral environments and their role on weathering rates and observed solute loads.  Bacterial enumerations of tailings materials from the Greens Creek Mine (Alaska, USA) by Lindsay et al. (2009) revealed a diverse consortium containing autotrophic neutrophilic and acidophilic sulfur-oxidizing bacteria and iron- and sulphur-reducing bacteria.  CHAPTER 1  18  A study by Power et al. (2010) revealed the presence of Acidithiobacillus spp. in chrysotile-bearing tailings helped to catalyze a 10x higher release of magnesium relative to control systems.  As mentioned by Lottermoser (2010), the knowledge of microbiology of neutral mine wastes is incomplete.  Therefore microbial work conducted as part of this thesis research built upon previous work by Dockrey et al. (2014) to advance the research of microbial diversity in NRD mine wastes and microbial impact(s) to mineral weathering rates.     1.5. Thesis Scope and Organization  The research included in this thesis involves experimentation and data collection at three scales; two constructed waste rock piles (i.e., 10 m in height; 19,000 tonnes), fifteen field barrels (i.e., <1 m in height; ~ 300 kg) and two laboratory-based column experiments (i.e., <1 m in height; ~150 kg).  Data was collected from constructed waste rock piles and field barrels for 3.5 years, between 2009 and 2012 whereas laboratory column experiments were conducted for approximately 1 year and mimicked monthly rainfall amounts during the first year of field site conditions.  Additional experimentation included a 2 year tracer test on constructed piles and a 4 month tracer test on laboratory column experiments.  Measured flow and solute transport was incorporated in numerical and/or geochemical model simulations to permit a quantitative analysis of the interaction between hydro-geochemical processes and provide insight to observed/measured results.   Specific research objectives/questions of this thesis are outlined below:  Quantify infiltration, evaporation, storage and drainage of constructed waste rock piles to understand water balance components of waste rock in a bimodal, wet-dry climate.    Evaluate flow mechanisms/regimes and the proportion of preferential to matrix flow paths associated with unsaturated waste rock at variable experimental scales.  Can flow and solute transport from unsaturated waste rock be reasonably simulated using a uniform flow model, or does it necessitate a dual domain model approach?   Similar to the work by Bailey et al. (2013) at the Diavik mine, can blasting residues be used as a resident tracer to further the understanding of internal flow regimes?  CHAPTER 1  19   Can larger structures, such as waste rock piles, be adequately replicated in the laboratory by accurately defining the particle size discrepancy between the tested scales?   Following work by Dockrey (2010) and Dockrey et al. (2014), how does microbial community/diversity impact geochemical reactions within waste rock piles?     Quantify the impact of microbial activity on mineral reaction rates and waste rock drainage chemistries. The amalgamation of these specific objectives has a high-level purpose to add to the knowledge base for three main research areas, namely: unsaturated waste rock hydrology, geochemical processes associated with NRD mine wastes, and scale-up.  This thesis consists of seven chapters, Chapter 2 and 3 provides a detailed description of three experimental scales used in this thesis, namely: field-scale waste rock piles (Chapter 2) and smaller scale field barrels and laboratory columns (Chapter 3).  These chapters present data used to develop conceptual models of flow and solute transport from the Antamina mine. Chapter 4 presents a comparison of unsaturated flow and solute transport through waste rock at two experimental scales using temporal moment analysis and numerical modeling.  Chapter 5 presents results from tracer tests to investigation preferential and matrix flow in waste rock.  Chapter 6 investigates microbiological and geochemical controls on sulphide oxidation from unsaturated waste rock.  Chapters 4 through 6 are intended as submissions to peer-reviewed scientific journals and satisfy the major thesis objectives outlined above.  Chapter 4 has been published in the Journal of Contaminant Hydrology (Vol. 171 (2014); page 49 – 65).  Chapter 7 synthesizes the salient research findings, their relevance to waste rock piles at other mine sites, their implications for Antamina’s waste rock management, and provides recommendations for future work.     CHAPTER 1  20  1.6. Tables Table 1.1 Comparison of Three Studies using Mesoscale and/or Multi-scale Waste Rock Experiments.  Cluff Lake Mine Diavik Mine Aitik Mine Temperature (oC)  [Average range] Mean value  [-40 – +35] 0  [-44 – +27] -12  [-15 – +15 ]  0 Annual Precipitation (mm) 455 280 680 Infiltration (%) 55 40 - 60† 77 Pile footprint (m) (w x l x h) 8 x 8 x 5 50 x 60 x 15 200 x 400 x 24 Drainage collection system 16 – 2 m x 2 m basal sub-lysimeters 1 Basal lysimeter (~2800 m2) 4 – 4 m x 4 m and 4 – 2 m x 2 m basal sub-lysimeters 2 – 4 m x 4 m and 2 – 2 m x 2 m upper sub-lysimeters Drainage ditches Total sulphide %  0.64 <0.04 (Type I) 0.04 – 0.08  (Type II) > 0.08 (Type III) ~0.7 Mean flow velocity (m/yr) 1.1 – 1.8 5.5 - 11 0.11 – 2.1 Preferential flow (%) 0.1‡, 6§ n/a¶ 55 – 70 † Inifiltration calculated as a percentage of rainfall, which is 65% of total annual precipitation ‡ Non-capillary flow (Nichol et al., 2005) § Macropore flow (Nichol et al., 2005) ¶ Richard’s equation modeling and measured parameters were successful in simulating wetting front velocity and cumulative outflow volumes from high-intensity rainfall events, thereby suggestive of a low to insignificant preferential flow component to waste rock flow regimes at Diavik (Neuner, 2009).     CHAPTER 1  21  1.7. Figures  Figure 1-1. Location of mining projects in Canada. Base metals = green; Precious metals = dark blue; base metals/precious metals = purple; Iron ore = dark brown; Uranium = olive green; other metals = light grey; Industrial minerals = light brown; diamonds = agua-blue; Coal = black; Upgraded crude oil = burgundy. Oil and Gas: Bitumen = organe; Gas = blue; Oil = dark brown; Oil/Gas = Grey http://geoappext.nrcan.gc.ca/MMS/MIB.Map.Presentation.aspx?config=configMMS_e.xml&language=en&styletype=Design (from NRCan, 2013)   CHAPTER 1  22             Figure 1-2. Location of the Antamina Mine. Note: modified from Dockrey, 2010.200 km CHAPTER 1  23             Figure 1-3. Map location of three waste rock pile studies; Cluff Lake uranium mine and Diavik diamond mine (A) and Aitik copper mine (B). Cluff Lake Mine Aitik Mine A B CHAPTER 1  24       Figure 1-4. Comparison of four conceptual models for water flow and solute transport.  Note: Adapted from Šimůnek and van Genuchten (2008). Water Solute Immob. Mobile Water Solute Immob. Mobile Immob. Mobile Water Solute Slow Fast Slow Fast Mobile-Immobile (MIM) Dual-Porosity Dual-Permeability Water Solute Uniform CHAPTER 2  25  CHAPTER 2: METHODOLOGY AND CONCEPTUAL MODEL DEVELOPMENT – EXPERIMENTAL WASTE ROCK PILES 2.1. Introduction Many of this thesis’ objectives can be singularly described as: assessing and quantifying flow/solute transport through Antamina waste rock and evaluating if and how manageable, smaller-scale experiments can be used to predict the behavior of its larger mine-scale structures.  Central to this thesis is an understanding of mechanisms and controls on infiltration through (and drainage released from) waste rock produced at the Antamina Mine (Antamina, Peru).   The majority of data used in this thesis is based on two (of 5) experimental waste rock piles (Pile 4 and 5) constructed at Antamina between 2008 and 2009.  The waste rock piles are 10 m in height and contain approximately 19,000 tonnes of waste rock that were deposited in three to six waste rock discharges of smaller tonnages (i.e., 1000’s tonnes).   Other experiments used in this thesis include two ~1 m tall columns containing material similar to Pile 5 only and 15 (of 45) 1 m tall field barrels, which contain material comparable to waste rock used in Pile 4 and/or 5.  The field barrels were constructed at the same time as Pile 4 and 5 (i.e., 2008 – 2009), whereas the laboratory column experiments were completed in 2011.  Further details of these smaller scale (i.e., ≤1 m) experiments are discussed in Chapter 3.   This chapter is divided into four parts, namely: 1) A description of the waste rock classification scheme employed at Antamina, which is applied to waste rock material in all experiments used for this research; 2) The construction/instrumentation and data collection of Pile 4 and 5;  3) A description of the dominant flow characteristics and drainage chemistries; and, 4) Development of a conceptual model to describe infiltration in the Pile 4 and 5 waste rock experiments.       A conceptual model can be defined as a simplified representation of the hydrologic system that is to be modeled (Anderson and Woessener, 1992) and typically considers three main aspects; processes, scale CHAPTER 2  26  and objectives (Sivakumar, 2007).  The conceptual model for flow and solute transport associated with Pile 4 and 5 is described in this chapter and establishes the foundation for detailed quantitative analysis provided in subsequent chapters.  Specifically, the degree of preferential to matrix flow and the characteristics/parameters associated with either flow paths are quantitatively assessed in Chapter 4 and 5.  Chapter 4 looks examines a vertical cross section through Pile 5 and, using tracer breakthrough curves, estimates flow and solute parameters using a temporal moment analysis and complementary model simulations.  Chapter 5 applies the same parameterization method to both Pile 4 and 5 and extends the analytical approach to include breakthrough curves of blasting-related solutes to estimate matrix flow characteristics. 2.2. Antamina Waste Rock Classification The Antamina mine is a Cu-Zn skarn deposit that began production in 2001 and contains proven reserves of Cu, Zn, Ag and Mo.   The concentrator milling rate at Antamina is approximately 130,000 t per day, which equates to a daily production of 400,000 tonnes of waste rock (Harrison et al., 2012).  Waste rock produced at Antamina generally consists of highly carbonaceous assemblages, with varying sulphide mineralization and minor ore minerals, which is placed in one of two waste rock dumps reaching over 300 m in height (i.e., East Dump or Tucush Dump).  These waste rock dumps are located in two separate drainage basins and the placement of mine wastes at either location is primarily based on its classification, which is a function of its lithology (Section 2.2.1.1) sulphur content and metal content (Section 2.2.1.2) (Aranda, 2010).    2.2.1. Lithology An initial environmental impact assessment (EIA) on the Antamina deposit was completed in 1998 (KC-SVS, 1998) and included an extensive geochemical review of waste rock types and involved 179 acid-base accounting (ABA) tests.  Five types of waste rock material from Antamina were geochemically tested; brown and green garnet skarn, marble, limestone, and intrusive (Table 2.1).   Results from ABA tests are used to describe materials as ‘potentially acid generating’ or PAG material, ‘non-acid generating’ or NAG and ‘uncertain acid-generating potential’ (MEND, 2009).  These CHAPTER 2  27  classifications are based on neutralization potential ratio (NPR) values, which is the ratio of a sample’s neutralization potential (NP) and its acid potential (AP) values.  Criteria used to classify samples as PAG, NAG or ‘uncertain’ assumes there are no analytical errors in ABA test results, and assumes minerals contributing to a sample’s NP and AP (e.g., carbonates and sulphides, respectively) are readily available and not occluded.  A sample is described as NAG material if its NPR value is greater than 2, whereas a sample is described as PAG material if its NPR is less than 1.  Samples with NPR values between these two values (i.e., 1 ≤ NPR ≤ 2) are classified as ‘uncertain’ acid-generating potential.   Typically a conservative approach is applied such that samples with NPR values less than or equal to 2 are assigned a ‘PAG’ classification.  The application of these criteria indicate samples of skarn and intrusive material from Antamina (shown in Table 2.1) may be labeled as PAG or NAG material, whereas samples of marble and limestone were classified as NAG material.   2.2.2. Sulphides and Metal Content Sulphide percentages of the 179 samples were measured as part of ABA tests and are shown in Table 2.1.  The highest mean sulphide contents and maximum values were observed from brown garnet skarn samples, followed by intrusive materials.  High sulphide contents associated with brown garnet skarn samples, median sulphide values from intrusive material, and comparably low sulphide ranges from green garnet skarn, marble and limestone samples is suggestive of a geological control on measured sulphide proportions.  This observation is supported by the early study by Petersen (1965) that described the mineralized skarn assemblage at Antamina as being dominated by grandite garnet, which grades from brown to green with increasing proximity to the host limestone material.  Similarly, a later study of the Antamina deposit by Love et al. (2004) identified the mineralized skarn unit straddling the original intrusive contact between the porphyrytic monzogranitic stock and the Upper Cretaceous carbonate strata.   Each sample submitted for ABA analyses underwent solid-phase analysis, via total acid digestion and metal analysis using inductively coupled plasma (ICP) technique (KC-SVS, 1998).  Results from these analyses identified arsenic and zinc as primary metal(loid)s of concern (see Table 2.1).   CHAPTER 2  28  The highest median and maximum As concentrations were observed from brown garnet skarn, whereas limestone and marble contained the lowest mean and maximum As values.  Both brown and green garnet skarn and marble materials contained similar mean Zn concentrations; however marble samples presented 2x – 6x higher maximum concentrations relative to skarn maximum values.  Limestone and intrusive samples presented a narrower Zn concentration range and mean values are 3x – 10x lower than the other three waste rock types.   2.2.3. Classification Criteria The results from ABA and solid-phase metal analyses conducted on waste rock from Antamina are shown in Table 2.1.  Antamina’s classification scheme for waste rock is shown Table 2.2 and indicates all skarn and intrusive material is labelled Class A, whereas marble, hornfels and limestone material is classified as Class A, B or C based on its metal and sulphide compositions.    The classification of waste rock excavated at Antamina, based on the criteria shown in Table 2.2, designates its placement in one of two waste rock dumps.  Specifically, waste rock classified as Class A and B is placed in the East Dump and leachate from this waste rock pile flows into a tailings pond (Beckie et al., 2011).  Less reactive waste rock classified as Class C is placed in the East Dump.  The Tucush Dump only receives the lower reactivity Class C and a small percentage of Class B.  Tucush drainage flows into an engineered wetland (Beckie et al., 2011).   2.3. Experimental Waste Rock Piles Five (5) experimental waste rock piles were built at Antamina using run-of-mine waste rock and constructed according to the methods in Corazao Gallegos (2007) and Bay (2009).  Each pile is 10 m in height and has an areal footprint of 1,296 m2 (i.e., 36 m (w) x 36 m (l)).  The first three piles (Pile 1, 2 and 3) contain single waste rock classes (i.e., Class B, Class A-intrusive, Class A-skarn, respectively) whereas Pile 4 and 5, which are the focus of this research, contain mixed waste rock materials of Class B and C and Class A and C (respectively).  Pile 4 and 5 were constructed over the course of one year between 2008 and 2009, and data acquisition began on June 1, 2009.  This thesis uses data collected CHAPTER 2  29  over the period of 3.5 years (i.e., June 2009 to December 2012), which corresponds to 3.5 water years.  For this study a “water year” refers to June 1 through to May 31 of the following year.     2.3.1. Construction and Instrumentation An impermeable, bitumen-based geomembrane covers the 36 m x 36 m base of each pile and is described herein as the basal lysimeter or Lysimeter D.   Three smaller interior sub-lysimeters (Lysimeter A, B and C) are located along the centerline of each pile with a footprint of 4 m x 4 m.  Lysimeter A and B are located below the crown or upper trafficked surface and Lysimeter C below the outer slope (or batter) of the pile.   A schematic of Pile 4 and 5 is provided in Figure 2-1. Infiltration through the waste rock overlying these lysimeters is conveyed through corrugated 4" HDPE tubing to an instrumentation hut located at the base of each pile.  A tipping bucket is present at the terminus of each lysimeter’s tubing to record flow rate and volume, along with two sensors that measure electrical conductivity and temperature of drainage waters.  In total, there are 6 instrumentation lines located in each constructed pile (shown in Figure 2-1).  Lines 1 through 4 were installed sub-vertically from the crown to the base of the pile, along the 37o slope of waste rock discharges.  The installation of these instrumentation lines involved removing a 1 m – 2 m trench along its centerline to lay instrument wires (Figure 2-2).  Lines 5 and 6 are located along the base of each pile (refer to Figure 2-1A and B).  The positions of each sensor along the six instrumentation lines in Pile 4 and 5 are found in Appendix A.    Instrumentation lines include several sensors to record a variety of parameters.  Specifically, 10 thermistors were installed in each pile to measure temperature, along with 24 Decagon Devices Inc. (5TE) sensors to record temperature, electrical conductivity and moisture contents along the six instrumentation lines.   Five time domain reflectometer (TDR) probes were installed along the outer instrumentation line (i.e., Line 4) of each pile and provided additional information of the migrating wetting front.  The design and construction of TDR probes corresponded to methods described in Nichol (2002) and TDR soil moistures were measured with a Moisture Point MP-917 instrument and Campbell Scientific CR1000 datalogger.    CHAPTER 2  30  Fifteen soil water solution samplers (SWSSs) and 65 gas ports were also installed along each instrumentation line to sample in situ water and gas (respectively) migrating through the pile pore space.  The SWSS design was similar to those instruments installed at Cluff Lake, which sampled porewater from centimeters to tens of centimeters surrounding the tip of the porous cup.  As noted by previous studies (Harvey, 1993; Litaor, 1988; Parker and van Genuchten, 1984), SWSSs sample a biased fraction of the water present in the matrix, following the application of an artificial pressure gradient, and therefore are neither a volume-averaged nor a true flux-averaged concentration.  However, these instruments can measure wetting front migration, assuming infiltration is mostly governed by matrix flow.   Gas ports consist of one-eightth inch tubes and gas samples are collected by pumping gas from the interior of the pile through a fine mesh filter and measuring concentrations with a portable gas chromatograph (Beckie et al., 2011).  Seasonal and daily fluctuations of O2 and CO2 are measured using an Apogee O2 sensor (SO-112) and LiCOR (LI-820) CO2 gas analyzer.  The installation of instrumentation line sensors was a delicate process in order to ensure adequate contact between the surrounding soil matrix and sensors.  All sensor cables and tubing were housed within 2” corrugated plastic tubes to protect the sensors from damage during the remainder of pile construction.  Following placement of instrumentation lines, coarser cobbles/boulders were carefully placed (by hand) over top of the tubing and an excavator subsequently covered lines with previously trenched material.  An example of the instrumentation installation process is shown in Figure 2-3, which shows the installation of instrumentation line 4.    Data outputs from all sensors located along an instrumentation line are connected to a CR1000 datalogger located in the instrumentation hut.  Power for this datalogger and other instruments is supplied by a 56 W solar panel, which was installed in 2009 at the Antamina site.  Further details of these sensors can be found in Peterson (2014), which describes a similar set up for Pile 1 through Pile 3.  2.3.2. Waste Rock Physical Characteristics Physical characteristics of waste rock play an important role in the unsaturated hydrology of waste rock.   Specifically, the parameterization of the physical characteristics of waste rock (e.g., total mass, particle CHAPTER 2  31  size distribution, bulk density, porosity, hydraulic conductivity) are necessary to complement conceptual models, and are used as input parameters for numerical simulations.  Those parameters without estimated or measured values introduce assumptions and/or uncertainty to model simulations.  Therefore every effort should be made to constrain parameter estimates using tests and measurements of site-specific materials.  The following sections briefly describe the physical characteristics of waste rock used in waste rock piles (i.e., Pile 4 and 5).        2.3.2.1. Total mass Approximately 20,000 tonnes of waste rock was used in each of Pile 4 and 5, which was built in five and six smaller discharges, respectively.  Each discharge is labeled by a reference polygon, which relates to a unique location in Antamina’s open pit.  In Pile 4, two (of 5) discharges include Class B material that contributed to approximately one-quarter of the pile’s total waste rock tonnage (as equal parts hornfels and marble rock types).  The remainder (i.e., 75%) of Pile 4, or three (of 5) discharges, is of Class C waste rock (i.e., hornfels material; Table 2.3).   In Pile 5, three (of 6) waste rock discharges consist of Class A material and the other half as Class C material (Table 2.4).  Individual discharges of Pile 5 contain similar waste rock tonnages, such that 51% (by weight) of Pile 5 consists of Class A waste rock (i.e., quartz-monzonite intrusive and lesser skarn) and the remainder as Class C waste rock (i.e., mostly hornfels with trace marble).  The design and composition of Pile 4 and 5 was based on constructing a mixed waste rock pile that represents a smaller scale analog of material deposited in the two mine-scale waste rock dumps at Antamina.  The Tucush Dump at Antamina receives primarily Class C with lesser Class B waste rock, whereas the East Dump receives all Class A material, some Class C and a small amount of Class B material.      Pile 4 and 5 were constructed in tandem at Antamina, such that Class C waste rock used in the first and fifth discharges were comparable between the two piles.  This is shown in Table 2.3 and 2.4 by identical reference pit location numbers for these Class C discharges.  CHAPTER 2  32  2.3.2.2. Particle size distribution (PSD) curves Particle size distribution (PSD) analyses were conducted on eight waste rock samples used in each of Pile 4 and 5.  These analyses were conducted by Golder Associates and adhered to standard guidelines regarding the PSD methodology (i.e., ASTM D5519-94).  Figure 2-4 presents results from PSD analyses on waste rock from each pile.  For reference, PSD analyses described as ‘Pile 4/5 – Class C’ in Figure 2-1 relates to the first or fifth discharges in both piles, which contained comparable Class C waste rock material.  For comparison purposes, average Class B curves (from Pile 1) and Class A (from Pile 2) are shown in addition to Pile 4- and 5- specific data.  Full PSD results can be found in Appendix A.  Although large boulders are present in constructed experimental piles at Antamina, they were relatively infrequent and do not contribute to a significant portion of the pile’s total mass.    The hydraulic properties of waste rock are heavily influenced by its PSD (Chapter 1).  Waste rock particle diameters can range over more than six orders of magnitude.  A study by Yazdani et al. (2000) showed that particles with grain diameters greater than 4.75 mm do not have significant capillarity in unsaturated conditions.  Therefore the amount passing the 4.75 mm sieve size can be used to characterize waste rock as relatively coarse or fine.  Waste rock used in Pile 4 and 5passed this sieve size and suggests materials used in both piles is relatively coarse and likely support preferential flow paths.   The calculation of a uniformity coefficient (𝐶𝑈; [-]) is another method to characterize material by PSD curves as well-graded versus poorly-graded.  Specifically, a material’s 𝐶𝑈 describes the ratio between the size fraction diameter representing 60% of the total sample (𝐷60) relative the 10% size fraction (𝐷10).   Samples with 𝐶𝑈 values greater than or less than 20 are described as well-graded and poorly graded, respectively (Morin et al., 1991).  Calculated 𝐶𝑈 values of Pile 4 and 5 waste rock (Figure 2-1) show Class A and B waste rock can be classified as ‘well-graded’ and Class C as ‘poorly graded’ material.   2.3.2.3. Density, porosity and saturated hydraulic conductivity The measurement of wet and dry densities (𝜌𝑡and 𝜌𝑏; [M∙L-3]) of two Class C samples from Pile 4 and 5 were conducted by Golder Associates, according to ASTM D5030-89 standards.  These tests were based CHAPTER 2  33  on 6.8 and 5.5 tonnes of material per sample and estimates of wet density and dry density were 1,947 kg∙m-3 and 1,960 kg∙m-3 versus 1,909 kg∙m-3 and 1,928 kg∙m-3 (respectively).   Density measurements are not known to have been conducted on Class A and B materials associated with Pile 4 and 5; however, density analyses were conducted on Class A material from Pile 2 by Golder Associates (Peru).  Wet and dry density measurements on Pile 2 Class A were estimated as 1,953 kg∙m-3 and 1,928 kg∙m-3, respectively (Peterson, 2014), which are similar to Class C density measurements.   Bay (2009) inferred a bulk density of Class B material contained in Pile 1 by using GPS data (to calculate areal volume) and total mass estimates used to construct the waste rock pile.  From this method, Bay (2009) estimated bulk density of 2,000 ± 400 kg∙m-3 for Class B waste rock in Pile 1. Porosity measurements (𝜃𝑇; L3∙L-3]) were performed on all three waste rock classes by Golder Associates (Peru) at the Antamina mine and/or the University of British Columbia (UBC).   Porosities were measured by a water displacement method, which included saturating the pore space of dried waste rock and allowing the water to displace air-filled voids over the period of one day.  Tests performed by Golder include material with particle diameters less than 2 m, whereas those tests conducted at UBC used grain sizes passing the 3 inch sieve size.  Average porosity measurements for Class A, B and C waste rock were estimated as 0.33 m3∙m-3, 0.28 m3∙m-3and 0.29 m3∙m-3 (respectively). Saturated hydraulic conductivities (𝐾𝑠𝑎𝑡; [L∙T-1] of waste rock were measured at site using single-ring (area = 0.28 m2) infiltrometer tests.  Approximately 8 to 10 infiltrometer tests were performed on the upper traffic surface (or crown) of Pile 1 (Class B) and 2 (Class A – intrusive) waste rock piles during March and April, 2009.  Average 𝐾𝑠𝑎𝑡 values of Pile 1 and 2 materials were calculated as 1.3 x 10-5 m∙s-1 (Class A – intrusive) and 2.7 x 10-5 m∙s-1 (Class B), with full results in Appendix A.  Infiltrometer tests were conducted prior to the completion of Pile 4 and 5 and therefore 𝐾𝑠𝑎𝑡 measurements are not available for Class C material.   2.3.2.4. Soil water characteristic curves (SWCCs) The results from PSD curves, density analyses and porosity measurements were used by Speidel (2011) to calculate material-specific soil water characteristic curves (SWCCs).  SWCCs were generated using CHAPTER 2  34  SoilVision software (Fredlund, 1996) and results are shown in Figure 2-5.  Further details can be found in Speidel (2011) and full results of SWCC values can be found in Appendix A.   The SWCCs indicate Class A material had the highest air-entry value (AEV), which is an order of magnitude greater than AEV values associated with Class B and C material (respectively).  Estimated AEV values are an inverse of the mean particle diameter and this result implies Class A material was more susceptible than Class B or C waste rock to ponding and run off during large rainfall events due to finer grain sizes.  In contrast, at low rainfall rates, infiltration was more likely to infiltrate Class A matrix material relative to Class B or C material.  2.3.3. Waste Rock Geochemical Characteristics 2.3.3.1. Bulk geochemical analysis Following the excavation of waste rock from the Antamina open pit, a sub-sample (i.e., 0.25 g) of material was submitted for bulk geochemical analysis.   Geochemical tests were carried out by ALS Chemex (Peru) via a four acid digestion with inductively coupled plasma metal analysis.  Results are shown in Table 2.5 and 2.6 for Pile 4 and 5 waste rock (respectively).   Bulk geochemical results show most concentrations (i.e., Ag, As, Co, Fe and Mo) were reasonably similar from all samples analyzed.   However, the small sub-sample analyzed per waste rock discharge did reveal some material-specific characteristics.  Concentrations of Zn, Bi and (to a lesser degree) Pb in Pile 4 were higher in Class B waste rock relative to Class C samples.  In Pile 5, Cu concentrations from Class A waste rock were greater than 10x those values from Class C waste rock.   2.3.3.2. X-ray fluorescence (XRF) Ten samples of the three waste rock classes were submitted for geochemical characterization using X-ray fluorescence (XRF).  Four (of 10) samples were associated with Pile 4 and 5 materials, with 1 Class A sample (Pile 5) and three Class C samples (Pile 4 and 5).  The remaining six samples were associated with Class B material from Pile 1 and Class A (intrusive) material from Pile 2.   CHAPTER 2  35  Whole-rock analyses included analyses of major oxides using fusion XRF (i.e.,SiO2, TiO2, Al2O3, Fe2O3, MnO, MgO, CaO, K2O, Na2O, P2O5 and L.O.I.).  Trace element analysis of Pb, Ga, As, Zn, Cu, Ni, Co, Mn, Cr, V, Ba, Ti, La, Ag and Mo was provided using pressed powder pellet X-ray diffraction (XRD) and percent carbon and sulphur values were determined using a LECO CNS-analyzer.  All equipment was provided courtesy of the Biotron Institute for Experimental Climate Change Research at the University of Western Ontario (London, Ontario, Canada).  Results are provided in Table 2.7, with full results found in Appendix A. Class A material (n = 6) contained a significantly lower (> 10x) CaO value and up to 4x higher SiO2 weight % values, relative to those values from Class B and C samples.   This distinction between Class A and Class B or C material indicated that Class A contained a significantly higher silicate, as opposed to carbonate, mineral composition.   In regards to metal contents, all samples presented similar Ag, Co, Cr and Ni concentrations with a few waste rock-specific enrichments or higher concentrations noted, namely:  Class A: Cu  Class B: Pb and Zn  Class C: As The presence of high Cu and Zn in Class A and B material was complementary to results from bulk geochemical analyses.    Although As was higher in Class C material, it is within the range of As concentrations (i.e., <400 ppm; Table 2.2) required to classify this material as Class C as opposed to Class B or A.  2.3.3.3. X-ray diffraction (XRD) The mineralogical compositions of eight samples from Pile 4 and 5 waste rock (i.e., Class A: n = 2; Class B: n = 3; Class C: n = 3) were analyzed using XRD to complement the preliminary bulk geochemical and XRF characterizations.  For each sample, a small representative sub-sample (i.e., <100 g) was pulverized to <10 µm using a McCrone micronizing mill and analyzed for mineralogy on a Siemens D8000 Diffractometer.  Qualitative identification of mineral phases was completed by referencing X-ray power diffraction patterns using EVA software.  Comparative quantitative analyses were completed using Topas CHAPTER 2  36  v.3.0 software in order to determine relative proportions of minerals detected in each sample.  XRD analyses were performed at The University of British Columbia (Vancouver, BC) and more information of these analyses/techniques can be found in Peterson (2014).  Results are provided in Table 2.8.   The mineralogical composition of Class A material supported previously described trends from XRF analyses.  Specifically, calcite was not detected in Class A material, which was suggested by low CaO percentages from XRF data, and supported the absence of acid-buffering carbonate minerals.  Secondly, Class A samples were comprised of ~ 50% silicate minerals (i.e., oligoclase and K-feldspar) and coincident to previously noted high SiO2 values (refer Table 2.7).  Thirdly, chalcopyrite was only detected in Class A material and likely the host mineral for the previously noted high copper concentrations (refer to Table 2.6 and 2.7).  XRD analyses also noted the presence of the garnet mineral andradite in Class A material only (i.e., Pile 5, discharge 2), which supports the geological association between skarn and intrusive units as noted by Love et al. (2004).     Both Class B and C materials had high proportions of calcite, which comprised of the majority its buffering capacity and is supported by the previously noted high CaO values from XRF analyses (Table 2.7).  Silicate mineral buffering also contributed to Class B and C neutralization capacity, as shown by moderate proportions of anorthite.   The major acid-generating minerals observed in Class B and C material were pyrrhotite and (lesser) pyrite, with Class B having a slightly higher sulphide proportion (than Class C).  2.3.4. Water Balance Recording and Calibration The volume and rate of infiltration reaching the base of the waste rock pile is recorded by a tipping bucket, which is based on a design described in Bay (2009) and Bay et al. (2008).  Each tip of a tipping bucket contains a specific volume that decreased as the flow increases.  The relationship between flow rate and tip volume is unique to each tipping bucket and described by its individual calibration curve and equation.   An example of a calibration curve for Pile 4 lysimeter C is provided in Figure 2-6 and calibration equations are provided in Table 2.9.  The calibration curve (Figure 2-6) describes the relationship between a tipping bucket’s tip rate (i.e., tips/second) versus its volume per time (as mL or L per tip).  Faster tip rates result in lower volumes per tip.  In general, the tipping bucket associated with the CHAPTER 2  37  basal lysimeter (i.e., Lysimeter D) captures approximately 2.2 – 2.5 L of water per tip at a maximum flow rate of 60 L per minute.  In contrast, tipping buckets associated with the sub-lysimeters (i.e., Lysimeter A, B and C) are smaller and capture approximately 25 - 32 mL per tip with a maximum flow rate of 16 mL per second.  The application of the calibration equations to tipping bucket data is provided in Appendix A. Precipitation data is collected at the experimental waste rock pile site using two Hobowise® rain gauges.  Rain gauges employ a tipping bucket design and therefore records the rain volume and rate with each ‘tip’.  Two gauges were installed to act as a backup in case of instrument failure in one or the other.  Although these contingencies were in place, precipitation data was not recorded by either rain gauge between August 2011 and May 2012 and therefore resulted in a ‘data gap’.  To ameliorate this data gap, precipitation was inferred by establishing two proxy relationships: 1) correlation with rain data from the nearby Yanacancha weather station; and 2) a nearly proportional relationship between rapid flow through Pile 4 and daily precipitation.  The details of these proxy relationships are discussed in Appendix A. 2.3.5. Aqueous Sampling  Water samples are collected from SWSSs and from the four lysimeters in both piles.  Samples are taken at varied frequencies, from weekly to tri-annually, which is dependent on the station, parameter, time of year and pile age (Bay, 2009).  Proper sampling and storage are applied to each sample taken from SWSSs and in-line lysimeter drainage (i.e., filtering, acid-preservation, cool temperatures (4oC)).  Details of the proper sampling technique are provided in Appendix A.   Samples are analyzed for total and dissolved metals by Envirolab Peru S.A.C using inductively coupled plasma mass-spectrometry (ICP-MS).  The elements analyzed for dissolved suites include: Al, Ag, As, Ba, Be, Bi, Bo, Ca, Cd, Ce, Co, Cr, Cs, Cu, Fe, Ga, Ge, Hg, Hf, K, La, Li, Lu, Mg, Mn, Mo, Na, Nb, Ni, P, Pb, Rb, Sb, Se, Si, Sr, Sn, Ta, Te, Th, Ti, Tl, U, V, W, Yb, Zn.  The majority of these elements are also analyzed for total concentrations.   Other parameters recorded from each station include non-metals (total alkalinity, Cl-, F-, SO42-), pH, temperature and nutrients (NO3-, NO2-, NH4+).  Refer to Corazao Gallegos (2007) for details of the analytical method and parameter detection limits.   CHAPTER 2  38  2.3.6. Tracer Test Tracers used in the Antamina tracer test were selected based on the following three criteria:   Natural concentrations of tracers were low to negligible;   Tracers were non-toxic to surrounding habitat;   Tracer analyses were cost-efficient and able to be analyzed using an inexpensive, rapid method to produce results quickly and with high-reproducibly.  From these criteria, one conservative and one non-conservative tracer were selected for this study.  Bromide was selected as the conservative tracer due to its extremely low background concentrations at Antamina (i.e., 0.2 mg∙L-1 – 0.8 mg∙L-1), high stability, low toxicity to flora/fauna, and the ease of analyses through spectrophotometry or ion chromatography.  No preservation methods were required for bromide due to its high stability.  The non-conservative tracer was uranine (C20H10O5Na), a sodium-fluorescein that produces a green fluorescence when dissolved in water.  The measurement of this tracer involved a simple fluorescent spectrophotometric analytical method and could be detected macroscopically at higher concentrations.  Antamina waste rock drainage contained negligible background fluorescent concentrations (i.e., 0 µg∙L-1 – 0.5 µg∙L-1) and preliminary batch and column experiments were conducted to test the possibility of uranine adsorption to waste rock.  Results from column experiments indicated uranine adsorbed to Class A material (found in Pile 5) and transport would be significantly retarded (Figure 2-8A).  However, uranine did not adsorb to Class B material (found in Pile 4) and transport would behave conservatively (Figure 2-8B).  Column tests were not conducted on Class C waste rock and it was assumed results would be similar to those of Class B.  Uranine was utilized in tracer application to both Pile 4 and 5, mostly as a macroscopic indicator of tracer breakthrough.     CHAPTER 2  39  2.3.6.1. Tracer event characteristics Tracer application on Pile 4 and 5 was conducted on Jan 23rd and 24th, 2010 (respectively).  The tracer solute contained ~ 3000 mg∙L-1 bromide and ~10 mg∙L-1 uranine, and was applied to the crown of each pile.  The tracer event corresponded to a 4-year return period event, which was based on intensity-duration-frequency curves developed for Antamina during a previous hydrological investigation (Golder, 1999).  Detailed characteristics of the tracer application for each pile are shown in Table 2.10.   The uniformity of tracer distribution on the crown of each pile was evaluated by calculating the Christiansen Uniformity Co-efficient (CUC; Christiansen, 1942).  CUC values were determined by measuring the volume of tracer solutions following tracer application from 91 evenly distributed cups across the pile crown and applying Equation 2-1.   𝐶𝑈𝐶 = 100 [1 − ((∑ |(𝑋𝑖 − ?̅?)|𝑛𝑖=1 )𝑛)]  (Eq. 2-1) A CUC value of 82 and 77 was calculated from the two tracer applications (Pile 4 and 5, respectively), indicating a good to very good uniform distribution.  Volumes used to calculate CUC values are found in Appendix A.   2.3.6.2. Tracer sampling and analyses Following tracer application, tracer samples were taken from lysimeter outflows and sampling frequency varied throughout the tracer test, from frequent (every 15 – 30 minutes) for the first 2 weeks of tracer application to infrequent (i.e., weekly) at the onset of the dry season in late April.   In general, a weekly sample frequency was maintained with the exception of the first 2 weeks of the 2010 wet season, where samples were taken daily.   All samples were dispensed into 15 ml opaque HDPE bottles and preserved at cool temperatures to decrease uranine degradation with light or heat. Samples obtained from soil water solution samplers, representing in situ pore waters, were sporadically analyzed for bromide concentrations.  Samples were taken infrequently (i.e., weekly to monthly) in the first year following tracer application.     CHAPTER 2  40  Bromide was analyzed using a spectrophotometric method developed by Presley (1971).  Specific details of this method are provided in Appendix A.   Uranine concentrations were measured using a Varian® Cary Eclipse fluorescence spectrophotometer (courtesy of the Laboratory of Molecular Biophysics, Spectroscopy and Kinetics Hub, University of British Columbia, Vancouver, BC, Canada).     2.3.6.3. Tracer data gap Tracer samples were not collected between early March 2010 and mid-April 2010; however, lithium is a measured solute included in the Antamina aqueous sampling program.  Therefore bromide concentrations were estimated by comparing lithium concentrations and measured bromide concentrations from each lysimeter outflow in the weeks preceding or succeeding the data gap in order to establish an adequate correlation.   A reasonably strong (i.e., r2 > 0.7) and positive correlation was observed between lithium and bromide concentrations during these periods, which suggested Li was an adequate proxy for the 1.5 month ‘tracer data gap’ period.     2.4. Observations 2.4.1. Precipitation Precipitation recorded by rain gauges at site for the first two years (i.e., June 2009 to August 2011) is shown in Figure 2-9.  Note that precipitation was not recoded by 2011 to 2012, due to instrument malfunctions.  Yearly cumulative precipitation amounts for each water year (i.e., June 1 through May 31) for 2009 – 2010 and 2010 – 2011 were calculated as 1272 mm and 1290 mm, respectively.  Yearly precipitation totals for 2011 – 2012, which were based on establishing a ‘rainfall proxy’ with waste rock pile outflow and nearby climate stations, were estimated as 1260 mm – 1300 mm.      2.4.2. Flow Daily flow measurements from Pile 4 and 5 is shown in Figure 2-10 as area-normalized outflow or specific discharge (as m3∙m∙2d-1 or mm∙d-1; [L∙T-1]), for the purpose of comparing flow rates between the basal lysimeter (Lysimeter D) and smaller lysimeters (Lysimeter A, B and C).   CHAPTER 2  41  Pile 4 responds quickly to rain events in the wet season, which is shown in Figure 2-10 by rapid, high outflow responses from all lysimeters daily between the months of November and March and suggestive of a preferential flow component.  Pile 5 also responds quickly to large rain events, as shown by large spikes in daily outflow (Figure 2-10), but a more muted or slower response to moderate or lower rainfall amounts in the wet season (in comparison to Pile 4).   Drainage recovered from Lysimeter C in Pile 5 shows a relatively constant outflow flux rate, with subtle changes throughout the wet season and shallow drain-down curves at the onset of the dry season.  Class A material, the finest-grained material of the three waste rock classes with the highest AEV, represents almost half of the material overlying Lysimeter C in Pile 5 and likely contributes to a slower (possibly more matrix-dominated) flow regime.    Table 2.11 shows the individual contribution of each lysimeter to total annual drainage and its average outflow flux.  As expected, the basal lysimeter for both piles contributes the bulk of the total annual drainage (i.e., ~ 95 % and ~96 % for Pile 4 and 5, respectively).  However, Lysimeter A and C in Pile 4 yields the highest yearly and wet season flux rates (i.e., approximately 3 mm∙d-1 and 7 mm∙d-1, respectively).  Dry season flux rates are substantially lower (i.e., 24x – 50x) than wet season values associated with Pile 4, with estimated lysimeter values of 0.12 mm∙d-1 and 0.18 mm∙d-1.  In Pile 5, yearly averaged flux rates are similar or lower than those from Pile 4 with values between 1 mm∙d-1 and 2.5 mm∙d-1.  Pile 5 wet season (only) flux rates are 6x to 15x higher than estimated flux rates in the dry season (i.e., 0.20 mm∙d-1 – 0.44 mm∙d-1).   2.4.3. Evaporation Evaporation was estimated by applying a water balance method, or measured precipitation minus recorded drainage outflow, since runoff was not observed from Pile 4 and 5 during the 2009 to 2012 experimental period.  Pile 4 evaporation estimates over three water years were: 56% (2009 – 2010), 33% (2010 – 2011) and ~31% (2011 – 2012).  Pile 5 showed similar evaporation estimates, and values were calculated as: 58% (2009 – 2010), 37% (2010 – 2011) and ~33% (2011 – 2012).  The significant decrease in evaporation observed between the first to second water year suggest both piles required 1 CHAPTER 2  42  year to sufficiently ‘wet-up’ or waste rock to return to residual saturation (after drying during construction).  Similarities in evaporation estimates for year 2 and 3 indicate changes to storage are minimal.      2.4.4. Tracer Results Tracer results are shown in Figure 2-11 and 2-12 for Pile 4 and 5, respectively.   2.4.4.1. Pile 4 Bromide concentrations for the first 2 years following tracer application on Pile 4 are shown in Figure 2-11A.  Initial breakthrough of bromide in Pile 4 was observed in all four lysimeters 4 to 6 hours following tracer application.  Lysimeter B produced the highest concentrations of 1300 mg∙L-1 (or C/Co = 0.44), 1.5 days following tracer release.  Peak concentrations from Lysimeter A (located at the center of the pile base) were approximately one-third lower than the concentration of Lysimeter B (i.e., 438 mg∙L-1 or C/Co = 0.15).  Based on initial breakthroughs and peak concentrations, estimated preferential flow velocities range between 7 m∙d-1 and 60 m∙d-1.  Trace bromide concentrations of 3 mg∙L-1  to 12 mg∙L-1  (i.e., C/Co < 0. 005) measured from Lysimeter C drainage, which is located outside of the pile crown and the tracer application area, suggest flow pathways have a minor to negligible horizontal component to a depth of 10 m and/or are not impacted by end-dumped slopes or pile geometry.  Tracer concentrations from Pile 4 decreased until the onset of the dry season approximately 3.5 months following the tracer application (i.e., May 2010), from which point tracer concentrations measured from Lysimeters A, B and D showed similar steady, low values (i.e., C/Co = 0.001).   The onset of the 2010 – 2011 wet season coincided with a decrease in bromide concentrations by approximately half an order of magnitude.  Preferential flow paths are more likely to be activated at the height of the wet season (Shipitalo and Edwards, 1996), therefore the significant dilution in bromide concentrations from previous dry season or low precipitation months supports the likelihood of rapid or preferential flow with limited interaction with matrix waters in Pile 4.  Bromide concentrations from all lysimeters continued to decrease through the second dry season and the onset of the third (2011-2012) wet season.   CHAPTER 2  43  Figure 2-11B presents bromide transport through the matrix pore space, as shown with bromide concentrations measured from SWSSs that collect samples of matrix porewaters.  Tracer breakthrough curves from SWSSs were delayed and showed lower peak concentrations, in comparison to lysimeter-recovered samples.  There did not appear to be a clear relationship between the position of the SWSS and the timing/concentration of peak concentrations.  Specifically, maximum concentrations occurred at the same time in multiple locations along the second instrumentation line, with P4-L2A and P4-L2B reflecting SWSS located along instrumentation line 2 at the base of the line (L2A) and 2 m higher in height (L2B).  Also, the dissimilarity in maximum concentrations suggested flow is spatially variable within relatively short distances.  This dissimilarity is likely attributed to material heterogeneities observed at a local- or small- scale that can significantly impact flow rates and therefore water-rock interaction times. Approximately 72% of the total bromide applied was recovered after 2 years, with the remainder following extremely slow flow paths or associated with water held in immobile domains.  Eight percent of the tracer mass was recovered in the first three days and is assumed to represent the range of preferential flow paths.  The remaining 64% is then assumed to reflect matrix flow paths, with a median velocity of 0.36 m∙d-1.  Although preferential flow paths are a dominant characteristic of Pile 4 flow regimes, the majority of infiltration follows matrix flow paths.   Uranine breakthrough curves from Pile 4 lysimeters during the first year following tracer application are shown in Figure 2-11C and D.  Uranine concentrations (as C/Co) from the first 2 weeks were similar to those from the conservative (bromide) tracer analyses.  Flury and Wai (2003) indicate moderate to long water-rock interaction times results in an increased likelihood for uranine adsorption to porous materials.  As such, the similarity in C/Co values from bromide and uranine indicates a small proportion of mixing/dilution occurred and supports the likelihood for short water-rock interaction times and/or flow along preferential pathways or fast macropore paths in Pile 4. 2.4.4.2. Pile 5 Tracer breakthrough from Pile 5 lysimeters occurred sporadically over a period of weeks (Figure 2-12A), unlike Pile 4 which showed relatively uniform tracer breakthrough within hours of tracer application.  Initial bromide breakthrough was measured in Lysimeter D at near detection limit concentrations (i.e., 0.01 CHAPTER 2  44  mg∙L-1 or C/Co = 3.5 x 10-6) at four hours following tracer application.  Unlike Pile 4, which showed continual tracer breakthrough following initial measurements, bromide concentrations were measured sporadically and at similarly low or near detection limit values for two weeks.  After this period, bromide was detected continuously from Lysimeter D at concentrations greater than 0.1% of the applied concentration (i.e., >3 mg/L).   The first occurrence of bromide at Lysimeter B occurred 48 hours following release (i.e., 3.2 mg∙L-1 or C/Co ~ 10-3); however bromide was measured sporadically over a similar period as Lysimeter D (i.e., 2 weeks).  The highest or maximum bromide concentration measured from Pile 5 was sampled from Lysimeter B nine days following the tracer application (i.e., 289 mg∙L-1 or C/Co = 0.10) and during the period of sporadic tracer release.  Notably, this concentration was approximately 4x lower than peak values measured from Pile 4 outflow.   Bromide breakthrough from Lysimeter A required twice the time and was 10 times lower than that in Lysimeter B (i.e., 4 days; 0.30 mg∙L-1 or C/Co ~ 10-4).  Unlike Lysimeter B and D, bromide was released continuously following its first arrival.   Maximum concentrations of 238 mg∙L-1 or C/Co = 0.08 were measured from Lysimeter A outflow at the same time as peak concentrations from Lysimeter B.  These results suggest the presence of preferential flow paths is spatially variable, with peak estimated velocities of 3 m∙d-1 to 5 m∙d-1. Bromide was measured sporadically in Lysimeter C outflow for approximately one month after application, before systematically releasing low to near detection limit bromide concentrations (i.e., C/Co < 10-4).  The low bromide concentrations from Lysimeter C suggested flow had a minor to negligible horizontal component and/or not significantly impacted by end-dumped slopes or pile geometries, which was comparable to Pile 4.  Approximately 67% of the total bromide applied was recovered after 2 years, with the remainder following extremely slow flow paths or associated with water held in immobile domains.  Significantly less than 0.01% percent of the tracer mass was recovered in the first three days, which indicates preferential flow CHAPTER 2  45  paths are minimally present in Pile 5.  Therefore the majority of flow in this pile follows matrix flow paths, with a median velocity of 0.12 m∙d-1.   Comparison of bromide concentrations taken from matrix pore space (from SWSSs) versus lysimeter outflow is shown in Figure 2-12B.   Similar to results shown in Pile 4 (Figure 2-11B), bromide concentrations measured from SWSS samples showed spatial variability in the timing and magnitude of peak concentrations.  For example, the SWSS located closest to the surface with the shortest travel distance (P5-L2D) shown in Figure 2-11B presented a peak concentration that was delayed by 2 months relative to peak concentrations measured from SWSSs located at 4 m and 6 m from the surface (i.e., P5-L2B and –L2C, respectively).  Therefore Pile 5 flow is highly variable at relatively short distances within Pile 5 and may be attributed to material heterogeneities at a similar spatial scale, which is comparable to the observation in Pile 4 (Section 2.4.4.1).   In general, measured uranine concentrations were near or below detection limit values (i.e., ~0.5 µg∙L-1) for the majority of the first year following tracer application.  The one exception was a single uranine peak measured from Lysimeter A and B outflows at 9 days following tracer application (i.e., ~0.1 mg∙L-1  or C/Co = 0.01), which was coincident with the bromide peak maximum recorded in in the same two lysimeters.  Uranine tracer analysis was discontinued following the first year due to poor recovery and likely retardation due to adsorption to Class A materials.      2.4.5. Aqueous Chemistry Measured concentrations from lysimeter outflows are provided in Figure 2-13, 2-15 and 2-16 (Pile 4) and Figure 2-14, 2-17 and 2-18 (Pile 5).   2.4.5.1. pH and anion concentrations Measured pH, sulphate, alkalinity, and nitrate concentrations from Pile 4 and 5 are shown in Figure 2-13 and 2-14 (respectively).  Measured pH values from both piles are neutral to alkaline (i.e., pH 7.0 to pH 8.5) and show a slight increasing trend with each year.  Dissolved sulphate concentrations from Pile 4 fluctuate between 400 mg∙L-1 and 2000 mg∙L-1, with minimum and maximum values associated with the height of the wet and dry season (respectively).  Pile 5 sulphate values (Figure 2-14B) show less CHAPTER 2  46  seasonality and a narrower range, with most values between 800 mg/L and 2000 mg/L.  The basal lysimeters (i.e., Lysimeter D) present the most dramatic shift in sulphate concentration throughout a single year with the highest values at the onset of the wet season.  Pile 4 and 5 present a steady decrease in maximum (dry season) sulphate concentration, with average concentration decreases of 10 – 25 % over the 3 water years.   Alkalinity values from both piles are similar, with dry season highs of 50 – 60 mg∙L-1 HCO3- and wet season lows of approximately 30 mg∙L-1.  Nitrate values decrease by 0.5 – 2 orders of magnitude each water year from Pile 4 and 5, respectively, which suggests solutes associated with blasting materials may be used as an internal or resident tracer.   2.4.5.2. Metal concentrations 2.4.5.2.1 Pile 4 Figures 2-15 and 2-16 show seasonal variations to major dissolved cations for Pile 4 lysimeter outflow.  Concentrations of Zn, Se, Co and Cd present decreasing trends and some seasonal fluctuations, with wet season low values and dry season high values.  Several metal(loid)s show a weak seasonality (i.e., Mo, Ni), relatively stable values (Pb: 0.01 mg∙L-1), or stable to increasing values (Cu).  Three exceptions to a seasonal and/or decreasing trend are noted in Pile 4, namely:   As: As concentrations show an opposite seasonal trend, with higher concentrations at the height of wet season relative to dry season values.  Additionally, measured concentrations range three orders of magnitude (i.e., 0.001 mg∙L-1 and 1 mg∙L-1) and differ between lysimeter outflows: o Lysimeter A: equal to or less than 0.1 mg∙L-1, with a decreasing trend; o Lysimeter B: similar to Lysimeter A, with a decreasing trend, but at an order of magnitude lower concentrations (i.e., ≤0.01 mg∙L-1); o Lysimeter C and D: Present an increasing trend, with maximum As concentrations (during the dry season) increasing by up to half an order of magnitude each year.   CHAPTER 2  47   Fe: Fe concentrations range three orders of magnitude (i.e., 0.001 mg∙L-1  – 1 mg∙L-1) and, in general, resemble a similar pattern as As with increasing values in the wet season and dry season lows.    Sb: Similar to As, presents an opposite seasonal trend with wet season highs and dry season low values and lysimeter-specific concentrations; o Lysimeter A and B: Shows a weak seasonality; with decreasing values  (≤0.1 mg∙L-1) yearly and presented a decreasing trend with time; o Lysimeter C and D: Sb concentrations show a stronger seasonality, relative to Lysimeter A and B chemistries, with similar concentrations (i.e., 0.2 mg∙L-1 and 0.1 mg∙L-1, respectively).    The strong ‘wet season low and dry season high’ seasonal trends observed in Pile 4 implies flow regimes significantly influence measured drainage chemistries.  In particular, tracer results suggest the majority of infiltration in Pile 4 follows matrix flow paths and low wet season concentrations imply preferential flow paths (activated in the wet season) may dilute matrix flow chemistries upon mixing in the lysimeter and prior to sample collection.  An opposite seasonality is observed for As, Fe and Sb, which is suggestive of a geochemical/solubility control or secondary minerals precipitation along preferential flow paths during drier periods that are subsequently flushed in the wet season.    2.4.5.2.2 Pile 5 In general, solute concentrations from Pile 5 demonstrate a weaker and opposite seasonality relative to Pile 4 (i.e., higher concentrations in the wet season).  This opposite trend is noted for concentrations of As, Fe, Mo, Sb and Zn.  Most solutes, however, display a steady or decreasing trend in measured concentrations (i.e., alkalinity, Cd, Co, Pb and Se).  Copper concentrations are also observed to be relatively stable with respect to season, with increasing values over time.   Some metal(loid)s exhibit lysimeter-specific trends, namely:  As: Magnitude and temporal trends in As concentrations differ between lysimeters; CHAPTER 2  48  o Lysimeter B and D: As concentrations increase with time, with Lysimeter B values approximately 10x higher than those from Lysimeter D ( i.e., 0.1 mg∙L-1  versus 0.011 mg∙L-1).   o Lysimeter A and C: As concentrations are similar (i.e., ~0.01 mg∙L-1) and decreased with time.    Mo: Measured Mo concentrations from all four lysimeters increase with time, specifically: o Lysimeter A and D: presents the highest Mo concentrations (i.e., up to 1 mg∙L-1)  o Lysimeter B: Mo concentrations are approximately 10x lower (i.e., 0.1 mg∙L-1) than those from Lysimeter A and D o Lysimeter C: Mo concentrations are less than half those from Lysimeter B (i.e., 0.04 mg∙L-1) Tracer results indicate a minor to insignificant amount of infiltration follows preferential flow paths and the majority of flow is described as matrix flow.  Therefore the general observation of wet season high and dry season low concentrations from Pile 5, relative to Pile 4, support a high matrix flow proportion.  Matrix flow paths are more likely to have longer water-rock interaction times and drainage chemistries are controlled by geochemical reactions occurring in the matrix pore spaces.  Increasing As and Mo concentrations also implied pH-controls and mineral solubilities may have played a role in observed drainage chemistries.  2.5. Conceptual Models for the Large-scale 2.5.1. Pile 4 Pile 4 is a 10 m high constructed pile covering an areal footprint of 36 m x 36 m and contains 19,000 tonnes of waste rock.  This pile was built from five end-dumped, ~2,000 – 5,000 tonne discharges of alternating Class C and B waste rock at a 37o slope.  In total, approximately three-quarters of waste rock in Pile 4 is classified as Class C waste rock and the remainder as Class B.    Approximately two-thirds of yearly precipitation infiltrates Pile 4, with minimal changes to storage after one year of wet-up.  Outflow from the four lysimeters at the pile base responds quickly during large rain CHAPTER 2  49  events, as shown by the similarity between peak daily rain events (i.e., 10 – 30 mm/d; Figure 2-9) and similar maximum area-normalized outflow fluxes (Figure 2-10).   Quick response to large rain events supports the presence of a preferential flow component, in addition to matrix flow.  Preferential flow in Pile 4 is also evidenced by rapid tracer breakthrough within hours of tracer application and strongly seasonal drainage chemistries of wet season low and dry season high concentrations.  The recovery of low to near detection limit tracer concentrations from Lysimeter C, which was located below the pile batter and outside of the tracer application area (i.e., crown), implies flow paths are not likely impacted by internal pile geometries of end-dumped slopes.  Seasonal variability in solute concentrations (i.e., sulphate, alkalinity, Cd, Co, Mo, Ni, Se, and Zn) suggest drainage chemistries are significantly influenced by the bimodal climate, with lower values in the wet season and higher values in the dry season.   This seasonality suggests reactions involved in the release of these elements occur in the matrix pore space, which flows year round, and are diluted during wet season periods or when preferential flow paths are more likely to be activated.  The dilution may occur due to mixing of preferential and matrix flow in pile lysimeters prior to sample collection.   For those elements with an opposite seasonal trend (i.e., wet season highs; As, Sb), one possibility is these reactions occur along preferential flow paths.   Therefore increased concentrations during the wet season may be conceptualized as the dissolution or rapid flushing of secondary precipitates (e.g., Fe-oxy(hydr)oxides) containing sorbed  metals that precipitated during drier months.  A similar process may occur with other elements, but is not obvious from this high level review of Pile 4 drainage chemistries.  Seasonal changes to flow rates did not appear to influence Cu and Pb concentrations and values remained relatively stable year round.   The degree of preferential flow versus matrix flow in Pile 4 cannot be resolved from this high-level review of flow and solute transport; however, it is evident that both were present.  The observation that nitrate concentrations decrease yearly suggests it is likely reflective of a slow flushing of blasting residues on waste rock surfaces.  As well, nitrate concentrations typically plateau during dry season months and decrease significantly with the onset of the wet season.  This pattern suggests nitrate was released from CHAPTER 2  50  the matrix pore space of Pile 4 and the degree of matrix flow may be estimated by using nitrate (or other blasting solute; e.g., Cl, NH4+) as an internal tracer.   2.5.2. Pile 5 Pile 5 is constructed of an identical geometry as Pile 4, with the exception it contains 6 end-dumped discharges of approximately 2,000 – 4,000 tonnes of Class A or C waste rock.  A little more than half of the waste rock in Pile 5 is classified as Class A (i.e., 51%) and the remainder as Class C.  Two-thirds of precipitation infiltrates Pile 5, with minimal changes to storage after one year of wet-up.  Outflow from the four lysimeters at the pile base responds quickly during large rain events, as shown by large spikes in daily outflow between the two mixed piles.  However Pile 5 shows a more muted or slower overall response to daily rainfall in the wet season.  This was also shown by comparing average specific discharges in the wet season to those from the dry season (Table 2.11).  Specifically, these values differ by 5x – 16x in Pile 5, which is much lower than the 24x – 50x difference between wet and dry season fluxes from Pile 4.   Quick outflow responses in Pile 5 following large rain events are suggestive of a preferential flow component (in addition to matrix flow).  However results from the Pile 5 tracer test indicate preferential flow paths contribute to a small to insignificant proportion of this pile’s flow regimes relative to those from Pile 4.  Specifically, although bromide was measured within hours of the Pile 5 tracer application, concentrations were near detection limits and measured sporadically in lysimeter outflow for the first 2 weeks.  As well, bromide recovery three days following tracer application was four orders of magnitude lower than recovery for Pile 4 (i.e., 0.0004% versus 0.09%).  The majority of Pile 5 infiltration follows matrix flow paths, which is estimated to have a median velocity of 0.12 m∙d-1.  Similar to Pile 4, the recovery of low to near detection limit tracer concentrations from Lysimeter C, which was located below the pile batter and outside of the tracer application area (i.e., crown), implies flow paths are not likely impacted by internal pile geometries of end-dumped slopes.  With the exception of sulphate, most solute concentrations from Pile 5 demonstrate a weaker and opposite seasonal trend (i.e., alkalinity, As, Fe, Mo, Sb and Zn), relative to Pile 4, or fairly stable values CHAPTER 2  51  that decreased with time (i.e., Cd, Co, Pb and Se).  The absence of a strong seasonality for most solutes supports tracer test breakthrough results, such that the majority of water infiltrating Pile 5 follows matrix flow paths.  Nitrate concentrations decrease yearly and, like Pile 4, this solute may be applicable as an internal tracer to further evaluate matrix flow parameters/characteristics.  2.6. Summary This thesis research is part of a large, multi-year study at Antamina that included the construction of five mesoscale experimental waste rock piles (i.e., 10 m (h); 19,000 – 25,000 tonnes), in addition to 45 field barrels (i.e., 1 m (h); ~300 kg).  This focus of this research is on two (of 5) experimental piles, 15 (of 45) field barrels, two 1 m (high) column experiments, with smaller scale experiments discussed in Chapter 3.  All waste rock used in the multi-year and multi-scale study is characterized by a waste rock classification developed by Antamina.   This scheme classifies materials as Class A, B or C based primarily on its sulphide, Zn and As concentrations.  A mixture of two waste rock types was used in the construction of the two experimental piles (i.e., Pile 4 and 5) included in this research; Class B and C (Pile 4) and Class A and C (Pile 5).  Comparison of the physical characteristics and mineralogy from the three waste rock classes indicated there are distinct physical and geochemical differences between Class A, B and C, which have the potential to impact flow and solute transport through stockpiles of each waste rock material.  Both piles show indications of a rapid flow component; however results suggest preferential flow paths contribute to a larger portion of the flow regime in Pile 4, relative to Pile 5.  Results suggest that matrix flow and a minor preferential flow component characterizes the dominant flow characteristics associated with Pile 5.      CHAPTER 2  52  2.7. Tables Table 2.1 Acid-base Accounting and Bulk Geochemistry of Major Antamina Waste Rock Types. Waste Rock Type  Acid-Base Accounting (ABA) Solid-Phase Analyses Paste  pH Sulphide-S (%) NNP  (kg CaCO3/t) NPR (-) As  (ug/g) Zn  (ug/g) Skarn† – Brown garnet  (n =11) Mean 6.9 7.0 -176 0.20 104 4,909 Range 4.7 – 8.9 0.43 – 24 -732 - 128 0.0 – 0.4 6.0 – 1,730 32 – 28,500 Skarn† – Green garnet  (n =17) Mean 8.5 0.70 105 5.8 77 3,967 Range 6.4 – 11 0.03 – 3.2 -81 - 315 0.20 – 180 2.0 – 856 2.0 – 60,100 Marble  (n = 47) Mean 8.6 0.53 692 43 19 4,648 Range 7.8 – 9.6 0.03 – 4.0 149 – 904 2.5 - 354 1.0 – 190 2.0 – 136,600 Limestone  (n = 21) Mean 8.6 0.46 666 47 13 1,264 Range 8.1 – 9.3 0.09 – 1.8 11 - 947 2.4 - 332 1.0 – 55 32 – 13,500 Intrusive  (n = 83) Mean 8.2 1.4 -9 3.3 89 393 Range 6.5 – 9.5 0.00 – 12 -352 - 335 0.0 - 60 1.0 – 846 4.0 – 9,120 From KC-SVS (1998). † The mineralized skarn is dominated by grandite garnet, which grades from brown to green as skarn assemblages are closer to the host limestone (Petersen, 1965).   CHAPTER 2  53  Table 2.2 Antamina Waste Rock Classification. Class Reactivity Main Rock Types Zn  (ppm) As (ppm) Sulphides (%) A Reactive Skarn, Intrusive† - - - A Reactive Marble, Limestone, Hornfels > 1,500 > 400 > 3.0 B Slightly Reactive 700 – 1,500 N/A < 3.0 C Non-reactive < 700 < 400 < 3.0 Notes: Updated in 2010. †All skarn and instrusive material classified as Class A.CHAPTER 2  54  Table 2.3 Waste Rock Classification and Composition of Pile 4. Discharge Reference Pit Location* Date Excavated Waste Rock Class Weight (tonnes) 1 5-SP-4643-11-02 13-Nov-08 C†,‡ 2,545 1 4-NP-4268-03-01 14-Nov-08 C†,‡ 2,872 3 5-SP-4643-10-01 06-Dec-08 C† 4,215 5 5-SP-4598-01-01 14-Jan-08 C¶ 4,659   Class C Total 14,291 2 4-NP-4268-12-02 26-Nov-08 B† 2,214 4 4-NP-4268-27-05 09-Jan-08 B§ 2,660   Class B Total 4,474   Pile 4 Total 18,765 * Each blast corresponds to a polygon within the main pit; as referenced here.   † Grey – café hornfels ‡ No data from this discharge ¶ Café hornfels § Black/white marble    CHAPTER 2  55  Table 2.4 Waste Rock Classification and Composition of Pile 5. Discharge Reference Pit Location* Date Excavated Waste Rock Class Weight (tonnes) 1 5-SP-4643-11-02 13-Nov-08 C†, 1,909 1 4-NP-4268-03-01 14-Nov-08 C†,‡ 2,434 4 5-SP-4613-01-01 14-Dec-08 C¶ 2,319 5 5-SP-4598-01-01 14-Jan-09 C†,# 2,601   Class C Total 9,263 2 2-NP-4028-08-01 01-Dec-08 A§ 2,606 3 2-NP-4028-10-01 12-Dec-08 A§ 4,098 6 2-NP-4013-06-01 04-Feb-09 A §,‡‡ 3,033   Class A Total 9,737   Pile 5 Total 19,000 * Each blast corresponds to a polygon within the main pit; as referenced here.   † Grey – café hornfels ‡ No data from this discharge ¶ Grey hornfels # Marble § Intrusive – quartz monzonite ‡‡ Endoskarn CHAPTER 2  56  Table 2.5 Bulk Geochemical Characteristics of Pile 4 Waste Rock Material. Discharge Pit Location Date Material Ag (ppm) As (ppm) Bi (ppm) Co (ppm) Cu  (%) Fe  (%) Mo  (%) Pb  (%) Zn  (%) 1 5-SP-4643-11-02 13-Nov-08 C†,‡ 1 102 2 6 0.01 1.65 0.01 0.01 0.02 1 4-NP-4268-03-01 14-Nov-08 C†,‡ n.d n.d n.d n.d n.d n.d n.d n.d n.d 3 5-SP-4643-10-01 06-Dec-08 C† 1 45 1 5 0.01 1.35 0.001 0.02 0.01 5 5-SP-4598-01-01 14-Jan-08 C¶ 1 122 7 4 0.01 1.15 0.01 0.01 0.01 2 4-NP-4268-12-02 26-Nov-08 B† 1 36 22 8 0.01 2.12 0.01 0.03 0.05 4 4-NP-4268-27-05 09-Jan-08 B§ 2 16 12 3 0.01 0.61 0.001 0.04 0.08 n.d = no data † Grey – café hornfels ‡ No data from this discharge § Black/white marble ¶ Café hornfels    CHAPTER 2  57  Table 2.6 Bulk Geochemical Characteristics of Pile 5 Waste Rock Material. Discharge Pit Location Date Material Ag (ppm) As (ppm) Bi (ppm) Co (ppm) Cu  (%) Fe  (%) Mo  (%) Pb  (%) Zn  (%) 1 5-SP-4643-11-02 13-Nov-08 C†, 1 102 2 6 0.01 1.65 0.01 0.01 0.02 1 4-NP-4268-03-01 14-Nov-08 C†,‡ n.d n.d n.d n.d n.d n.d n.d n.d n.d 4 5-SP-4613-01-01 14-Dec-08 C¶ 1 130 10 7 0.01 1.57 0.001 0.02 0.01 5 5-SP-4598-01-01 14-Jan-09 C†,# 1 122 7 4 0.01 1.15 0.01 0.01 0.01 2 2-NP-4028-08-01 01-Dec-08 A§ 1 74 6 5 0.16 1.58 0.013 0.01 0.01 3 2-NP-4028-10-01 12-Dec-08 A§ 3 127 6 7 0.30 1.95 0.013 0.01 0.01 6 2-NP-4013-06-01 04-Feb-09 A §,‡‡ 1 73 6 5 0.14 1.78 0.013 0.01 0.01 n.d = no data † Grey – café hornfels ‡ No data from this discharge § Intrusive – quartz monzonite ¶ Grey hornfels # Marble ‡‡ Endoskarn    CHAPTER 2  58  Table 2.7 X-ray Fluorescence (XRF) Whole Rock Geochemistry of Antamina Waste Rock. Major Oxide or Element Units Class A  Pile 5 Class A  Pile 2 Class B  Pile 1 Class C Pile 4 and 5 n = 1 n = 5 n = 1 n = 3 Al2O3 % 7.31 7.76 8.39 5.39 CaO % 2.30 2.45 31.1 35.9 CrO3 % <0.01 <0.01 <0.01 <0.01 Fe2O3 % 3.88 4.43 5.18 2.34 K2O % 4.54 4.78 1.90 1.09 MnO % 0.05 0.07 0.18 0.06 MgO % 0.58 0.72 1.98 2.38 Na2O % 0.53 0.68 0.18 0.23 P2O5 % 0.06 0.09 0.10 0.09 SiO2 % 78.6 75.8 35.9 18.5 TiO2 % 0.25 0.27 0.35 0.26 Ag ppm <10 13 <10 <10 As ppm 66.0 97.4 138 253 Co ppm 9.40 10.0 9.20 7.70 Cr ppm 61.5 59.4 119 99.6 Cu ppm 1765 4309 1369 166 Mn ppm 416 522 1534 680 Mo ppm 289 902 182 <20 Ni ppm 6.00 8.48 13.6 11.2 Pb ppm 28.2 123 1019 298 Zn ppm 189 325 1096 337 C % 0.08 0.09 3.37 2.02 S % 0.87 1.19 0.75 0.13    CHAPTER 2  59  Table 2.8 Mineralogical Composition of Pile Waste Rock using X-ray Diffraction (XRD) Analysis.  Class A Class B Class C Waste Rock Pile Pile 5 Pile 4 Pile 4/5 Pile 5 Discharge 2 6 2 4 4 1 5 4 Albite low NaAlSi3O8 1.0 1.2 2.4 1.1 0.0 1.4 5.7 11 Andradite Ca3Fe2(SiO4)3 7.9 1.5 - - - - - - Anorthite CaAl2Si2O8 - - 9.3 0.2 0.7 7.7 - - Biotite 1M K(Mg,Fe)3AlSi3O10(F,OH)2 2.3 3.1 - - - -- - - Calcite CaCO3 - - 61 87 92 66 82 60 Chalcopyrite CuFeS2 0.64 0.74 - - - - - - Diopside CaMgSi2O6 - - 1.4 3.2 1.1 1.0 0.66 - Kaolinite Al2Si2O5(OH)4 - - - - - - 0.39 - Muscovite 2M1 KAl2(AlSi3O10)(F,OH)2 - - 1.2 - 0.5 1.6 - 3.8 Oligoclase (Ca,Na)(Al,Si)4O8† 7.0 8.4 - - - - - - K-feldspar‡ KAlSi3O8 34 44.3 4.8 1.7 0.91 2.4 - 4.7 Phlogopite 1M KMg3AlSi3O10(F,OH)2 - - 4.7 0.18 0.78 3.7 4.4 4.6 Pyrite FeS2 2.7 0.72 0.31 1.2 - - 0.46 0.15 Pyrrhotite 4M Fe1-xS§ - - 1.8 0.91 0.83 1.3 0.85 2.1 Quartz SiO2 46 40 8.7 1.2 0.43 9.7 5.0 10 Total  101.5 99.7 95.6 96.7 97.3 94.8 99.5 96.7 “-” = not detected  † where Ca/(Ca + Na) or % anorthite = 10%–30% ‡ Orthoclase § x = 0 – 0.2  CHAPTER 2  60  Table 2.9 Pile 4 and 5 Calibration Equations and Average Tipping Bucket Volume.  Calibration equation (L/sec) Volume per tip (mL) Pile 4   Lysimeter A 0.0261946 x-0.9831 27.33 ± 1.63  Lysimeter B 0.0274856 x-1.033 25.07 ± 2.19  Lysimeter C 0.0290336 x-1.0413 25.72 ± 2.11  Lysimeter D 2.8790733 x-1.0498 2523 ± 264 Pile 5   Lysimeter A 0.0367928 x-1.0515 31.67 ± 2.64 Lysimeter B 0.0356665 x-1.0602 29.63 ± 3.41 Lysimeter C 0.0282855 x-1.0009 28.30 ± 2.51 Lysimeter D 2.2768929 x-1.0069 2261 ± 402 Note: x = tip rate or sec/tip    CHAPTER 2  61  Table 2.10 Pile 4 and 5 Tracer Test Characteristics. Pile Description Tracer footprint (m2) Volapplied (L) Duration (hr) [Bromide] (mg/L) [Uranine] (mg/L) CUC† (--) 4 B and C 336 m2 8,829 4.75 2,900 9.14 82 5 A and C 311 m2 8,354 5.17 3,046 11.2 77 † Christiansen Uniformity Co-efficient; Note: tracer was applied using 45 – 50 psi and flow rates of ~ 1,600 L/hr.   CHAPTER 2  62  Table 2.11 Lysimeter Contribution (in m3) to Annual Drainage in Pile 4 and 5.  Lys A Lys B Lys C Lys D Total Pile 4      2009 – 2010 21.5 11.6 21.2 665 719 2010 – 2011 20.6 8.78 11.7 1082 1123 2011 – 2012  20.1 11.7 21.2 1086 1139 Area- normalized outflow flux (mm/d)     Yearly ‡ 3.6 1.8 2.8 1.9  Height of wet season # 7.6 3.7 6.5 3.6  Height of dry season ¶ 0.18 0.12 0.13 0.15   Pile 5      2009 – 2010 18.8 17.2 9.15 647 693 2010 – 2011 14.1 14.1 4.56 1022 1055 2011 – 2012  7.61 12.3 2.45 1088 1110 Area- normalized outflow flux (mm/d)     Yearly ‡ 2.3 2.5 1.0 1.9  Height of wet season # 3.9  4.5 1.5 3.2  Height of dry season ¶ 0.44 0.31 0.30 0.20  † Evaporation calculated by P – Q; total annual precipitation calculated to be 1272 mm (2009 – 2010), 1290 mm (2010 – 2011), and 1260 – 1300 mm (2011 – 2012).  Latter values estimated from precipitation proxies (discussed in Section 2.1.1.1).  ‡ Average calculated from yearly cumulative area normalized flux rate (m3/m2). # Wet season averages calculated from January monthly outflows in 2010, 2011 and 2012.  ¶  Dry season averages calculated from July monthly outflows in 2010, 2011 and 2012.   CHAPTER 2  63  2.8. Figures           Figure 2-1. Schematic of experimental waste rock piles, plan view (A) and side view (B). A B 4 m 4375 m Pile 4 Pile 5 36 m 10 m 4385 m CHAPTER 2  64               Figure 2-2. Trenching of Pile 4 instrumentation line 4. A 1-2 m deep trench along the discharge centerline is observed in (A) along with a side view (B).    A B Pile Crown Pile Base CHAPTER 2  65              Figure 2-3. Instrumentation line placement (A), sensor installation (B) and covering (C).   A B C Soil water solution sampler Decagon 5TE Gas line TDR probe  Careful placement of larger boulders to protect sensors CHAPTER 2  66    Figure 2-4. Particle size distribution of waste rock used in Pile 4 and 5.   01020304050607080901000.010.1110100100010000% Passing Particle Size (mm) Pile 4/5 - Class CPile 4/5 - Class CPile 4/5 - Class CPile 4/5 - Class CPile 5 - Class CPile 4 - Class BPile 4 - Class BPile 5 - Class APile 1 - Class B avgPile 2 - Class A avgCu (D60/D10) P4/5-Class C:  13.2 (avg) P5-Class C:      9.4 P4-Class B:      25.5 (avg) P5-Class A:      46.5 P1-Class B avg: 21.3 P2-Class A avg: 343 CHAPTER 2  67   Figure 2-5. Soil water characteristic curves (SWCCs) from Pile 4 and 5 waste rock.  Note: adapted from Speidel (2011).   0.000.050.100.150.200.250.300.350.01 0.1 1 10 100 1000 10000 100000 1000000Volumetric Water Content  (m3∙m-3) Soil Suction (kPa) Pile 4/5 - Class CPile 4/5 - Class CPile 4/5 - Class CPile 4/5 - Class CPile 5 - Class CPile 4 - Class BPile 4 - Class BPile 5 - Class APile 1 - Class B avgPile 2 - Class A avgCHAPTER 2  68   Figure 2-6. Calibration curve from Pile 4 lysimeter C.   y = 29.551x-1.029 R² = 0.9927 05101520250 20 40 60 80 100 120Flow Rate (ml/sec) Tip Rate (sec/tip) CHAPTER 2  69     Figure 2-7. Comparison of weekly precipitation from Punto B and Yanacancha rain gauges between Sept 14, 2009 and August 15th, 2010.   y = 0.8566x + 0.42 R² = 0.8312 0204060801000 20 40 60 80 100Yanacanch Precipitation (mm/week) Punto B Precipitation (mm/week) CHAPTER 2  70          Figure 2-8. Comparison of uranine adsorption in Class A and B waste rock (A and B, respectively).   0.000.200.400.600.801.000.0 1.0 2.0 3.0 4.0C(t)/Co A BromideUranine0.000.200.400.600.801.000.0 1.0 2.0 3.0C(t)/Co Pore volume B CHAPTER 2  71    Figure 2-9. Daily precipitation (mm) recorded from research site rain gauge.   010203040Jun-09 Oct-09 Feb-10 Jun-10 Oct-10 Feb-11 Jun-11Daily Precipitation (mm) CHAPTER 2  72    Figure 2-10. Pile 4 (A) and Pile 5 (B) area-normalized daily outflow recorded from pile lysimeters. 00.010.020.030.04Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Area normalized outflow flux (m3·m-2·d-1) A: Pile 4 Lys ALys BLys CLys D00.010.020.030.04Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Area normalized outflow flux (m3·m-2·d-1) B: Pile 5 Lys ALys BLys CLys DCHAPTER 2  73   Figure 2-11. Pile 4 tracer breakthrough curves of bromide (A and B) and uranine (C and D). 00.10.20.30.40.5Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12C/Co A: Pile 4 - Bromide  LysALysBLysCLysDB inset 00.10.20.30.40.5Jan Feb Mar Apr May JunC/Co B: Pile 4 - Bromide LysALysBLysCLysDP4-L1DP4-L2AP4-L2BP4-L2CP4-L2DP4-L4E00.10.20.30.4Jan-10 Mar-10 May-10 Jul-10 Sep-10 Nov-10 Jan-11C/Co C: Pile 4 - Uranine LysALysBLysCLysDD inset 00.10.20.30.422-Jan 27-Jan 1-Feb 6-Feb 11-Feb 16-Feb 21-FebC/Co D: Pile 4 - Uranine LysALysBLysCLysDCHAPTER 2  74   Figure 2-12. Pile 5 tracer breakthrough curves of bromide (A and B) and uranine (C). 0.000.020.040.060.080.10Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12C/Co A: Pile 5 - Bromide LysALysBLysCLysD0.000.020.040.060.080.10Jan Feb Mar Apr May JunC/Co B: Pile 5 - Bromide LysALysBLysCLysDP5-L2DP5-L2CP5-L2BP5-L2A00.0030.0060.0090.0120.015Jan-10 Mar-10 May-10 Jul-10 Sep-10 Oct-10 Dec-10C/Co C: Pile 5 - Uranine LysALysBLysCLysDB inset CHAPTER 2  75   Figure 2-13. Comparison of drainage chemistries from Pile 4 lysimeter outflows. 6789Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12pH A: Pile 4 pH  05001000150020002500Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Sulphate (mg/L) B: Pile 4 Sulphate  Lys ALys BLys CLys D020406080Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Alkalinity (mg/L HCO3- ) C: Pile 4 Alkalinity 0.010.11101001000Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Nitrate (mg/L) D: Pile 4 Nitrate Lys ALys BLys CLys DCHAPTER 2  76   Figure 2-14. Comparison of drainage chemistries from Pile 5 lysimeter outflows.  6789Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12pH A: Pile 5 pH 0500100015002000250030003500Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Sulphate (mg/L) B: Pile 5 Sulphate Lys ALys BLys CLys D020406080Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Alkalinity (mg/L HCO3- ) C: Pile 5 Alkalinity 0.0010.010.11101001000Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Nitrate (mg/L) D: Pile 5 Nitrate Lys ALys BLys CLys DCHAPTER 2  77   Figure 2-15. Comparison of As, Fe, Mo, Sb, Se, and Zn concentrations from Pile 4 outflows. 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) A: Pile 4 - Lys A 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) B: Pile 4 - Lys B As Fe MoSb Se Zn0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) C: Pile 4 - Lys C 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) D: Pile 4 - Lys D As Fe MoSb Se ZnCHAPTER 2  78   Figure 2-16. Comparison of Cd, Cu, Co, Pb, and Ni concentrations from Pile 4 outflows. 0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) A: Pile 4 - Lys A 0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) B: Pile 4 - Lys B Cd Cu Co Pb Ni0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) C: Pile 4 - Lys C 0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) D: Pile 4 - Lys D Cd Cu Co Pb NiCHAPTER 2  79      Figure 2-17. Comparison As, Fe, Mo, Sb, Se, and Zn concentrations from Pile 5 outflows. 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) A: Pile 5 - Lys A 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) B: Pile 5 - Lys B As Fe MoSb Se Zn0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) C: Pile 5 - Lys C 0.0010.010.1110Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) D: Pile 5 - Lys D As Fe MoSb Se ZnCHAPTER 2  80   Figure 2-18. Comparison of Cd, Cu, Co, Pb, and Ni concentrations from Pile 5 outflows.  0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) A: Pile 5 - Lys A 0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) B: Pile 5 - Lys B Cd Cu Co Pb Ni0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) C: Pile 5 - Lys C 0.000010.00010.0010.010.11Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12Concentration (mg/L) D: Pile 5 - Lys D Cd Cu Co Pb Ni  CHAPTER 3    CHAPTER 3  81  CHAPTER 3: METHODOLOGY AND CONCEPTUAL MODEL DEVELOPMENT – FIELD BARRELS AND COLUMN EXPERIMENTS 3.1. Introduction Prediction of mine site drainage chemistry typically involves extrapolating results from small-scale experiments to those expected from mine-scale structures.  This process is referred to as up-scaling or scale-up and generally involves calculating scaling factor(s) to relate the two environments/conditions.  Although the concept is simple, the process of up-scaling can involve many scaling factors to capture differences in the physical, hydrological, geochemical, and/or microbiological properties at either scale.  The magnitude of the differences between the experimental scale and the scale needed for prediction may result in hundreds of unique scaling factors; however most authors choose a small sub-set of factors (Morin, 2013).  This scale-dependency or discrepancy has been shown in many studies, which can result in results that are order(s) of magnitude different between two scales (Drever and Clow, 1995; Malmström et al., 2000; Schnoor, 1990; Swoboda-Colberg and Drever, 1993; Velbel, 1993, Vereecken et al., 2007).  Understanding the scale-up question is important for accurate prediction of future drainage and the proper design of mine-waste disposal facilities (Blowes et al., 2007).  Under- or over- estimation of the quality of future drainage can be a costly error in many respects.  Therefore the construction, composition, and test conditions of these smaller scale tests should be assessed to ensure test results are applicable to larger scale structures and the scale discrepancy can be quantified.    This thesis research includes experiments at two scales differing by an order of magnitude (in height).  Chapter 2 described the larger scale experiments, which consist of two 10 m (height) experimental waste rock piles at the research site (Antamina) containing 19,000 tonnes of waste rock and covering an areal footprint of 1,296 m2.  This chapter describes two types of smaller, 1 m (high) waste rock experiments; fifteen field barrels at Antamina containing approximately 300 kg of relatively homogeneous waste rock, and two column experiments in the laboratory containing 170 kg of mixed waste rock.  Field barrels contain material characterized as a single waste rock class (i.e., Class A, B or C) and originate from a single location from Antamina’s open pit.  Laboratory columns contain a mixture of Class A and C waste   CHAPTER 3    CHAPTER 3  82  rock types and consist of material from several open pit locations.  Columns are mimicked after a vertical cross-section through Pile 5 – Lysimeter B (refer to Figure 2-1B), which contains Class A material overlying Class C material.  Notably, the larger of two waste rock dumps at Antamina (i.e., East Dump) contains Class A and C waste rock and these columns may be interpreted as a smaller scale analog of this mine-scale structure.  Chapter 3 provides details of the composition and construction of the small-scale experiments and provides a review of flow results and drainage chemistries.  A conceptual model of results from field barrels and columns is provided, with a brief comparison to larger-scale experiments (Chapter 2).  The presence of similarities or differences between experimental piles and field barrels or columns helps to form the basis of more detailed, quantitative studies to be presented in chapters 4, 5 and 6 and therefore quantifies the need or absence of scaling factors to relate the two scales.    3.2. Small-scale Waste Rock Experiments 3.2.1. Construction and Instrumentation Two (2) columns were filled with Class A and C waste rock in volumes proportionate to the vertical column above Lysimeter B in Pile 5 (~80% and ~20%, respectively; Figure 2-1B).  This material combination was selected as it represents a ‘mixed material’ composition found at the larger scale East Dump at Antamina.  Additionally, Class A material was characterized as PAG waste rock material due to its higher likelihood to release metals and generate acidity (via sulphide oxidation) relative to Class B or C waste rock (Antamina, 2007). 3.2.1.1. Field barrels All field barrels at Antamina were constructed in the same manner, as described in detail in Aranda (2009), and contain between 220 – 320 kg of waste rock from a single location from the open pit.   Briefly, the selection of waste rock for field barrel experiments began with 4 tonnes of material not used in experimental pile construction.  From this initial mass, approximately one-quarter was removed and cone-and-quartered twice with material larger than 10 cm removed.  This process resulted in approximately 300 kg of material remaining to be used in field barrels and/or other analyses.     CHAPTER 3    CHAPTER 3  83  Field barrel waste rock was deposited into clean 205-L (55-Ga) plastic drums with four 1” holes along drum sides to allow for aeration.   Waste rock was placed above a 15-cm layer of 30-mesh sand to allow drainage to be conveyed through a hole at the base of each drum and collected in a nearby bucket.  Field barrels are situated on constructed berms at a 5% slope to ensure drainage collection.  Figure 3-1 provides a schematic of the field barrel set up, with associated photos taken at Antamina.  Further details of the construction of field barrels can be found in Aranda (2009).  Field barrel experiments do not contain in situ instruments or sensors.  Four years of data from the 12 field barrels associated with Pile 4 or 5 is used in this thesis (i.e., January 2009 to December 2012).  Pile 2 field barrels (i.e., P2-1A, P2-2A and P2-3A) were constructed in October 2007 and approximately 5 years of data from these experiments is used in this research (i.e., October 2007 to December 2012).  3.2.1.2. Laboratory columns  The geometry of both column experiments (Column 1 and Column 2) was identical, ~ 1 m height and 0.45 m diameter, and outfitted with eight ½″ (outer diameter; o.d.) holes along the height of the column and twelve ¼″ (o.d.) holes dispersed evenly among 4 height intervals (i.e., 0.16 m, 0.44 m, 0.66 m, and 0.92 m from the base; Figure 3-2).  Columns contained comparable material and geometries, but differ in regards to its initial microbiological characterization, which is described in detail in Section 3.2.4. A fine, well-sorted silica sand layer (i.e., 5 cm height) was placed at the base of the columns to ensure drainage is conveyed to collection buckets.  Two types of waste rock were used in the columns: Class A and Class C, in a ratio of 2:1.  Class A waste rock was divided into two types, labeled Class A-1 and Class A-2 and columns can be described as containing three waste rock strata.  Class A-1 material was comprised of 3 unique Class A discharges (i.e., P2-1A, P2-2A and P5-6A) in a 2:1:1 ratio, whereas Class A-2 was comprised entirely of a single discharge (i.e., P2-3A).  Class C waste rock was representative of Class C material used in Pile 4 and 5 and three field barrels (i.e., P4/5-1A, P4/5-5A/B).   Strata-specific waste rock (i.e., Class A-1, Class A-2 and Class C) used in the columns were individually combined, cone-and-quartered, and divided into two sets (for the two columns).  This method aimed to ensure a relatively homogeneous bulk composition in each of the two columns.     CHAPTER 3    CHAPTER 3  84  Material segregated for Column 1 was dried at 50oC, to exterminate the majority (or all) bacteria associated with this material.  Microbial enumerations,’ using the most probable number (MPN) method (Cochran, 1950) were conducted on this material, which showed a lower total microbial population was initially present in Column 1.  The enumeration method (in detail) and full results are found in Appendix B. Following heat treatment of Column 1 material, a sterile, dilute salt solution was percolated three to four times through waste rock segregated for Column 1 and 2 to encourage growth of bacteria present and remove mobile salts on mineral surfaces that may have precipitated during storage or drying prior to placement in the columns.  The dilute salt solution is an amalgamation of known growth media for Acidithiobacillus ferrooxidans, Acidithiobacillus thiooxidans and neutrophilic thiosulphate oxidizing bacteria (e.g., Thiobacillus spp) and contains (per litre of deionized water): (NH4)2SO4 0.03 g; KH2PO4 0.01 g; MgSO4*7H2O 0.04 g; CaCl2 0.023 g.   Following heat treatment and/or salt percolation, waste rock was placed in the columns.  Specifically, Class C material was placed above the silica sand layer, with Class A-2 as the middle stratum and Class A-1 material as the upper layer (Figure 3-2A).  Total mass of these materials per column is provided in Table 3.2.  Instrumentation Each column was instrumented with 8 Decagon 5TE sensors, 7 Decagon MPS-1 sensors, 3 Vaisala GMT-220 CO2 sensors, 3 Apogee O2 sensors and 6 (manual sampling) gas lines.  Decagon 5TE sensors measured temperature, EC and volumetric water content, which were recorded by Decagon EM50 dataloggers at 5 minute intervals (Figure 3-2B).  Sensors were located at 19 cm, 25 cm, 30 cm (Class C); 46 cm, 53 cm (Class A-2); 63 cm, 71 cm, 74 cm (Class A-1) depth and calibrated prior to installation.  Decagon 5TE sensors were surrounded by a 2:1 mixture of silica sand and sieved (<5 mm) waste rock to ensure contact with sensor tips and collected data reflects in situ physicochemical conditions of matrix pore waters (Figure 3-3B and C ).   Decagon MPS-1 sensors measured the dielectric permittivity of a porous matrix to estimate its water potential (or matric potential), on similar 5 minute intervals.  The MPS-1 sensors were located at 16 cm,   CHAPTER 3    CHAPTER 3  85  25 cm (Class C); 40 cm, 48 cm (Class A-2); 56 cm, 67 cm (Class A-1) depths and were calibrated prior to installation.  Similar to the 5TE sensors, water potential sensors required an adequate contact with water and these sensors were surrounded by clean silica sand.   All Decagon sensors were connected to Decagon EM50 dataloggers for data collection and storage of data records for up to 4 months. Three Vaisala GMT-220 sensors were installed in each column (1 per material type) for the measurement of carbon dioxide concentrations (Figure 3-2B, Figure 3-3D).  These sensors employed an infrared gas analyzer (IRGA) technique using a single-beam dual-wavelength non-dispersive infrared (NDIR) light source and a silicon-based sensor.  The sensors (2 cm diameter x 15 cm length) were protected by a plastic cover with an open end to the environment.   Measurements were taken every 5 minutes and are controlled through a CR1000 datalogger (see Figure 3-2B).  Further details of these sensors can be found in Appendix B.   Apogee SO-series O2 sensors are a galvanic-cell type oxygen sensor that measures the current flow between electrodes and relates the voltage differential to an oxygen concentration value (Figure 3-3C).  Oxygen measurements were taken every 5 minutes and data is recorded through a Campbell Scientific CR1000 datalogger.  Details of calibration of Apogee sensors can be found in Appendix B.  Gas lines for manual sampling exited along 6 NPT ports along the side of each laboratory column (12 cm, 26 cm (Class C); 37 cm, 54 cm (Class A-2); 65 cm, 74 cm (Class A-1)) and used a Swagelok reduction fitting to connect 1/8″ (o.d) tubing from the in situ sampling location to a sampling port at the exterior of the columns.  Each tube pointed downwards within the waste rock and contained a cloth mesh at the termination of each line to prevent moisture from accumulating in the tubing.  A close-off valve and spiral connector (for syringe) was located at the sampling end to limit atmospheric contamination. The basal lysimeter was built into the original column design as a low angle slope to a basal NPT port.  The thin silica sand layer was placed at the base of the column to ensure drainage flows to the basal lysimeter (Figure 3-5A).   Drainage released from the column was conveyed to a tipping bucket (Texas Instruments TR525-M) that recorded flow rate and volume on 5 minutes via a connection with a CR1000   CHAPTER 3    CHAPTER 3  86  datalogger (see Figure 3-2C).  Tipping buckets were calibrated to have a 0.1 mm per tip resolution, with a maximum tip rate of 3 seconds per tip.  Drainage waters from each column were siphoned into large 10 gallon buckets for collection of monthly alkalinity samples.  Additionally, the secondary collection acted to crudely quantify drainage volumes in the event of malfunctioning tipping buckets.   One sub-lysimeter was installed at ~ 0.5 m (from the basal lysimeter) to sample drainage specific to Class A waste rock.  Sub-lysimeter drainage was collected through a NPT port located below the internal sub-lysimeter.  A thin 5 cm layer of coarse silica sand was placed at the base of the sub-lysimeter, to ensure drainage were conveyed to the collection bucket.  Waste rock overlying the sub-lysimeter consisted of 1 part Class A-2 and 1 part Class A-1, each with a thickness of approximately 12 cm; Figure 3-3E and F).  Sub-lysimeter walls were extended to the top of the column at 0.76 m, to ensure water entered the lysimeter and flowed to the collection vessel.   The column experiment was initiated on March 7, 2011 and ran for 460 days, or 1.5 water years.  Refer to Appendix B for schematics outlining the location of each instrument within Column 1 or 2, and the CR1000 datalogger program script (to record O2, CO2 and tipping bucket data).   3.2.2. Waste Rock Physical Characteristics  Twelve (of 15) field barrels contained representative waste rock used to constructed Pile 4 and 5.  The remaining three (of 15) field barrels were associated with Class A material from Pile 2, which was also used in the two column experiments.  Specifically, column experiments consisted of Class A and C material in a 2:1 ratio.  Class C material was comparable to waste rock used in the first and fifth discharge of Pile 4 and 5, whereas the Class A proportion consisted of material from  four unique open pit locations and associated with Pile 2 (n = 3) and Pile 5 (n = 1).  The purpose of using several sub-sets up Class A material was to be reflective of the heterogeneity within a single waste rock class.     CHAPTER 3    CHAPTER 3  87  3.2.2.1. Total Mass Details of the reference pit location and mass were used to construct field barrels and columns can be found in Table 3.1 and 3.2 (respectively).  The naming of each field barrel in this thesis includes its reference pile and discharge number.  For example; P2-2A refers to waste rock used in Pile 2 – discharge 2.  Material labeled as ‘P4/5-1A’ implies the same material was used in the first end-dump discharge for both Pile 4 and 5. The subscript of ‘A’ or ‘B’ after each field barrel name refers to one of three situations, namely: 1) A duplicate sample (e.g., P4/5-5A and -5B); 2) Two unique reference locations from the open pit (e.g., P4/5-1A and -1B); or, 3) The same pit origin, but two rock types (e.g., P4-4A: white marble; P4-4B: black marble).   Field barrels associated with Pile 4 and 5 were constructed in the first quarter of 2009, whereas Pile 2 field barrels were initiated in August to September 2007 (Table 3.1).  The total mass of waste rock contained in each of the 15 field barrels ranges between 220 kg and 320 kg.   Table 3.2 provides information of waste rock used in column experiments.  A total of 170 kg and 177 kg or waste rock was used in Column 1 and 2, respectively, which contained Class A and Class C material in proportions of approximately 2:1 (respectively).  Class A material is shown as two types: Class A-1 and Class A-2.  Class A-1 material consists of waste rock used in Pile 2 and Pile 5, whereas Class A-2 contains waste rock from the third discharge of Pile 2 only (i.e., P2-3A).  Class C material is a combination of waste rock used in the first and fifth discharges of Pile 4 and 5.   3.2.2.2. Particle size distribution (PSD) curves Waste rock material used in field barrels and columns did not reflect the range of grain diameters present in experimental piles, as described in Chapter 2, or mine-scale dumps due to smaller experiment geometries.  Instead, waste rock with grain diameters exceeding 10cm were excluded from field barrels and columns.  Therefore particle size distribution (PSD) curves conducted on these waste rock materials (ASTM-D-5519) were re-normalized to reflect the size fractions included in the small-scale experiments   CHAPTER 3    CHAPTER 3  88  (Figure 3-4).  Note that PSD analyses were performed on a limited number of samples associated with Pile 4 and 5.  Specifically, PSDs were not available for; Class A: P5-2A and P5-3A; Class B: P4-2A; Class C: P4/5-1A/B and P4-3A.  Full grain size curves that include grain diameters greater than 10 cm can be referenced in Figure 2-1 for Pile 4 and 5 and Peterson (2014) for Pile 2 curves.  Results shown in Figure 3-4 show there was a distinct difference between Class A and C material.  The 4.75 mm sieve size is assumed to be the upper boundary between particles contributing to matrix material, which is based on a study by Yazdani et al. (2000) that showed that material with grain diameters greater than 4.75 mm do not have significant capillarity in unsaturated conditions.  Therefore Class A waste rock used in Antamina field barrels contained a much higher matrix material proportion (i.e., 21% - 43%) relative to Class C material (i.e., 8% - 15%).  Only two samples of Class B material are shown in Figure 3-4 and, although this is a small sample size, these results suggest Class B material was finer grained than Class C and reflective of the coarser boundary of Class A waste rock (i.e., 23% - 25%).  Comparison to the amount of material passing the same sieve size from the full PSD curves (i.e., in large-scale piles) shows the truncated grain diameters resulted in a small 1.5x to 2x increase in its matrix component.    Calculated uniformity coefficient (𝐶𝑈 or 𝐷60 𝐷10⁄ ;  [-]) values for Class A-bearing field barrels were greater than 100 and were classified as well-graded, whereas those for Class C were less than 20 and considered to be poorly graded (Morin et al., 1991).  Field barrels containing Class B material were an intermediate between Class A and C, with calculated values of approximately 35 and 40 and also classified as well-graded material.   3.2.2.3. Density, porosity and saturated hydraulic conductivity Bulk dry density and porosity measurements for field barrels containing Pile 4 and 5 waste rock were described in Chapter 2 (Section 2.2.2.3).  Measurement of Pile 2 densities and porosities are described in Peterson (2014) and are also presented in Chapter 2.   The bulk dry density and porosities of each waste rock mixture used in column experiments (i.e., Class A-1, Class A-2 and Class C) were estimated using a laboratory-based technique.  Specifically, the bulk   CHAPTER 3    CHAPTER 3  89  density was estimated by first oven-drying all material marked for the Column 1 (at 50oC for 1 – 2 days).  Dried material was weighed and loosely compacted in a 0.5 m3 container to estimate its occupied volume and calculate bulk dry density.  Porosity measurements were determined by saturating oven-dried waste rock and recording the volume of water needed to reach a thin but visible film of water at the material surface (i.e., porosity = water volume/total volume (m3∙m-3)).   Following porosity calculations, water was allowed to drain in order to calculate field capacity values (i.e., total volume – drained volume).     Bulk density, saturated porosity and field capacity values associated with waste rock used in Column 1 are shown in Table 3.3.  Class C material had the highest saturated porosity and the lowest field capacity, which complemented its lower proportion of finer grain sizes (relative to Class A or B).  The finer grained texture of Class A waste rock likely contributed to its significantly higher field capacities, relative to those from Class B or C material.  Waste rock used in both columns was compositionally similar and parameters shown in Table 3.3 were applicable to Column 2 materials.   3.2.2.4. Soil water characteristic curves (SWCCs) Similar to the discussion in Section 2.2.2.4 of Chapter 2, Speidel (2011) imported PSD results and estimates of density and porosity measurements into SoilVision to generate SWCCs for waste rock used in field barrel and column experiments (Figure 3-5).  Results show that Class A material had the highest maximum and narrowest range of air entry values, followed by Class B and C (i.e., Class A: 0.06 – 6 kPa; Class B: 0.06 – 0.09 kPa; Class C: 0.03 – 0.04 kPa).     3.2.3. Waste Rock Geochemical Characterization 3.2.3.1. Bulk geochemical analyses Material used in field barrels and column experiments was collected from various regions in the open pit at Antamina.  A geochemical analysis was performed by Antamina immediately following excavation to ensure waste rock is assigned the proper classification.  These results for field barrels associated with Pile 4 and 5 are found in Table 2.5 and 2.6, whereas results associated with waste rock from three Pile 2 discharges (i.e., P2-1A, P2-1B and P2-3A) used in column experiments are shown in Table 3.4. Bulk geochemical characteristics of waste rock used in column experiments are shown in Appendix B.   CHAPTER 3    CHAPTER 3  90  Results show Class A-1 material contained significantly higher metal percentages than Class C (i.e., Cu, Mo, Pb, Zn) and was the only material to contain detectable copper sulphides.  Class A-2 contained greater than twice the Fe and As percentage than Class A-1.  Class C material showed the lowest bulk metal composition of the three materials.  However, Class C did show elevated arsenic, which was likely the main element of concern from this relatively non-acid-generating material.  3.2.3.2. X-ray fluorescence (XRF) X-ray fluorescence of the waste rock in individual field barrels and column experiments is shown in Table 3.5.  XRF methods applied to these samples are identical to those described in Section 2.2.3.2 for materials used in the large-scale experimental piles.  Similar to results discussed in Chapter 2, Class A waste rock contained a higher proportion of silicates relative to carbonate minerals, as observed from high SiO2 (66% – 79%) versus low CaO (1.5% – 4.0%) percent values (respectively).  In contrast, Class B and C waste rock contained high CaO percentages (i.e., 34 – 39 %) and low SiO2 percentages (13 – 22%) that suggest carbonate minerals contributed to the majority of the buffering capacity in these waste rock types.  Other unique differences between the three waste rock classes included 2x to two orders of magnitude lower Cu and Mo concentrations in Class B/C waste rock relative to Class A.  As well, Class A waste rock contained higher S percentages that suggest this material hosted a higher sulphide component was present than those values from Class B and C materials (i.e., Class A: 0.29% - 2.72% S; Class B/C: 0.04% - 0.21% S). 3.2.3.3. X-ray diffraction (XRD) XRD analyses were conducted at UBC on multiple discharges and were described in detail in Section 2.2.2.2.  Similar to bulk geochemical results, Table 3.5 displays the mineralogical composition of waste rock used in Pile 4 and 5 field barrels and column experiments. XRD analyses supported observations from XRF results, namely: Class A material contained a higher silicate (i.e., K-feldspar, oligoclase, quartz), and Cu- and Mo- bearing mineral proportion (i.e., chalcopyrite and molybdenite), with as no detectable calcite relative to Class B and C waste rock.  However a small   CHAPTER 3    CHAPTER 3  91  proportion of an iron-carbonate mineral, siderite, was observed in 3 (of 5) Class A samples (i.e., P2-1A: 0.22%; P2-2A: 0.47%; P5-6A: 0.06%).   Sulphide minerals contributed to approximately 1% – 5% of the mineral assemblage of Class A, as mostly chalcopyrite and pyrite, with lesser molybdenite and pyrrhotite.  Material from the field barrel P2-3A, which was used in column experiments, contained the highest sulphide percentage (i.e., 5%) which was up to 5x larger than sulphide percentages from the other 4 Class A samples analyzed (i.e., <2%).   In regards to Class B and C waste rock, pyrrhotite was observed in all samples submitted for XRD and contributed to the majority of the sulphide proportion (i.e., 0.8% – 2.6%) in these material types.  In general, Class B and C waste rock were mineralogically similar with the exception Class B contained slightly higher diopside and actinolite proportions and Class C assemblages contained slightly higher amounts of Na-plagioclase (i.e., albite).  3.2.4. Waste Rock Microbiology Enumerations of microbial populations associated with waste rock material used in laboratory columns (i.e., Column 1 and Column 2) were conducted prior to the initiation of the experiment.   Specifically, populations of three bacterial phenotypes were determined using the Most Probable Number (MPN) technique (Cochran, 1950).   Enumerated MPN scores (as number of cells or bacteria per gram of solid material; bacteria/g) of initial bacterial populations are shown in Table 3.7.  Enumerations of Class A-2 material (i.e., P2-3A) show this material contained a relatively robust microbial population of near equal proportions of acidophilic and neutrophilic bacterial phenotypes.  Class A-1 waste rock (i.e., P2-1A, P2-2A, and P5-6A) contained significantly different microbial communities, which consisted of four orders of magnitude higher neutrophilic,relative to acidophilic, phenotypes (i.e., 105 bacteria/g vs. 101 bacteria/g).  Interestingly, this microbial consortium was more closely comparable to microbial populations associated with Class C material.     CHAPTER 3    CHAPTER 3  92  3.2.5. Water Balance and Calibration  3.2.5.1. Field barrels Drainage was collected from field barrels on a monthly basis to record outflow volumes and collect samples for aqueous chemistry (to be discussed in Section 3.3.3).  Precipitation was recorded at one of two weather stations: the Yanacancha weather station located on the west shore of the Tailings Pond and the experimental test pile site, both of which are located at the Antamina mine.  The former station was located approximately 4 km from the experimental pile site and at a lower elevation.   3.2.5.2. Laboratory columns Rainfall was applied to laboratory columns using an air-atomizing full-cone spray nozzle (Lechler Airmist® Series 136.2) to mimic daily precipitations at the Antamina mine.  Activation of the nozzle required a simultaneous air pressure of ~ 100 psi and water pressure of greater than 10 psi.  Rainwater was pumped from a large, acid-washed carboy containing filtered tap water using a peristaltic pump set between 3.05 rpm and 3.15 rpm to ensure a flow rate of approximately 100 mL per minute.   Precipitation events mimicking wet and dry season rates at Antamina were dictated by the rainfall duration as opposed to intensity.  Wet versus dry season durations were calculated by dividing the total monthly rainfall (from site data) into daily amounts and dividing by the known pump flow rate.  For logistical reasons, it was decided that rain would be applied to columns 5 (of 7) days in a week.   At two months into the experiment, eaves were installed on the interior rim of each column to capture side-flow that may disrupt hydrological flow patterns.  Water collected by the eaves was conveyed to a collection bottle outside of the columns, to ensure it did not infiltrate into column materials.  Based on applied versus captured water volumes, it was calculated that 42% of applied water was captured with these eaves and rain durations were increased by 42% to ensure precipitation amounts mimicked Antamina observations.   Table 3.8 presents the monthly precipitation rate to be applied to columns each month.    CHAPTER 3    CHAPTER 3  93  Although eaves were not installed on the columns for the first two months, it was assumed this did not significantly affect the flow since these dry season months contributed to 2% of the total annual precipitation.  3.2.6. Aqueous Sampling 3.2.6.1. Field barrels Samples for aqueous chemistry were collected from field barrel drainage on a weekly or bi-weekly basis, except during the dry season when there was no infiltration through the field barrels.  Water samples from field barrels were measured for several physical parameters in the field (i.e., pH, specific conductance, temperature, and dissolved oxygen) and samples were collected for total and dissolved solutes.  The following elements were analyzed for dissolved and total concentrations: Al, As, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, Pb, S, Sb, Se, Si, Ti, Zn.  Total alkalinity is measured on a monthly basis from collected drainage waters.  The dissolved and total samples are submitted to an external laboratory.  Details regarding the analytical method and parameter detection limits provided by this external laboratory can be referenced in Corazao-Gallegos (2007).  3.2.6.2. Laboratory columns Like field barrels, drainage from column experiments reported to the basal lysimeter in the wet season only.  Therefore at the onset of the wet season, samples were initially taken daily to capture first flush chemistries and subsequently on a less frequent basis (i.e., weekly to twice-monthly) later in the experimental period.  Samples were analyzed for both total and dissolved elements using inductively coupled plasma optical emission spectroscopy and mass spectrometry (i.e., ICP-OES and ICP-MS, respectively).   Proper sampling and storage (i.e., filtering, acid-preservation, cool temperatures (4oC)) was maintained for all samples collected.  Details of these steps can be found in Appendix A.        CHAPTER 3    CHAPTER 3  94  3.2.7. Tracer Test A tracer test was applied to column experiments at the height of the first wet season (i.e., time = 240 days or 0.65 yr).  Bromide (as LiBr) was selected as the applied tracer due to its conservative nature and low to negligible background concentrations.  Applied concentrations to the upper surface of each column were 3100 ± 0.1 mg·L-1 and administered as a single, ~10 mm rainfall event over a duration of 17 minutes.   The tracer was applied with a moderate uniformity or distribution to the upper surface of column waste rock, as shown by calculated CUC values of approximately 67 (Christiansen, 1947; refer to Equation 2-1).   Samples for bromide analyses were taken on an hourly to weekly basis and concentrations were measured using the same spectrophotometric method (Presley, 1971) described previously in Section 2.3.4.  Duplicates were performed every 25 samples for quality assurance and quality control purposes.   3.3. Observations 3.3.1. Flow and Evaporation 3.3.1.1. Field barrels Drainage volumes recorded from field barrels are shown in Figure 3-6.  Drainage volumes are presented as area-normalized cumulative volumes to enable comparison between experiments at different scales.  An estimation of the average yearly specific discharge from field barrels experiments shows Class A waste rock contained the lowest values at 167 mm·a-1 to 185 mm·a-1.  In contrast, Class C waste rock had the highest values of 183 mm·a-1 to 218 mm·a-1 and Class B waste rock shows intermediate values between Class A and C (i.e., 155 mm·a-1 to 203 mm·a-1).  Therefore evaporation in field barrels ranged between 83% (Class C) to 86% (Class A), which was significantly higher than evaporation estimated from experimental piles (i.e., ~33%; Section 2.4.3). The overall finer grain sizes associated with waste rock used in field barrels, relative to larger scale piles, is attributed to scale geometries that limit the largest particle size to be placed in a field barrel.  Finer grain sizes result in a higher likelihood for material to behave in a soil-like manner (Dawson and Morgenstern, 1995) or present lower flow velocities and therefore evaporation estimates may be higher.   CHAPTER 3    CHAPTER 3  95  3.3.1.2. Laboratory columns Figure 3-7 presents the cumulative outflow and estimated evaporation percentages from columns over the experiment duration of 460 days.   Column 1 received close to 2 m of precipitation and approximately 58% or 1.1 m infiltrated the column.  Column 2 received less precipitation (i.e., 1.8 m) and 51% or 0.9 m is reported as infiltration.  The difference between the two columns was attributed to the placement of eaves in the columns.  Specifically, eaves were placed closer to the material surface in Column 2, which resulted in a greater capture of precipitation contacting the column walls.   Outflow flux values for columns were estimated as 4 mm∙d-1 – 5 mm∙d-1 for the wet season only, which were similar to the wet season flux rates for Pile 5 Lysimeter B (i.e., 4.5 mm∙d-1; Table 2.13).  Based on average moisture contents measured from in situ sensors in each column (results shown in Appendix B), flow velocities from Column 1 and Column 2 were calculated as 21 – 33 mm∙d-1 and 16 – 28 mm∙d-1, respectively.   Evaporation was calculated using the water balance method and assumes negligible change in storage and the absence of run off.  At the end of the experiment, evaporation is estimated as 42% and 49% for Column 1 and 2 (respectively) and is considerably lower than estimated evaporation from field barrels (i.e., ~85%).      3.3.2. Tracer Results  Figure 3-8A shows initial tracer breakthrough from both columns at 1.5 days following application, with peak concentrations at 5 and 7 days (for Col. 1 and Col. 2, respectively).   Although the timing of peak tracer values from Column 2 was slower than Column 1, maximum concentrations were similar (i.e., C/Co = 0.12 or 360 mg∙L-1  Br-).   At the cessation of the wet season, both columns released 75% of the applied tracer mass (Figure 3-8B).  These results strongly suggest efforts to maintain material homogeneity between the two columns were successful and, in general, hydrological regimes were comparable.  As well, the presence of a quick breakthrough at 1.5 days following application suggests preferential flow is present; however the majority of infiltration likely followed matrix flow paths.    CHAPTER 3    CHAPTER 3  96  Although a tracer test was not applied to field barrels, nitrate concentrations in field barrel drainages show a decreasing trend with time.  Specifically, nitrate concentrations decreased each year by 5 – 100x (Figure 3-10), or from initial values between 10 mg∙L-1 and 50 mg∙L-1 to less than 1 mg∙L-1 by the end of the experiment.   3.3.3. Aqueous Chemistry 3.3.3.1. Field barrels Temporal plots of field barrel drainage chemistries are presented in Figure 3-9 (pH and sulphate), Figure 3-10 (alkalinity and nitrate), Figure 3-11 (As and Zn), Figure 3-12 (Mo and Sb), and Figure 3-13 (Cu and Co).  Temporal plots associated with other dissolved metal concentrations can be found in Appendix B.   pH and sulphate  Measured pH from field barrels bearing Class B and C waste rock (i.e., n = 9) typically showed neutral to alkaline values between pH 7 and pH 9 (Figure 3-9).  In contrast, leachate from Class A field barrels reached pH values less than pH 6.5 in 4 (of 6) experiments by December 2012.    Drainage from one field barrel associated with Class A waste rock used in Pile 2 (i.e., P2-3A) had acidic pH values in the first water year (i.e., October 2007 – June 2008), with pH values decreasing by 1 to 2 pH units each subsequent year, and reach extremely acidic drainage (i.e., pH 2.5) at the end of 2012.  The other 14 field barrels, with acidic or neutral drainage, showed more of a slight decreasing pH trend with time or a rate of approximately 0.1 pH units per year (Figure 3-9).  Sulphate concentrations generally ranged between 20 mg∙L-1 and 1000 mg∙L-1 for most field barrels, with higher concentrations at the start of the wet season or when secondary minerals (that likely precipitated on rock surfaces during drier periods) were dissolved and flushed (Figure 3-9).  Differences in sulphate concentrations between the start and end of the wet season can vary by an order of magnitude in some field barrels; however the range of sulphate concentrations observed remained relatively stable throughout the experimental duration.  Two exceptions to this stable sulphate trend were observed with two Class A field barrels (i.e., P2-3A and P5-3A) that present increasing values with time and surpass 1000 mg∙L-1.  Sulphate concentrations from P2-3A, which released acidic leachate following one water   CHAPTER 3    CHAPTER 3  97  year, presents concentrations that increase by up to 5x per year with maximum values of 16,000 mg∙L-1 at the start of its sixth wet season.    Acid-buffering capacity Alkalinity concentrations were relatively stable throughout the 4 to 5 years of data shown in Figure 3-10, with values ranging between 20 mg∙L-1 and 80 mg∙L-1 HCO3- for 12 (of 15) field barrels.  This suggests that most field barrel materials contained a significant and available (i.e., non-passivated) carbonate-buffering capacity.  Exceptions to this trend were observed for three Class A field barrels (P2-2A, P5-3A and P5-2A) that contain acidic drainage and, in these scenarios, sulphide oxidation outpaced dissolution of available carbonate minerals or carbonate buffering capacity is exhausted.      Measured Si values (not shown; Appendix B) were typically between 0.1 mg∙L-1 and 7 mg∙L-1and decreased yearly in field barrels containing neutral to alkaline pH values.  For those field barrels with acidic pH values (i.e., Class A field barrels), Si concentrations were approximately 10 mg∙L-1 or higher and increased with time.  This trend suggests silicate mineral dissolution was the main acid-neutralization mechanisms for this material type, which was also supported by the low carbonate mineral composition observed from XRD analyses (Table 3.6).  Metal concentrations Figure 3-11 presents dissolved As and Zn concentrations measured from field barrel drainages, which were two (of 3) main qualifiers used by Antamina to classify waste rock as Class A, B or C (i.e., with the other qualifier is sulphide %; Table 2.1).  Arsenic and zinc are elements with an increased mobility at neutral pH conditions since they form oxyanion metalloids (i.e., AsO43-) or are present as weakly hydrolyzing cations (i.e., Zn) that may form complexes with inorganic/organic ligands.  Most soil surfaces have a net negative charge at higher pHs, which prevents most oxyanionic species from sorbing to soil surfaces and therefore they remain in solution.  For weakly hydrolyzing metals, the hydrolysis of these metals occurs at much higher pHs (relative to other metals such as Cu) or are out-competed for a finite number of soil adsorption surfaces by other species.  Therefore under the prevailing neutral pH conditions   CHAPTER 3    CHAPTER 3  98  from Antamina field barrels, metal(loid)s such as As, Cd, Mo, Sb, Se, and Zn may be released in higher concentrations relative to metals with higher mobility at lower pH values (i.e., Co, Cu, Fe, Ni, and Pb).    Arsenic Arsenic concentrations varied significantly between field barrels and, although bulk solid-phase concentrations are provided (Table 2.5, 2.6 and 3.4), a clear trend was not observed between drainage concentration and initial As abundance.  The highest As concentrations (i.e., 21 mg∙L-1) were measured from field barrels containing Class C waste rock used in Pile 4 and 5 (i.e., P4/5-5A and P4/5-5B; Figure 3-11E), however most Class A and C waste rock contained values between 0.01 mg∙L-1 and 1 mg∙L-1 with Class B field barrel drainage releasing the lowest As concentrations (i.e.,<0.001 mg∙L-1 to 0.01 mg∙L-1).   In general, As concentrations decreased with time with the exception of the field barrel containing a highly acidic leachate (i.e., P2-3A).  Measured concentrations from P2-3A increased by 4 orders of magnitude over 2 years (Figure 3-11A).  This increase was concomitant with a similar increase in Fe concentrations (shown in Appendix B), which suggests iron hydroxide precipitates that may have adsorbed As species were dissolving into solution at these lower pH conditions.   Zinc Zinc concentrations are shown in Figure 3-11 and, in contrast to As trends, showed an increasing temporal trend in field barrels containing Class A and B waste rock.  The highest values were observed from the acidic P2-3A field barrel and reached maximum concentrations of 362 mg∙L-1, whereas the majority of Class A and B field barrels showed values between 0.1 mg∙L-1 and 1 mg∙L-1.  Class C field barrels showed steady or decreasing Zn concentrations with time that were typically near 0.01 mg∙L-1.  Molybdenum and Antimony Concentrations of two other metal(loid) species with increased mobility at neutral pHs (i.e., Mo and Sb) are shown in Figure 3-12.  Measured Mo concentrations were two to three orders of magnitude higher in Class A field barrels, relative to those from Class B and C field barrels.  The only exception to this trend were measured Mo values from P2-3A (Figure 3-12A), which presented acidic drainage and therefore   CHAPTER 3    CHAPTER 3  99  was not favourable to Mo mobility.  Antimony values from Class A field barrels containing neutral pH drainages were typically near 0.1 mg∙L-1, whereas Class B field barrels showed similar or lower Sb values.  Class C waste rock showed a wide variation in dissolved antimony concentrations, with values between 0.001 mg∙L-1 and 2 mg∙L-1.  Copper and Cobalt  Figure 3-13 presents dissolved concentrations of Cu and Co measured from field barrel drainages.  Measured Cu concentrations varied between detection limit values (i.e., 0.001 mg∙L-1) and up to 100x higher in most field barrels.  Seasonal trends in Cu concentrations did not show the characteristic high initial wet season values that decreased to the onset of the dry season, which suggests metal attenuation mechanisms (e.g., malachite precipitation) may have occurred in these carbonate-dominant rock types.   Other metal(loid)s that did not show decreasing values in a single wet season included: Al, As, Fe, Mo, Sb, and Si, which suggests other minerals may have been at saturation and other secondary mineral precipitation reactions may have attenuated their release (e.g., gibbsite, goethite, powellite, amorphous SiO2).   Cobalt was sporadically measured in Pile 2 neutral pH field barrels (i.e., P2-1A and P2-2A) at concentrations between 0.0001 mg∙L-1 and 0.001 mg∙L-1.  Copper and cobalt concentrations from P2-3A were significantly higher (i.e., 3 – 5 orders of magnitude) than those observed from other field barrels, with Cu and Co maximum concentrations of 4,700 mg∙L-1 and 5 mg∙L-1 (respectively).   Summary of waste rock specific characteristics  In regards to drainage chemistries characteristic of each waste rock class, a few observations can be made, namely:   Class A field barrels (n = 6): P2-1A, P2-2A, P2-3A, P5-2A, P5-3A, P5-6A o Contained the lowest average yearly infiltration (i.e., 171 mm∙a-1) and highest evaporation (86%), which was also observed by high AEV values from SWCCs; o Greatest likelihood to release acidic drainage due to low carbonate mineral composition and high sulphide percentage, which was supported by drainage from 4 (of 6) field   CHAPTER 3    CHAPTER 3  100  barrels showing acidic pH values, and decreasing alkalinity and increasing sulphate concentration trends; and, o Drainage from Class A waste rock tended to be enriched in metal(loid) concentrations at both neutral and acidic pH conditions.  At acidic pHs, Al, As, Cd, Co, Cr, Cu, Fe, and Ni increased by 0.2 to 1 order of magnitude each year for the dataset presented here.  In contrast, at neutral pHs, Mo and Cd concentrations were released at significantly higher concentrations, relative to those from Class B and C waste rock.   Zinc was released at significantly higher concentrations from Class A waste rock, relative to Class B or C, at both acidic and neutral pHs, and Zn values increased with time.    Class B field barrels (n = 3): P4-2A, P4-4A, P4-4B o Slightly higher average yearly infiltration (i.e., 179 mm∙a-1) with similar evaporation estimates (i.e., 85%) from Class B waste rock, relative to Class A;   o Drainage was alkaline (i.e., pH 8), with relatively stable alkalinity values around 40 mg∙L-1 HCO3- and sulphate concentrations that vary an order of magnitude between the start and end of a single wet season; o Se was released at detectable values from Class B field barrels only, whereas Se concentrations were near or below detection limit values for Class A and C field barrel drainages; and, o Zn is another metal of concern from most Class B field barrels, which presented concentrations that steadily increase with time and was similar to the trend observed from Class A field barrels.      Class C field barrels (n = 6): P4/5-1A, P4/5-1B, P4/5-5A, P4/5-5B, P4-3A, P5-4A, o Average yearly infiltration rates (i.e., 204 mm∙a-1) were 19% and 14% higher than those from Class A and B field barrels (respectively), and contained the lower evaporation estimates (i.e., 83%) relative to field barrels containing Class A or B material;   o Similar to Class B field barrels, drainage showed alkaline pHs and slightly higher alkalinities (i.e., 50 mg∙L-1); and,   CHAPTER 3    CHAPTER 3  101  o As and Sb were the main elements of concern, with concentrations up to 10 mg∙L-1 and 2 mg∙L-1, respectively.  3.3.3.2. Laboratory columns Temporal plots of drainage chemistries from column experiments are presented in Figure 3-14 (pH, sulphate, Mn, Si, Ca and Na), Figure 3-15 (As, Cd, Mo, Sb, Se, and Zn), and Figure 3-16 (Co, Cr, Cu and Ni).  Temporal plots associated with other dissolved metal concentrations can be found in Appendix B, along with full analytical results received from the laboratory.   pH and sulphate Initial flush drainages contained acidic pH water, with pH 6.0 and pH 4.5 from Column 1 and 2, respectively (Figure 3-14A).  Measured pH from Column 1 drainage was approximately 1 – 2 pH units higher than those values observed from Column 2 throughout the remainder of the wet season and, halfway through the wet season (i.e., t = 240 days), drainage pH from Column 1 increased to near neutral values (i.e., pH = 6.8 – 7.2).  In the same period, Column 2 drainage maintained acidic pH values (i.e., pH = 4.5 – 5.0).   Sulphate concentrations showed a similar trend from both columns, with high initial values (i.e., Column 1: 2,900 mg∙L-1; Column 2: 8,100 mg∙L-1) that decreased by 2x – 5x by the end of the wet season (Figure 3-14B).      Acid-buffering capacity Alkalinity (not shown) was very low (i.e., <1 mg∙L-1 HCO3-) for both columns, which was unsurprising given low measured pH values and high sulphate concentrations.  Although waste rock used in column experiments contained one-third carbonate-dominated material (i.e., Class C), low pH values suggest calcite buffering was not occurring in the columns.  Possible reasons for low alkalinity measurements include passivated carbonate mineral surfaces or rapid flow rates.    Instead, Mn-bearing carbonates (e.g., rhodochrosite) and silicate minerals were suggested as the main acid-buffering minerals present in waste rock used in column experiments.  Specifically, Mn   CHAPTER 3    CHAPTER 3  102  concentrations followed a similar trend as sulphate values, with high initial concentrations (i.e., Column 1: 1000 mg∙L-1; Column 2: 1600 mg∙L-1) that decreased by two to three orders of magnitude over the duration of the experiment.   Comparison of this trend with measured pH values suggests Mn-carbonates may be sufficient to maintain Column 1 pH values above pH 6 only.   The steady, low pH values measured from Column 2 implies silicate buffering is the main acid-buffering mechanism.  Figure 3-14D, E and F present concentrations of Si, Ca and Na and, in general, concentrations are higher in Column 2, relative to Column 1, drainage.    Metal concentrations Dissolved metals are shown in Figure 3-15 and 3-16.  Overall, dissolved concentrations were 2x – 10x higher in Column 2 relative to Column 1 when both columns present pH values less than pH 6.  After day-240, or when Column 1 pH values increased to more circumneutral or neutral pHs, observed metal concentrations were strongly governed by pH conditions.  Specifically, those metals with increased mobility at neutral pHs were present at higher concentrations in Column 1, relative to Column 2, with few exceptions (i.e., Zn and Cd).  Other metals, such as Cu, Co, Cd and Zn, were present at higher values in Column 2 relative to Column 1, by approximately one order of magnitude.   3.4. Conceptual Models for the Small-scale 3.4.1. Field Barrels Drainage reports to the base of the field barrels during the wet season only and yearly evaporation is estimated to be approximately 85%.  Waste rock material used in field barrels contain particle diameters less than 3”, whereas experimental piles contain boulders up to 2 m in diameter, and the overall finer grained field barrel material is more likely to behave in a ‘soil-like’ manner and present slower overall flow velocities.  Waste rock with water that slowly infiltrates through the upper few centimeters, a zone that can be impacted by evaporation, suggest yearly evaporation totals will be higher relative to those associated with coarser grained material and therefore faster flow rates.       CHAPTER 3    CHAPTER 3  103  Although field barrel drainage volumes were recorded weekly to twice-monthly, the absence of real-time records of outflow (e.g., tipping buckets) made it difficult to determine if preferential flow paths were indeed present.   A tracer test was not conducted on these field barrels; however blasting residues (e.g., N-species) were released from field barrels at a release rate that decreased significantly with time.  Therefore, similar to experimental piles, it is likely these release rates may be used as an in situ or internal tracer to estimate matrix or preferential flow regimes at this smaller scale.   The large carbonate mineral proportion in most waste rock types at Antamina resulted in field barrel drainages containing neutral to alkaline pH conditions.  As well, concentrations were typically higher at the onset of the wet season and decreased steadily to the onset of the dry season.  This trend suggests drainage chemistries were likely governed primarily by flow regimes and mineral dissolution, with lesser sulphide oxidation, during wet season periods.  In contrast, the absence of flow during the dry season, suggests sulphide oxidation and, consequently, secondary mineral precipitation, likely occurred in the matrix pore space and contributed to the metal-rich initial flush chemistries from field barrels at the onset of the next wet season.    Elements such as Al, As, Fe, Mo, Sb and Si, that showed relatively steady concentrations within a single wet season, imply attenuation mechanisms such as secondary mineral precipitation (e.g., malachite, gibbsite, goethite, powellite) may have been present and helped to control elemental release rates.   Very few field barrels showed acidic drainages during the experiment duration and those field barrels with low pH drainages contained Class A waste rock (i.e., P2-2A, P2-3A, P5-2A, P5-6A).  Class A material contained low carbonate contents and (therefore) a much higher silicate buffering capacity relative to Class B or C (Table 3.6).  As well, Si concentrations steadily increased in Class A field barrel drainages only and complemented known relationships between silicate dissolution rates and pH (i.e., increasing dissolution with decreasing pH; Marini, 2007; White and Brantley, 1995).   A comparison of mass loadings (in mg/kg/wk) from field barrels and experimental piles is shown in Table 3.9.  Although the mixed composition of Pile 4 and 5 makes a direct correlation between the two experiments difficult, it is clear that loadings from field barrels were five times to several orders of   CHAPTER 3    CHAPTER 3  104  magnitude higher than those calculated from the larger scale experiments.  Although smaller scale experiments contain shorter flow paths than those in larger scale structures, there is an increased likelihood for slower flow velocities and less preferential flow due to a larger proportion of finer grain sizes.  Therefore water-rock interaction times will be longer at the smaller scale and mass loading rates are more likely to be higher.   3.4.2. Laboratory Columns Hydrological flow regimes from both columns were similar, as is shown by comparable flow rates, evaporation estimates and tracer breakthrough curves.  The estimated evaporation percentages from columns (42% and 49%) were also closer to those values from experimental piles (i.e., 33%) than those from field barrels (i.e., 85%)  In regards to flow, columns mimicked trends noted from field barrels with drainage reporting to the base of the columns in the wet season only; however outflow flux rates (i.e., 4 mm∙d-1 – 5 mm∙d-1) were comparable to those observed from the larger scale experimental pile (i.e., 4.5 mm∙d-1) during the wet season.   As well, tracer results suggest a small component of preferential flow was present in the columns and the majority of infiltration likely infiltrates matrix flow paths at reasonably fast velocities (i.e., 16 mm∙d-1 – 33 mm∙d-1).   Aqueous chemistries of first flush drainages from both columns were similar, with acidic pH values and high concentrations.  However, drainage from Column 1, which contains the initially heat-treated material and lower initial microbial populations, showed more neutral pH values with time.   Column 2, which had higher initial microbial populations, remained acidic and presented approximately 10x higher concentrations for most solutes analyzed (i.e., sulphate, Ca, Cd, Co, Na, Ni, Si, Zn).   This observation suggests microbial population size, and possibly its community composition, had the potential to impact drainage chemistries.   Metal(loid)s with increased mobility at neutral pH were released at higher concentrations in Column 1, relative to Column 2, which suggests pH plays an important role in drainage chemistries.     Weekly mass loading rates from columns for sulphate, As, Mo and Zn are shown in Table 3.9 along with its larger scale analog (i.e., Pile 5-Lysimeter B), which was the basis of the column design.   Column 1 mass loads were approximately 10x – 100x higher than those from Pile 5 Lysimeter B.   In regards to   CHAPTER 3    CHAPTER 3  105  Column 2, metals with increased mobility at acidic pH conditions showed an even greater magnitude of difference to estimated mass loads at the larger scale (i.e., Pile 5 Lysimeter B).   Since columns presented comparable flow regimes and contained comparable geometry and waste rock assemblages, the higher mass loading rates associated with Column 2 (relative to Column 1) implies heat treatment substantially impacted Column 1 microbial communities and is evidence of microbially-enhanced or greater weathering rates in the former.  3.5. Summary This chapter presented details of the construction, instrumentation, and composition of two types of small-scale experiments that are part of this thesis research.  Specifically, these experiments included 15 x 1 m (h) field barrels containing between 220 kg and 320 kg of waste rock and 2 x 1 m (h) columns containing approximately 170 kg of waste rock.   Field barrels experiments were located at the research site at Antamina, Peru and were exposed to similar climatic conditions as experimental piles (i.e., Pile 4 and 5).  The column experiments were constructed in the laboratory at the University of British Columbia and applied rainfall events mimicked the bimodal wet-dry climate observed at Antamina.  Field barrels were comprised of single waste rock types (i.e., Class A, B or C) whereas column experiments contained mixed waste rock (i.e., Class A and C).  The composition and structure of the column experiment was derived from a vertical cross-section through Pile 5 (i.e., Lysimeter B) and these columns reflected a smaller scale analog of a larger scale, mixed waste rock pile.   Up-scaling of soil water processes is one of the most important issues in vadose zone research (Vereeken et al., 2007) and generally involves applying properties that are calculated/observed at the small-scale to structures that are several orders of magnitude larger.   The extension of these small-scale tests to provide predictions at a larger scale is routinely accomplished by applying scaling factors; however the method is not well developed.   This thesis takes the approach that a complete assessment of the characteristics of small-scale tests would better assist its quantification of its applicability for up-scaling.  In this thesis, the four main differences between the smaller-, relative to larger-, scale experiments included:     CHAPTER 3    CHAPTER 3  106   Shorter flow paths (~ 10x);  Higher evaporation (i.e., 20% to 250%);   Smaller material volumes (i.e., 100x – 10,000x); and,   The exclusion of particles with grain diameters exceeding 10 cm. A review of these scale differences suggest preferential flow paths likely contributed to a smaller component of the flow regimes in columns and field barrels, due to the exclusion of large grain sizes and low probability for film flow or rapid flow via connected macropores.   As well, less infiltration suggests water is more likely to infiltrate the matrix pore space and follow matrix flow paths.   Smaller material volumes also promote a less heterogeneous granular matrix and mineralogical assemblage.  For the former, this may lead to fewer spatial variations in flow patterns (e.g., channeling or fingering) and lead to a smaller preferential flow component.   A less heterogeneous geochemical assemblage, coupled with shorter path lengths, suggest shorter water-rock interactions and drainage chemistries can differ considerably from those noted at the experimental pile scale.  A comparison of mass loading rates between field barrels and columns to those estimated from experimental piles (i.e., Table 3.9) show values from small-scale experiments were several orders of magnitude higher than those at a larger scale.  This is not a surprising result and has been commonly documented in the literature (e.g., Hollings et al., 2001; Malmström et al., 2000; Wagner, 2004). A complete characterization of the differences between experimental scales aimed to assess if one or more of these scale differences strongly govern observed flow and solute transport through unsaturated waste rock.  This thesis focuses on quantifying the degree of preferential to matrix flow at the small- and large- scale, which is described in Chapter 4 (columns versus experimental piles) and Chapter 5 (field barrels versus experimental piles).  In both chapters, the comparative analysis uses a flow-corrected time method (Eriksson et al., 1997) to estimate preferential to matrix flow proportions and numerical modeling to parameterize flow and solute properties.   The method developed by Eriksson et al. (1997) uses breakthrough curves from tracer tests.        CHAPTER 3    CHAPTER 3  107  Although it is not discussed in detail in this chapter, drainage from the column containing heat-treated waste rock or reduced microbial populations (i.e., Column 1) released significantly different drainage chemistries than Column 2.   Specifically, drainage from Column 2 contained lower or more acidic pH values and higher concentrations of most solutes (e.g., sulphate, Cu, Co, Cd, Ni, Zn).  This result implies the size and possibly the composition of microbial populations may have had a direct influence on drainage chemistries.  A more detailed assessment and discussion of the microbial population differences (i.e., size and community structure) between Column 1 and 2 is provided in Chapter 6.      CHAPTER 3    CHAPTER 3  108  3.6. Tables Table 3.1 Summary of Field Barrel Experiment Characteristics. Name Pit Location Excavation Date Field Barrel Start Date Waste Rock Class Rock Type Weight (kg) P2-1A 2-NP-4118-17-03 27-Aug-07 17-Oct-07 A Intrusive 220 P2-2A 2-NP-4118-21-01 03-Sept-07 17-Oct-07 A Intrusive 260 P2-3A 2-NP-4103-18-01 13-Sept-07 17-Oct-07 A Intrusive 260 P5-2A 2-NP-4028-08-01 01-Dec-08 07-Feb-09 A Intrusive 290 P5-3A 2-NP-4028-10-01 12-Dec-08 08-Feb-09 A Intrusive 310 P5-6A 2-NP-4013-06-01 04-Feb-09 08-Feb-09 A Intrusive 280 P4-2A 4-NP-4268-12-02 26-Nov-08 06-Feb-09 B Grey-café hornfels 270 P4-4A 4-NP-4268-27-05 09-Jan-08 06-Feb-09 B White marble 320 P4-4B 4-NP-4268-27-05 09-Jan-08 07-Feb-09 B Black marble 320 P4/5-1A 5-SP-4643-11-02 13-Nov-08 13-Jan-09 C Grey-café hornfels 256 P4/5-1B 4-NP-4268-03-01 14-Nov-08 14-Jan-09 C Grey-café hornfels 272 P4/5-5A 5-SP-4598-01-01 14-Jan-09 29-Mar-09 C Marble/grey-café hornfels 282 P4/5-5B* 5-SP-4598-01-01 14-Jan-09 29-Mar-09 C Marble/grey-café hornfels 274 P4-3A 5-SP-4643-10-01 06-Dec-08 06-Feb-09 C Grey-café hornfels 290 P5-4A 5-SP-4613-01-01 14-Dec-08 08-Feb-09 C Grey hornfels 284 *duplicate     CHAPTER 3    CHAPTER 3  109  Table 3.2 Waste Rock used in Laboratory-based Column Experiments. Waste Rock Class Name Proportion† Column 1 (kg) Column 2 (kg) Class A-1 P2-1A 0.50 51.9 52.6 P2-2A 0.25 P5-6A 0.25 Class A-2 P2-3A 1.00 54.9 64.9 Class C P4/5-1A 0.40 63.5 59.3 P4/5-5A 0.60 Total (kg) 170.3 176.9 † Proportion of each waste rock type contributing the mixed material used in the columns     CHAPTER 3    CHAPTER 3  110  Table 3.3 Physical Characteristics of Waste Rock used in Column Experiment. Material Bulk Density (kg·m-3) Saturated porosity (m3·m-3) Field capacity (m3·m-3) Class A-1 1503 0.22 0.17 Class A-2 1620 0.25 0.20 Class C 1575 0.29 0.05       CHAPTER 3  111  Table 3.4 Bulk Geochemical Characteristics of Pile 2 Waste Rock used in Column Experiments. Discharge Pit Location Date Material Ag (ppm) As (ppm) Bi (ppm) Co (ppm) Cu  (%) Fe  (%) Mo  (%) Pb  (%) Zn  (%) 1 2-NP-4118-17-03 25-Aug-07 A 1.15 39 7 5 n.d 1.58 0.05 0.01 n.d 2 2-NP-4118-21-01 2-Sept-07 A 3.33 41 6 7 0.004 0.01 0.009 n.d n.d 3 2-NP-4103-18-01 12-Sept-07 A 1.58 139 5 6 n.d 2.44 0.004 n.d n.d n.d = no data.         CHAPTER 3  112  Table 3.5 X-ray Fluorescence (XRF) Whole Rock Geochemistry of Antamina Waste Rock used in Field Barrel and Column Experiments. Major Oxide or Element Units Class A Class B Class C P2-1A P2-2A P2-3A P5-6A P4-4A P4/5-1A/B P4/5-5A Al2O3 % 7.58 8.69 6.71 7.31 5.90 3.90 6.37 CaO % 2.89 1.50 4.00 2.30 35.10 38.81 33.77 CrO3 % < 0.01 < 0.01 < 0.01 < 0.01 < 0.01 < 0.01 < 0.01 Fe2O3 % 1.50 2.70 10.93 3.88 2.61 1.71 2.71 K2O % 4.86 5.35 4.00 4.54 1.16 0.69 1.43 MnO % 0.06 0.06 0.10 0.05 0.06 0.06 0.06 MgO % 0.61 0.68 0.99 0.58 2.40 2.24 2.49 Na2O % 0.55 0.94 0.35 0.53 0.26 0.21 0.23 P2O5 % 0.08 0.09 0.11 0.06 0.10 0.09 0.09 SiO2 % 79.09 77.26 66.57 78.55 20.63 12.93 21.79 TiO2 % 0.27 0.28 0.28 0.25 0.27 0.21 0.30 Ag ppm < 10 < 10 18 < 10 < 10 < 10 < 10 As ppm 33 88 221 66 60 641 58 Co ppm 7 7 17 9 < 5 < 5 8 Cr ppm 45 90 49 62 107 84 108 Cu ppm 476 2361 14707 1765 194 248 55 Mn ppm 464 447 803 416 683 667 690 Mo ppm 1800 989 622 289 < 20 < 20 < 20 Ni ppm < 5 8 12 6 11 10 13 Pb ppm 109 206 138 28 293 440 160 Zn ppm 10 14 16 11 13 8 10 C % 0.10 0.22 0.02 0.08 2.20 1.95 1.90 S % 0.29 1.28 2.72 0.87 0.04 0.21 0.13         CHAPTER 3  113  Table 3.6 Mineralogical Compositions (by XRD) of Waste Rock used in Field Barrel and Column Experiments. Waste rock classification Class A Class B Class C Field Barrel Name P2-1A P2-2A P2-3A P5-2A P5-6A P4-2A P4-4A P4-4A P4/5-1A P4/5-5A/B P5-4A Actinolite Ca2(Mg,Fe)5Si8O22(OH)2 - - 1.0 - - 2.0 3.2 2.6 0.6 0.4 0.2 Albite low NaAlSi3O8 2.8 6.7 1.0 1.0 1.2 2.4 1.1 0.0 1.4 5.7 11 Andradite Ca3Fe2(SiO4)3 - 2.1 8.5 7.9 1.5 - - - - - - Anorthite CaAl2Si2O8 - - - - - 9.3 0.2 0.7 7.7 - - Biotite  K(Mg,Fe)3AlSi3O10(F,OH)2 2.8 3.0 2.0 2.3 3.1 - - - - - - Calcite CaCO3 - - - - - 61 87 92 66 82 60 Chalcopyrite CuFeS2 0.77 1.00 2.42 0.64 0.74 - - - - - - Clinochlore (Mg5Al)(AlSi3)O10(OH)8 - - - - - 2.0 0.4 0.3 3.7 - 3.0 Diopside CaMgSi2O6 1.9 - 2.7 - - 1.4 3.2 1.1 1.0 0.66 - Kaolinite Al2Si2O5(OH)4 1.0 1.6 - - - - - - - 0.39 - K-feldspar‡ KAlSi3O8 42.3 40.1 34.2 34.0 44.3 4.8 1.7 0.91 2.4 - 4.7 Muscovite  KAl2(AlSi3O10)(F,OH)2 2.2 - 0.9 - - 1.2 - 0.5 1.6 - 3.8 Magnetite Fe3O4 - - 0.7 - - - - -    Molybdenite MoS2 0.09 0.03 0.04 - - - - -    Oligoclase (Ca,Na)(Al,Si)4O8† 6.4 8.7 6.2 7.0 8.4 - - - - - - Phlogopite KMg3AlSi3O10(F,OH)2 - -  - - 4.7 0.18 0.78 3.7 4.4 4.6 Pyrite FeS2 0.46 0.73 2.73 0.49 0.72 0.31 1.2 - - 0.46 0.15 Pyrrhotite Fe1-xS§ - - - 1.0 - 1.8 0.91 0.83 1.3 0.85 2.1 Quartz SiO2 38.8 35.7 35.2 45.7 40 8.7 1.2 0.43 9.7 5.0 10 Siderite FeCO3 0.18 0.47 - - 0.06 - - - -  - Total 100 100 98 100 100 100 100 100 99 100 100 † where Ca/(Ca + Na) (% anorthite) is between 10%–30% ‡ Orthoclase § x = 0 – 0.2     CHAPTER 3  114  Table 3.7 Microbial Initial Population Enumeration via the Most Probable Number Technique. Waste rock classification Class A Class C Associated field barrels P2-1A, P2-2A, P5-6A P2-3A P4/5-1A,   P4/5-5A/B Acithiobacillus thiooxidans 98 4.4 x 103 1.58 x 102 Acidithiobacillus ferrooxidans - 2.2 x 105 46 Neutrophilic thiosulphate oxidizers 1.08 x 106 2.2 x 105 >4.8 x 108 Note: Population numbers reported as # of cells or bacteria per gram of sieved (<2 mm) material.       CHAPTER 3  115  Table 3.8 Calculated Daily Precipitation Rate and Duration for Laboratory-based Columns. Month Antamina Column  experiment  Monthly Precipitation (mm) Daily rain volume (ml) Application time  (min)† Jan  133 953 24.6 Feb 213 1690 43.6 Mar 184 1322 34.1 Apr 75.0 557 14.3 May 35.3 262 6.7 Jun 8.5 63 0.94 Jul 14.0 100 1.49 Aug 14.0 101 1.51 Sep 30.8 228 3.4 Oct 142 1022 26.3 Nov 167 1236 31.8 Dec 162 1164 30.0 † Time required for rain 5 days/week, assumes 100 ml/min flow rate and 42% increase from original measurements based on water captured from column eaves. Shaded areas denote dry season months at Antamina.        CHAPTER 3  116  Table 3.9 Field Barrel versus Experimental Pile Mass Loading Rates (mg/kg/week). Waste Rock Class Name Sulphate Alkalinity As Mo Zn Field Barrels A P2-1A 2 2 0.002 0.2 0.003 A P2-2A 4 1 0.002 0.2 0.03 A P2-3A 30 0.1 0.005 0.001 3 A P5-2A 9 1 0.006 0.1 0.002 A P5-3A 20 0.4 0.004 0.1 0.002 A P5-6A 10 1 0.004 0.1 0.002 B P4-2A 4 1 0.0001 0.0004 0.01 B P4-4A 4 1 0.0001 0.0003 0.0004 B P4-4B 150 1 0.0001 0.0003 0.0004 C P4/5-1A 130 2 0.01 0.001 0.0003 C P4/5-1B 160 2 0.0002 0.0005 0.0005 C P4/5-5A 4 2 0.3 0.0006 0.0003 C P4/5-5B* 7 2 0.3 0.0006 0.0003 C P4-3A 4 1 0.003 0.004 0.0001 C P5-4A 5 1 0.001 0.001 0.0004 Laboratory Columns† A and C Col. 1 30 - 0.001 0.0006 0.03 A and C Col. 2 60 - 0.0003 6 x 10-5 0.2 Pile 4‡ B and C Lys A 2  (0.02) 0.1 (0.0004) 0.0001 (6 x 10-7) 7 x 10-5  (6 x 10-7) 0.0009  (1 x 10-5) B and C Lys B 1  (0.03) 0.04 (0.0006) 7 x 10-6 (6 x 10-8) 2 x 10-5  (3 x 10-7) 0.0005  (2 x 10-5) C Lys C 2  (0.03) 0.09 (0.001) 0.001  (1 x 10-6) 6 x 10-5  (1 x 10-6) 0.001  (2 x 10-5) B and C Lys D 1  (0.09) 0.08 (0.002) 0.0004 (1 x 10-6) 8 x 10-5 (1 x 10-6) 0.0009  (2 x 10-5) Pile 5† A and C Lys A 2  (0.09) 0.07 (0.004) 2 x 10-5 (2 x 10-7) 0.0001 (6 x 10-7) 0.003 (0.0002) A and C Lys B 2  (0.01) 0.06 (0.0002) 0.0001 (8 x 10-7) 0.0001 (8 x 10-7) 0.0002  (1 x 10-6) A and C Lys C 0.6  (0.05) 0.02 (0.002) 3 x 10-6 (1 x 10-7) 9 x 10-6 (9 x 10-7) 0.0003  (3 x 10-5) A and C Lys D 2  (0.07) 0.1 (0.002) 3 x 10-5 (4 x 10-7) 0.002  (1 x 10-5) 0.0003  (1 x 10-5) †Laboratory columns are mimicked after Pile 5, Lysimeter B.  Mass load estimates calculated at day-300, or when pH differences between Column 1 and 2 are at the greatest. ‡Values in brackets reflect dry season average mass loads, whereas those without brackets represent average values during wet season months.      CHAPTER 3  117  3.7. Figures             Figure 3-1. Schematic of field barrel construction (A) and photographs of field barrels (B and C).  Note that photos shown in Figure 3-1B and C were taken at the first field barrel site, near the tailings storage facility (TSF).  The field barrels have since been moved to the same area as the experimental piles and are situated on a 1.5 m (in height) constructed berm.  Drainage collection bucket from volume recording and aqueous sampling 1” aeration holes drilled into side of field barrel drums A B C Waste rock < 4″ 30-mesh silica sand; h = 0.15 m 0.59 m     CHAPTER 3  118              Figure 3-2. Column experiment construction.  A: Stratification of 3 waste rock types; B:  Vaisala CO2 probe and Decagon EM50 dataloggers; C: Tipping bucket, at outflow from basal lysimeter, to record flow rate and volume; and D:  Rain event on Col. 1 column from Lechler air-atomizing nozzle.      Class A-1 Class A-2 Class C EM50 dataloggers D A Class A-1 Class A-2 Class C Silica Sand Tipping Bucket C Drainage Collection CO2 probes B EM50 dataloggerCol. 2 Col. 1 Col. 1 CR1000     CHAPTER 3  119                    Figure 3-3. Instrumentation of column experiments.  A: 5 cm basal sand layer; B: Placement of 5TE probe in Class C material; C: Sand and waste rock mixture to surround 5TE sensors and ensure contact with infiltrating waters: fine sand: < 5 mm material for 5TE sensors D: CO2 and O2 sensors in Class A-2; E, F Sub-lysimeter construction and placement.   A B D F C Apogee O2 sensor Vaisala CO2 sensor Sub-lysimeter Installation  Decagon 5TE sensor E Class A-1 Class A-2 Silica Sand 1:1:1 mixture of coarse sand, fine sand and sieved waste rock (< 5 mm) for 5TE sensors     CHAPTER 3  120   Figure 3-4. Particle size distribution of waste rock used in field barrels and column experiments. Note: PSD analyses were not conducted on:   Class A: P5-2A and P5-3A  Class B: P4-2A  Class C: P4/5-1A/B and P4-3A     01020304050607080901000.010.1110100% Passing Particle Size (mm) P4/5-5A/BP4/5-5A/BP4/5-5A/BP4/5-5A/BP5-4AP4-4AP4-4BP5-6AP2-1AP2-2AP2-3ACU (D60/D10) P4/5-5:    11 – 16 P5-4A:      7 P4-4A/B: 35 – 40 P5-6A:     125 P2-1A:     375 P2-2A:    130 P2-3A:    150      CHAPTER 3  121   Figure 3-5. Soil water characteristic curves (SWCCs) of Pile 4 and 5 field barrel materials and waste rock used in column experiments.   0.000.050.100.150.200.250.300.350.400.01 0.1 1 10 100 1000 10000 100000 1000000Volumetric Water Content (m3·m-3) Soil Suction (kPa) P4/5-5A/BP4/5-5A/BP4/5-5A/BP4/5-5A/BP5-4AP4-4AP4-4BP2-1AP2-2AP2-3AP5-6A    CHAPTER 3  122   Figure 3-6. Cumulative area-normalized outflow (m3∙m-2) from field barrels. 0.00.20.40.60.81.0Oct-07 Jun-08 Jan-09 Sep-09 May-10 Jan-11 Sep-11 May-12 Jan-13Cumulative area-normalized outflow (m3∙m-2) Pile 2 - Class A Field Barrels P2-1AP2-2AP2-3A0.00.20.40.60.81.0Jan-09 Sep-09 May-10 Jan-11 Sep-11 May-12 Jan-13Cumulative area-normalized outflow (m3∙m-2) Pile 4 - Class B and C Field Barrels P4/5-1A P4/5-1B P4-2AP4-3A P4-4A P4-4B0.00.20.40.60.81.0Jan-09 Sep-09 May-10 Jan-11 Sep-11 May-12 Jan-13Cumulative area-normalized outflow (m3∙m-2) Pile 5 - Class A and C Field Barrels P5-2A P5-3A P5-4AP5-6A P4/5-5A P4/5-5B    CHAPTER 3  123    Figure 3-7. Cumulative precipitation (dashed lines) and area-normalized outflow (solid lines) from column experiments.   0.00.51.01.52.00 100 200 300 400 500Cumulative Precipitation/Outflow (m) Time (days) Col.1Col.2    CHAPTER 3  124   Figure 3-8. Comparison of tracer breakthrough curves (A) and cumulative mass recovery (B) from column experiments.  Note: Co values are 3.1 mg/L (Col. 1), 3.2 mg/L (Col. 2); Mo values are 5.4 g (Col. 1) and 4.0 g (Col. 2). 0.000.050.100.15220 240 260 280 300 320 340 360C/Co [-] A Tracer application  0.00.20.40.60.81.0220 240 260 280 300 320 340 360M/Mo [-] Elapsed time (days) B Col.1Col.2    CHAPTER 3  125           Figure 3-9. Comparison of pH and sulphate concentrations from field barrels. 24681012Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12pH A 110100100010000100000Oct-07Oct-08Oct-09Oct-10Oct-11Oct-12Sulphate (mg/L) B P2-1A P2-2A P2-3A24681012Jan-09 Jan-10 Jan-11 Jan-12 Jan-13pH C 10100100010000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Sulphate (mg/L) D P4/5-1A P4/5-1BP4-2A P4-3AP4-4A P4-4B24681012Jan-09 Jan-10 Jan-11 Jan-12 Jan-13pH E 10100100010000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Sulphate (mg/L) F P5-2A P5-3A P5-4AP5-6A P4/5-5A P4/5-5B    CHAPTER 3  126        Figure 3-10. Comparison of alkalinity and nitrate concentrations from field barrels. 0.11101001000Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12Alkalinity (mg/L HCO3- ) A 0.010.11101001000Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12Nitrate (mg/L) B P2-1AP2-2AP2-3A1101001000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Alkalinity (mg/L HCO3- ) C 0.010.11101001000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Nitrate (mg/L) D P4/5-1A P4/5-1BP4-2A P4-3AP4-4A P4-4B1101001000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Alkalinity (mg/L HCO3- ) E P5-2A P5-3A P5-4AP5-6A P4/5-5A P4/5-5B0.010.11101001000Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Nitrate (mg/L) F P5-2A P5-3AP5-4A P5-6AP4/5-5A P4/5-5B    CHAPTER 3  127         Figure 3-11. Comparison of dissolved As and Zn concentrations from field barrels. 0.0010.010.1110100Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12As (mg/L) A 0.0010.010.11101001000Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12Zn (mg/L) B P2-1A P2-2A P2-3A0.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13As (mg/L) C 0.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Zn (mg/L) D P4/5-1A P4/5-1BP4-2A P4-3AP4-4A P4-4B0.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13As (mg/L) E 0.00010.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Zn (mg/L) F P5-2A P5-3AP5-4A P5-6AP4/5-5A P4/5-5B    CHAPTER 3  128         Figure 3-12. Comparison of dissolved Mo and Sb concentrations from field barrels.   0.00010.0010.010.1110100Oct-07Oct-08Oct-09Oct-10Oct-11Oct-12Mo (mg/L) A 0.00010.0010.010.1110Oct-07Oct-08Oct-09Oct-10Oct-11Oct-12Sb (mg/L) B P2-1AP2-2AP2-3A0.00010.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Mo (mg/L) C 0.00010.0010.010.1110Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Sb (mg/L) D P4/5-1A P4/5-1BP4-2A P4-3AP4-4A P4-4B0.00010.0010.010.1110100Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Mo (mg/L) E 0.00010.0010.010.1110Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Sb (mg/L) F P5-2A P5-3AP5-4A P5-6A    CHAPTER 3  129         Figure 3-13. Comparison of dissolved Cu and Co concentrations from field barrels.   0.0010.010.1110100100010000Oct-07 Oct-08 Oct-09 Oct-10 Oct-11 Oct-12Cu (mg/L) A 0.000010.00010.0010.010.1110Oct-07Oct-08Oct-09Oct-10Oct-11Oct-12Co (mg/L) B P2-1AP2-2AP2-3A0.0010.010.11Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Cu (mg/L) C 0.000010.00010.0010.010.1Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Co (mg/L) D P4/5-1A P4/5-1BP4-2A P4-3AP4-4A P4-4B0.0010.010.11Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Cu (mg/L) E 0.000010.00010.0010.010.1Jan-09 Jan-10 Jan-11 Jan-12 Jan-13Co (mg/L) F P5-2A P5-3AP5-4A P5-6AP4/5-5A P4/5-5BCHAPTER 3  130       Figure 3-14. pH, sulphate and dissolved concentrations (Mn, Si, Ca, and Na) from column experiments. 2345678100 200 300 400pH A 100100010000100 200 300 400Sulphate (mg/L)  B 0.0010.010.1110100100010000100 200 300 400Mn (mg/L) C 110100100 200 300 400Si (mg/L) D 0200400600800100 200 300 400Ca (mg/L) Elapsed time (days) E Col. 1Col. 20.010.1110100100 200 300 400Na (mg/L) Elapsed time (days) F Col. 1Col. 2CHAPTER 3  131        Figure 3-15. Dissolved metal(loid) concentrations (As, Cd, Mo, Sb, Se, and Zn) from column experiments. 0.0010.010.1110100 200 300 400As (mg/L) A 0.00010.0010.010.11100 200 300 400Cd (mg/L) B 0.00010.0010.010.1100 200 300 400Mo (mg/L) C 0.0010.010.1100 200 300 400Sb (mg/L) D 0.0010.010.11100 200 300 400Se (mg/L) Elapsed time (days) E Col. 1Col. 20.010.11101001000100 200 300 400Zn (mg/L) Elapsed time (days) F Col. 1Col. 2CHAPTER 3  132       Figure 3-16. Dissolved metal(loid) concentrations (Co, Cr, Cu and Ni) from column experiments.   0.0010.010.1110100 200 300 400Co (mg/L) A 0.00010.0010.010.11100 200 300 400Cr (mg/L) Elapsed time (days) B 0.11101001000100 200 300 400Cu (mg/L) C Col. 1Col. 20.0010.010.11100 200 300 400Ni (mg/L) Elapsed time (days) D Col. 1Col. 2CHAPTER 4  133  CHAPTER 4: COMPARISON OF UNSATURATED FLOW AND SOLUTE TRANSPORT THROUGH WASTE ROCK AT TWO EXPERIMENTAL SCALES USING TEMPORAL MOMENTS AND NUMERICAL MODELING 4.1. Introduction The construction of waste rock piles is part of the mining process at many mine sites.  These piles are generally large, unsaturated structures containing material with grain diameters spanning at least six orders of magnitude (i.e., 10-6 m – 1 m; Nichol et al., 2005; Fala et al., 2005).  Predictions of solute loads released at the base of a waste rock pile require the characterization of fluid flow and solute transport.  Estimating solute transport parameters for use in predictive models becomes increasingly difficult at larger scales due to the effects of spatial variability in the many factors that control the infiltration of water and the mobility of solutes (e.g., initial/boundary conditions, material properties, preferential flow paths; see Jury and Fluhler, 1992; Li and Ghodrati, 1997; Vanderborght et al., 2000; Vanderborght and Vereecken, 2007).  These complexities observed at the large scale are generally viewed as difficult, or impossible, to reproduce in smaller experiments (e.g., field barrels, laboratory columns, or humidity cells).  For example, a number of  studies have compared mineral weathering rates across scales with the intention of determining correction (or scaling) factors (Otwinowski, 1995; Eriksson et al., 1997; Malmström et al., 2000; Frostad et al., 2005).  However, the results reveal large discrepancies in behaviour among scales, which suggest large field experiments may be necessary to adequately characterize system response (Malmström et al., 2000).   The material characteristics of waste rock can have a substantial impact on its hydrological response (Smith and Beckie, 2003). This observation is supported by the study of Buchter et al. (1995), who state that in stony soil the flow path of water is an intrinsic property of the soil medium (at a given water content).  Flow through waste rock can be categorized into two types; preferential flow through either open voids and/or the coarse-sized fraction of the matrix, and matrix flow through finer-grained (or matrix) materials.  Preferential flow paths can be expressed as channelized flow whereby the water flux is concentrated into spatially distinct areas smaller than the total cross-sectional flow area (Nichol et al., 2005).  In this paper, the term matrix flow describes the movement of water through the granular material CHAPTER 4  134  present between the cobbles and boulders and is dictated by moisture contents, capillary tension and gravity.   The occurrence of some degree of preferential flow in unsaturated porous media has long thought to be the rule rather than the exception (e.g., Brusseau and Rao, 1990; Flury et al., 1994).  In general, three factors exert a strong control on the degree of preferential flow in waste rock; the infiltration rate, the spatial arrangement of matrix-supported and matrix-free zones, and the particle size distribution (PSD).  The first factor is dependent upon the local climate and is not pertinent to this study as infiltration conditions are similar for all experiments reported here.  For the second factor, Herasymuik (1996) observed that the method of construction controls the spatial orientation of material within the pile.   For example, waste rock piles that are created by end-dumping may result in a fining-upwards gradation, as coarser material falls to the base of the lift and finer materials generally remain near the surface (Fala et al., 2005).  An alternative is push-dumping, whereby waste rock is deposited in a series of lifts starting at the top of a pile and ‘pushing’ the material to a level surface.  This method generally results in a coarse lower zone and a non-uniform upper zone with horizontal traffic surfaces between lifts (Corazao Gallegos, 2007).   The PSD of waste rock is a function of its mineralogical composition, material friability, mine-specific blasting techniques, among other factors.  PSD analyses generally encompass size fractions that range over several orders of magnitude and, regardless of the coarse proportions, typically exhibit a long tail at the finer size fractions (Smith and Beckie, 2003).   Structured porous media exhibiting flow heterogeneities, or a combination of preferential and matrix flow paths, are frequently described by multi-domain models (Gerke and van Genuchten, 1993a).  In the simplest multi-domain model (i.e., dual-porosity), water/solute are partitioned between a mobile and an immobile domain.  Using this approach, the mobile domain constitutes preferential flow and matrix flow paths, and the immobile domain includes stagnant (or non-flowing) waters.  The immobile domain is a dynamic region whereby water and/or solute may transfer in and out of this domain, as a result of pressure or concentration gradients (Šimůnek and van Genuchten, 2008).  Model formulations using a fixed water content in the immobile domain and solute only transfer between the immobile and mobile CHAPTER 4  135  domains represent the most basic dual-porosity approach and are termed mobile-immobile (MIM) models (e.g., van Genuchten and Wierenga, 1976).  A dual-permeability model also assumes two overlapping pore domains, like the dual-porosity approach, but replaces the immobile domain with a low-permeability domain, which allows active, albeit slow water movement (Gerke and van Genuchten, 1993a, b; Šimůnek and van Genuchten, 2008).   Water and solute transport in a dual-permeability model can occur within and between mobile and immobile domains.   In this study, flow and tracer response are examined at two experimental scales, in 0.8 m tall laboratory columns and in a 10 m high experimental waste rock pile constructed using the end-dumping method.  This study is akin to previous work in that it compares flow and conservative solute transport at two scales (Strömberg and Banwart, 1994, 1999a, 1999b; Eriksson et al., 1997), but differs in its effort to maintain similar compositions and proportions of the different waste rock types at both scales.  The objective of this research is two-fold: 1) to determine the main flow regimes controlling solute transport in these waste rock assemblages under site-specific and laboratory conditions, and 2) to determine if the principal influences on solute transport linked to waste rock flow regimes in larger test piles can be adequately reproduced at more practicable scales by managing ‘controllable’ variables (i.e., material compositions and precipitation inputs).  Both experiments share the same rock type and the rainfall applied to the laboratory columns mimics the daily rainfall recorded at the experimental pile.  Flow is recorded from basal lysimeters and a tracer test is used to characterize the partitioning of infiltration among different flow paths.  Tracer results are analyzed using a flow-corrected time approach (Eriksson et al., 1997) and numerical modeling.  The flow-corrected time approach has been employed in other studies with transient flow behaviour (e.g., Wierenga, 1977; Jury, 1982; Small and Mular, 1987; Destouni, 1991).  Numerical modeling aimed to apply the simplest modeling approach (i.e., the MIM model) to permit the diagnosis of dominant features of solute transport processes, while minimizing the number of model parameters. CHAPTER 4  136  4.2. Materials and Methods 4.2.1. Site Description The experimental field site is at the Antamina Mine, located in north-central Peru (9o 32 S and 77o 03 W), approximately 270 km north of Lima, Peru.  The average annual temperature and precipitation range between 5.5 – 6.0 OC, and 1200 – 1300 mm, respectively.  Climate at Antamina is governed by the tropical rain belt, which creates distinct ‘wet’ and ‘dry’ seasons.  Wet seasons occur between October and April, with the remainder of the year described as the dry season.  Approximately 80 - 90% of the total annual precipitation at Antamina occurs during wet season months.   Antamina divides its waste rock into three main classes based on geochemical parameters (Class A, B and C; Table 4.1).  Class A and C material represent end members, with the former considered reactive and the latter less-reactive; Class B material is intermediate between the two.  The majority of Antamina waste rock is carbonate-rich and contains variable sulfide mineral contents (i.e., >3 % Class A; <3 % Class B and C) and minor ore minerals (e.g., chalcopyrite, molybdenite, sphalerite) (Antamina, 2007). 4.2.2. Experimental Piles From 2006 – 2009, five experimental piles were built at Antamina for the purpose of understanding controls on waste rock leachate quality and quantity.  Each pile is 36 m (l) x 36 m (w) x 10 m (h) and contains between 19,000 to 25,000 tonnes of material (Figure 4-1A).  Piles were constructed through multiple tips (or tipping phases) of end-dumped material producing approximately 37o slopes. The focus of this study is on Pile 5, which contains ~19,000 tonnes of Class A and C waste rock.   Details of the construction of all five piles can be found in Corazao Gallegos (2007) and Bay (2009).   Pile 5 consists of six unique tipping phases, each containing 2,600 – 4,500 tonnes of Class A intrusive or Class C hornfels/marble material waste rock.  The ratio of Class A to C material in Pile 5 is ~ 9:11 (as wt%:wt%),  (Figure 4-1A).  Particle size distribution (ASTM D 5519-94) and bulk density (ASTM D 5030-89) were analyzed at the mine site (Golder, 2010).  Tests were completed on samples consisting of 4 – 8 tonnes of waste rock.  Averaged PSD curves of samples relevant to this study are shown in Figure 4-2.  Generally, a typical PSD curve of waste rock shows a well-graded distribution with a coefficient of CHAPTER 4  137  uniformity Cu (=D60/D10) of 20 or higher (Morin et al., 1991; McKeowen et al., 2000).  Calculated Cu values present very different physical properties for Class A and C materials.  Class A material shows high Cu values (i.e., 257) and a well-graded distribution whereas Class C materials contain low Cu values (i.e., 12) and a poorly graded distribution.  The footprint of Pile 5 is underlain by a 75-mil geomembrane liner serving as a basal lysimeter.  Within this large basal lysimeter are 3 smaller (4 m x 4 m) geomembrane liners located along the centerline of the pile.  These smaller catchments act as sub-lysimeters to characterize flow or infiltration through a smaller cross-sectional area within the pile (Figure 4-1A).  These smaller sub-lysimeters and the large basal lysimeter are referred to as Lysimeter A – C and Lysimeter D, respectively (Figure 4-1A) Drainage is channeled from the lysimeters to an instrumentation hut that records flow rate/volume, electrical conductivity, and temperature.  The experiment was initiated in the dry season and corresponds to the first day of recorded data (i.e., June 1, 2009).  The instrumentation hut includes sampling ports for analysis of drainage water quality.   Each experimental pile contains six instrumentation lines with multiple sensors to record in situ temperature, electrical conductivity, gas concentrations and water content.  Sensors recording in situ moisture contents (θ) [L3∙L-3] are used to determine initial conditions of waste rock material.  Data from the remaining instruments (i.e., conductivity, temperature and gas concentrations) are not discussed in this study, but further details may be found in Bay et al. (2009) and Peterson et al. (2012).   This study focuses on flow and transport to Lysimeter B of Pile 5.  Lysimeter B is located along the centerline of the pile and encompasses material from the first 3 end-dumping events used to construct the pile.  In total, approximately 430 tonnes of waste rock overlies the 4 m (w) x 4 m (l) footprint of Lysimeter B in a 1:4 ratio of Class C material underlying Class A materials.  Class C material was used in the first end-dumping event, whereas the second and third events contained Class A material. The placement of the two Class A materials were separated temporally by the installation of instrumentation line 1 (Figure 4-1A).  Both Class A materials were excavated from the same location within the Antamina open pit but at 9 days apart.  As such, these materials are assumed to be similar, and are not differentiated in this study.  CHAPTER 4  138  A lithium bromide tracer with 3050 mg·L-1 of Br- was applied on Pile 5 on January 24th, 2010 or 237 days following experiment initiation.  Bromide was selected as the tracer due to its low background concentrations at Antamina (i.e., 0.2 mg·L-1 – 0.8 mg·L-1) and its conservative nature.  Bromide was applied to the crown of the pile through a single applied rainfall event of 24 mm and 5 hours duration (4.8 mm∙hr-1).  The application rate corresponds to a heavy precipitation event with a 4-year return period at the Antamina mine.  The Christiansen uniformity coefficient value (CUC; Christiansen, 1942) of the tracer application was determined to be 77, which confirms a moderate to even tracer application across the pile crown.  Observed ponding was minimal and no surficial runoff from the pile occurred during tracer application. Bromide concentrations from water samples were measured using a spectrophotometric method (modified from Presley (1971)).  One of every 25 sample analyses was repeated for quality assurance.  For a brief time period when tracer samples were inadvertently not collected, bromide concentrations were estimated by correlation with dissolved lithium concentrations, which is a weakly sorbing cation that was co-applied with bromide.  Lithium concentrations were obtained from routine weekly water-quality samples collected and analyzed by the mine.  This study’s approach assumes vertical flow from the pile’s upper surface to the free-draining sub-lysimeter B.  Despite significant material property differences between Class A and Class C material, Blackmore et al. (2012) showed Pile 5 is dominated by vertical flow regimes with a negligible horizontal flow component.  Specifically, measured tracer concentrations from Lysimeter C, outside of the tracer application area, were at or below detection for the majority of the experiment and released an insignificant (i.e., 0.03%) amount of applied tracer mass over the duration of the experiment. 4.2.3. Laboratory Columns The smaller scale column experiments were designed to closely resemble the material assemblage and structure of Pile 5-Lysimeter B.  Class A and C materials for the column study were transported to the University of British Columbia (UBC, Vancouver, BC, Canada).  Class A material was comprised of two subsets, A-1 and A-2.   Class A-1 is comprised of Class A waste rock from three different regions of Antamina’s open pit, which includes material used in Pile 5.  Class A-2 is composed of material from a CHAPTER 4  139  single location and is more physically and geochemically homogeneous than Class A-1.  Although these materials are not an exact replica of waste rock overlying Lysimeter B in Pile 5, they exhibit the range of physical heterogeneity within the Class A waste rock classification, analogous to the materials used in the larger scale experiment.  The Class C material used in the columns was collected from 2 tipping phases during construction of the pile and is representative of the majority of the Class C material found in Pile 5. Waste rock encompassing each of the three subsets (Class A-1, Class A-2 and Class C) was combined individually and cone-and-quartered (Kellagher and Flanagan, 1956) prior to placement in two 0.8 m (h) x 0.45 m (d) columns (Figure 4-1B).  One part Class C underlies two parts Class A, which is equally divided between the two Class A samples, similar in design to the layering in Pile 5.  A 5 cm layer of silica sand was placed at the base of the waste rock mixture to promote drainage to the basal outlet.  The two columns are referred to as Column 1 and Column 2 and each contain ~ 170 kg of waste rock.   Waste rock was placed in each column and the construction and placement of material in each column required caution and care to: reproduce the gross internal structure in terms of layering Class A over Class C materials, ensure a random representation of grain sizes from each material type, avoid disruption of surrounding materials and prevent damage to instrument installations.  Eight sensors (Decagon ECH2O probes) are located within each column to record moisture content (θ) [L3∙L-3].  These sensors require an adequate hydraulic contact with the surrounding soil and are sensitive to air gaps and soil disturbance.  Accordingly, the moisture content sensors were placed in a 1:1 ratio of silica sand to sieved waste rock to ensure infiltrating waters contacted the sensors.  A grain size limitation is imparted by the column dimensions and material is restricted to cobbles with diameters less than 10 cm.  PSD curves for Class A and C material used in column experiments are truncated and normalized (i.e., 10 cm = 100 % passing) in order to reflect grain sizes used at this scale and enable a better comparison with field scale PSD results (Figure 4-2).  Class A and C materials showed comparable Cu values to those calculated from pile curves (i.e., Class A: 210 and 133 (Columns - Class A1 and A2, respectively) versus 257 (Pile 5); Class C: 11 (Columns) versus 12 (Pile 5). These results indicate the physical properties of Class A and C waste rock are comparable at both experimental scales, despite a grain diameter constraint imposed for the columns. CHAPTER 4  140  Porosity and field capacity measurements were made in the laboratory by completely saturating oven-dried material to obtain water content at saturation θsat [L3·L-3], which was assumed to equal porosity, and allowing this water to drain to estimate its field capacity value [L3·L-3] (Table 4.2). Distinct PSD curves were generated for Class A-1 (n = 4) and A-2 (n = 1) materials and averaged PSD curves of Class C (n = 4) are applicable to both the experimental pile and column experiment (Figure 4-2).  The percentage of material less than 2 mm in Class A and C material is 23% and 4%, respectively, implying that Class A waste rock is considerably finer-grained.  Each column contains a basal lysimeter, which channels all drainage to a tipping bucket for recording flow rate/volume and allows for collection of aqueous samples (Figure 4-1B).  Precipitation events were applied using an air-atomizing spray nozzle (Lechler Airmist®  Series 136.2), combining compressed air (at 100 psi) and flow rates of 100 mL·min-1  (377 mm∙hr-1).  Rainfall was applied once daily and five times weekly to columns for durations of 40 seconds to 35 minutes.  This range mimics dry season and wet season precipitation amounts (respectively) observed at Antamina over the June 2009 – May 2010 water year.  Applied rainfall yielded a moderate uniformity coefficient (CUC value = 67).  To avoid infiltration of precipitation along the column sides, a small eave was placed along the inner diameter of both columns (see Figure 4-1B).  Eave catchment water was funneled to a collection bottle for volumetric measurement and corrected applied precipitation volumes reflect approximately 70 – 80% of initial applied volumes (i.e., 250 – 300 mm∙hr-1).        Similar to the pile, a tracer was applied in a single rain event of 10 mm over 0.2 hours on both Column 1 and Column 2 at the height of the wet season (i.e., t = 241 days).  Bromide, as LiBr, was also used in the column experiments (Co = 3100 mg·L-1 Br-) and leachate concentrations were analyzed using the same spectrophotometric method as described for the pile tracer test.   4.2.4. Parameter Estimation Eriksson et al. (1997) investigated the existence of preferential flow in waste rock at the Aitik mine site, Sweden using a flow-corrected time (τ) approach.  This method accounts for temporal variability in flow CHAPTER 4  141  and allows for the comparison of breakthrough curves through their temporal moments, from multiple experiments at different scales.  The following is a brief discussion of the flow-corrected methodology.  Flow-corrected time (τ) reflects time as a relative proportion of the total flow exiting a boundary relative to the duration of the tracer experiment and related to the time lapsed following tracer application (Equation 4-1):   𝜏 ≡ ϒ𝑉(𝑡)𝑉𝑡𝑜𝑡  (4-1) where  ϒ [T] is the experimental tracer-test period, 𝑉(𝑡) [L3] is the cumulative flow up to time t and 𝑉𝑡𝑜𝑡 [L3] is the total flow volume for the experiment duration.  Breakthrough curves represented as functions of τ are therefore stretched out during periods of high flow and compressed during low-flow periods (relative to real-time curves).   The expression for the first temporal moment (𝜏̅) [T], or the arrival time of the center of tracer solute mass at depth z [L], is; 𝜏̅(𝑧) =  ∫ 𝜏𝑆(𝜏, 𝑧)𝜕𝜏∞0∫ 𝑆(𝜏, 𝑧)𝜕𝜏∞0   (4-2) In this equation (S) [M·T-1] is S(τ,z) = 𝑐(𝜏, 𝑧)?̃? where 𝑐(𝜏, 𝑧) [M·L-3] is the measured solute concentration at time 𝜏 and a vertical travel distance z and ?̃? [L3·T-1] = 𝑉𝑡𝑜𝑡/ϒ is the steady-state water flow that corresponds to flow-corrected time, τ.  In this analysis, the first moment is evaluated at the base of the pile or column (z = 10 or 0.8, respectively).  Assuming the conservative tracer is unaffected by reactions and mass diffusion into immobile domains (Eriksson et al., 1997), the 𝜏̅ value can be used to calculate the apparent mobile water content 𝜃𝑚∗  [L3·L-3],   𝜃𝑚∗ =𝜏̅?̃?𝐴𝑧 (4-3) where 𝐴 [L2] is the discharge area of drainage and z [L] is the travel distance.  CHAPTER 4  142  The degree of preferential flow (𝑣; [-]) can be assessed by comparing the apparent mobile water content to the actual mobile content values 𝜃𝑎𝑐𝑡𝑢𝑎𝑙 [L3·L-3], The actual mobile water content is assumed to equal the measured water content 𝜃𝑚𝑒𝑎𝑠 (from sensors) and measured water contents may be comprised of both immobile and mobile water domains (i.e., 𝜃𝑚𝑒𝑎𝑠 = 𝜃𝑚 + 𝜃𝑖𝑚).  However, the Eriksson et al. (1997) approach assumes no transfer between the immobile and mobile domains or the absence of an immobile domain (i.e., 𝜃𝑎𝑐𝑡𝑢𝑎𝑙 ≈ 𝜃𝑚𝑒𝑎𝑠 = 𝜃𝑚 + 𝜃𝑖𝑚; if 𝜃𝑖𝑚 ≈ 0 or negligible transfer) and is conceptualized as a uniform flow model.  Indeed, in the presence of immobile porosity, the assumption that measured water contents equal actual water contents (to be used in Equation 4-4) results in an underestimation of 𝑣 values.  Therefore, the ratio of ‘apparent to actual mobile water contents’ (i.e., 𝑣) can be used as an indicator of preferentially flowing water and/or an immobile domain.  A low ratio suggests that a system may be better represented by a dual-domain approach.  The spreading of the breakthrough curves around the mean arrival time (𝜏̅) is quantified by the second temporal moment (στ) [T] (Eriksson et al., 1997), as; Relative spreading around the mean arrival time can be expressed by the normalized second moment (𝐶𝑉𝜏) [T], which allows spreading to be compared between experiments of different scales (e.g., columns to Pile 5).   𝑣 =𝜃𝑚∗𝜃𝑎𝑐𝑡𝑢𝑎𝑙 (4-4) 𝜎𝜏 = √∫(𝜏 − 𝜏̅)2𝑆(𝜏, 𝑧)𝜕𝜏∫ 𝑆(𝜏, 𝑧)𝜕𝜏∞0∞0 (4-5) 𝐶𝑉𝜏 = 𝜎𝜏𝜏̅ (4-6) CHAPTER 4  143  A moderate to high amount of spreading coupled with a low ratio of apparent to measured mobile water contents (or 𝑣 << 1; Equation 4-4), is suggestive of the existence of preferential flow paths and/or diffusive mass-transfer between mobile and immobile domains (assuming measured moisture contents significantly over-estimate actual mobile water contents).   Finally, Eriksson et al. (1997) estimate the local longitudinal dispersivity 𝜆 [L] by fitting an advection-dispersion equation (ADE) (adapted from Kreft and Zuber, 1978) with a time-averaged velocity to the measured flow-corrected breakthrough curve:   𝑆(𝜏, 𝑧)𝑀𝑜= (  1𝜏̅ [4𝜋 (λz) (τ𝜏̅)3]12)  𝑒𝑥𝑝 [−(1 −𝜏𝜏̅)24 (λ𝑧) (𝜏𝜏̅)] (4-7) where 𝑀𝑜 [M] is the applied tracer mass; the dispersion coefficient 𝐷 [L2·T-1] (from Kreft and Zuber, 1978) is replaced by 𝐷 =  𝜆?̅?, and ?̅? is the time-averaged pore water velocity [L·T-1] (Destouni et al., 1994).  Three assumptions are associated with the use of Equation 4-7: the observed tracer spreading is due only to local hydrodynamic dispersion or a Fickian model of transport (Roberts et al., 1986) and not diffusional solute exchange to immobile domains, the model fits assume a constant velocity and dispersion, and the use of first temporal moment values implicitly accounts for potential early tracer arrival due to preferential flow (Eriksson et al., 1997).   The flow-corrected time approach and application of Equations 4-1 through 4-7 define the first and second temporal moment (Equation 4-2 and 4-5 or 4-6, respectively), assist in identifying the existence of preferential flow paths (Equation 4-3 and 4-4), and characterizing the dispersivity of the system, assuming no solute transfer to the immobile domain (Equation 4-7).   4.2.5. Numerical Simulations Flow and solute (i.e., bromide) transport are modeled using HYDRUS1D, which includes several formulations for variably saturated flow in one dimension including single and dual domain approaches (Šimůnek and van Genuchten, 2008).  For the former, variably saturated flow through a uniform, or CHAPTER 4  144  single-porosity, system is described by the Richards’ equation (Equation 4-8, for one-dimensional systems; Richards, 1931; Šimůnek et al., 1998). 𝜕𝜃(ℎ)𝜕𝑡=𝜕𝜕𝑧𝐾(ℎ) (𝜕ℎ𝜕𝑧+ 1) (4-8) where h is the pressure head [L], θ is the volumetric water content [L3·L-3], t is time [T], z is the spatial coordinate [L], and K(h) is the unsaturated hydraulic conductivity as a function of h [L·T-1].   Transport of a conservative solute is described by the classical advection-dispersion equation (Equation 4-9): 𝜕𝜃𝑐𝜕𝑡=𝜕𝜕𝑧(𝜃𝐷𝜕𝑐𝜕𝑧) −𝜕𝑞𝑐𝜕𝑧 (4-9) where c is the solution concentration [M·L-3], D is the dispersion coefficient accounting for hydrodynamic dispersion [L2·T-1], and q is the specific discharge [L·T-1].  Dual-porosity (including the MIM formulation) and dual permeability are two examples of a dual domain approach that may be modeled using HYDRUS1D (Šimůnek and van Genuchten, 2008).  The selection of an appropriate model to describe solute transport through a porous medium is difficult to establish a priori (Roth et al., 1991).  In an effort to maintain simplicity, the dual-porosity MIM model is selected as the applied dual-domain approach for model simulations in this study.     The dual-porosity approach assumes the total water content 𝜃 of a system can be partitioned into its mobile 𝜃𝑚𝑜 and immobile 𝜃𝑖𝑚 components (Equation 4-10).   𝜃 =  𝜃𝑚𝑜 + 𝜃𝑖𝑚 (4-10) In a dual-porosity model, either water and solute, or solute only, may transfer between the mobile and immobile domain.  The former assumes flow and transport is restricted to the mobile domain, and both may transfer between the mobile and immobile porosities (van Genuchten and Wierenga, 1976; Šimůnek and van Genuchten, 2008).  The latter, solute only transfer approach, is better described as a mobile-immobile (MIM) model in HYDRUS1D and assumes uniform water flow in the mobile domain, a constant immobile water content, and solute exchange between the two domains controlled by a mass-transfer coefficient (𝜔𝑚𝑖𝑚; [T-1]).   CHAPTER 4  145  Two equations are required to describe water flow in the mobile domain (Equation 4-11) and moisture dynamics in the immobile domain (Equation 4-12) (Šimůnek et al., 2003):  𝜕𝜃𝑚𝑜(ℎ𝑚𝑜)𝜕𝑡=𝜕𝜕𝑧𝐾(ℎ𝑚𝑜) (𝜕ℎ𝑚𝑜𝜕𝑧+ 1) − Г𝑤 (4-11) 𝜕𝜃𝑖𝑚(ℎ𝑖𝑚)𝜕𝑡=  Г𝑤 (4-12) where Г𝑤  [T-1] is the mass-transfer rate for water between the mobile and immobile domains.  Refer to Šimůnek et al. (2003) and Kӧhne et al. (2004) for alternative formulations of the mass-transfer rate.  The dual-porosity conservative solute transport model adapts the standard advection-dispersion equation (Equation 4-9) to represent solute exchange between the two domains as a sum of an apparent first-order diffusion and dispersion process and advective transport (where applicable).  Dual-porosity solute transport for a conservative solute is described below, with Equation 4-13a describing solute transport in the mobile domain, Equation 4-13b is the mass balance for the immobile domain, and Equation 4-13c is the transfer of solute mass between the mobile and immobile domains.  𝜕𝜃𝑚𝑜(𝑐𝑚𝑜)𝜕𝑡=𝜕𝜕𝑧(𝜃𝑚𝑜𝐷𝑚𝑜𝜕𝑐𝑚𝑜𝜕𝑧) −𝜕𝑞𝑚𝑜𝑐𝑚𝑜𝜕𝑧− Г𝑠 (4-13a) 𝜕𝜃𝑖𝑚(𝑐𝑖𝑚)𝜕𝑡= Г𝑠 (4-13b) Г𝑠 = 𝜔𝑚𝑖𝑚(𝑐𝑚𝑜 − 𝑐𝑖𝑚) +  Г𝑤𝑐∗ (4-13c) where 𝑐𝑚𝑜 and 𝑐𝑖𝑚 are solute concentrations of the mobile and immobile regions [M·L-3], respectively; 𝐷𝑚𝑜 is the dispersion coefficient in the mobile region [L2·T-1], 𝑞𝑚𝑜 is the specific discharge in the mobile region [L·T-1], 𝜔𝑚𝑖𝑚  is the mass-transfer coefficient [T-1], and Г𝑠 is the mass-transfer term for solutes between mobile and immobile regions [M·L-3·T-1].  In the dual-porosity model, 𝑐∗ is equal to 𝑐𝑚𝑜 or 𝑐𝑖𝑚 for Г𝑤 > 0 or Г𝑤 < 0, respectively).  The dual-porosity MIM approach used in this study applies the uniform flow equation (Equation 4-8), where 𝜃 refers to the 𝜃𝑚 (or the mobile water content), and solute transport shown in Equation 4-13.  In the latter, 𝛤𝑤 is zero as solute transfers between immobile and mobile CHAPTER 4  146  porosities are restricted to diffusive mass transfer controlled by a mass-transfer coefficient (𝜔𝑚𝑖𝑚) and the concentration gradient between the two domains. 4.2.5.1. Model discretization and parameterization Model simulations in this study assume vertical flow, therefore Pile 5 simulations are conceptualized as a 10 m column consisting of 2 material types; 8 m of Class A (emplaced during two separate discharge events) overlying 2 m of Class C.  The one-dimensional profile is discretized into 100 elements from the top surface of the pile to a lower seepage-face boundary (i.e., pressure head = atmospheric), with a uniform grid block size of 10 cm.  Evaporation and subsequent infiltration at the upper surface of the experimental pile was computed from observed site precipitation (see Appendix C) and the Hargreaves method was used for estimating evaporation (Jensen et al., 1997).  The Hargreaves method was selected for two reasons; a water balance approach would add uncertainty by assuming the absence of non-vertical flow beyond the areal footprint (4 m x 4 m) of Lysimeter B and the Hargreaves evaporation estimation approach involves fewer input parameters (i.e., minimum/maximum temperatures and latitude) than the Penman-Monteith method (i.e., net radiation, soil heat flux, wind speed, temperature, atmospheric pressure, relative humidity).  Temperatures are measured hourly at a weather station, near the experimental pile site.  Mean night-time and day-time temperatures are assumed to represent robust estimates of the minimum and maximum daily temperatures, respectively, required for the Hargreaves evaporation equation.  In the model, precipitation and evaporation are applied daily and a single instantaneous pulse of bromide is applied, representing the timing of the tracer test at the height of its first wet season.  The model simulates flow and solute transport for the first 804 days following completion of construction, with minimum and maximum time steps of 0.005 d and 1 d, respectively.  Daily input values for Pile 5 simulations can be found in Appendix C. The smaller scale columns are described as 1D-vertical domains, 0.8 m in length, consisting of 3 materials; 2 x Class A (as Class A-1 and Class A-2) and Class C, all in equal proportions.  The domain was discretized into 100 elements, with grid blocks of 8 mm, and a flux condition is assigned to the upper boundary.  A seepage-face (i.e., h = 0) is applied to the lower boundary, which assumes the boundary CHAPTER 4  147  flux will remain zero with a negative pressure head and outflow is triggered by the saturation of the lower soil profile.  Daily precipitation and evaporation values were applied to column simulations.  The former reflect area-normalized applied precipitation volumes.  Evaporation was estimated using a water balance method (i.e., E = P – Q), by applying two assumptions.  Firstly, all recorded outflow is assumed to be vertical and constrained by the spatial boundaries of the column geometry.   Secondly, daily precipitation or rate of infiltration is month-dependent (to reflect seasonal changes at Antamina) and assumed to be constant within the days of that month.  Therefore, the largest uncertainties with evaporation estimates occur at the first days of a month or when there is significant change in the daily applied precipitation rates.   A single pulse of bromide was applied on day 241, to represent the start of the tracer test, and flow and solute transport was simulated for 372 days with similar minimum and maximum time steps as used in the pile simulations.  Daily input values for column simulations can be found in Appendix C. Six physical parameters are required to describe the matrix properties of Class A and C material used in the pile and the laboratory columns; residual water content 𝜃𝑟 [L3·L-3], saturated water content 𝜃𝑠 [L3·L-3], dry bulk density 𝜌𝑏 [M·L-3], saturated hydraulic conductivity 𝐾𝑠𝑎𝑡 [L·T-1], and two van Genuchten parameters 𝛼 [L] and 𝑛 [-].   Pile 5 was constructed over a period of 6 – 8 months and the material had a significant ‘wet-up’ time.  Estimates of water content 𝜃𝑚 [L3·L-3] for pile simulations were based on measurement ranges from in situ volumetric water content sensors (i.e., 0.18 – 0.36 m3∙m-3) following the 2009 wet season (i.e., May – June 2009).  Materials used for the laboratory columns were flushed and drained prior to construction to measure material-specific soil parameter input values (i.e., 𝜃𝑟, 𝜃𝑠) described in previous section (Table 4.2).  A material’s residual water content value reflects the minimum water content value obtained at sufficiently high suction pressures (i.e., 106 kPa).  For this parameter, field capacity values were used as model inputs as it was assumed these values would not significantly differ from residual water contents. Therefore, initial water contents were assumed to reflect residual water content values.  Initial estimates for dispersivity 𝜆 [L] values were obtained by fitting flow-corrected data to an ADE (Equation 4-7), using CHAPTER 4  148  the method described previously by Eriksson et al. (1997), and are to be further refined in model simulations.   The van Genuchten parameters were obtained using a four-step procedure; 1) renormalizing PSD results to material passing the #4 sieve (i.e., < 4.75 mm); 2) apportioning the renormalized volumes to sand-silt-clay fractions (i.e., 4.75 mm > sand > 0.42 mm > silt > 0.074 mm > clay); 3) applying these percentages to the USDA soil classification triangle (Brown, 2003; see Appendix C); and 4) referencing a table developed by Carsel and Parrish (1988) describing typical van Genuchten parameters for the 12 USDA-classified soil types.  Reference to this table provided estimates for the shape parameters 𝛼 and 𝑛 (Equation 4-14), and saturated hydraulic conductivity 𝐾𝑠𝑎𝑡 values, which are used as initial input values for the simulations at both scales.   𝜃(ℎ) =  𝜃𝑟 +(𝜃𝑠 − 𝜃𝑟)(1 + |𝛼ℎ|𝑛)𝑚 (4-14) The #4 sieve size, described in step 1, was selected based on studies by Yazdani et al. (2000), which found porous media containing median grain diameters greater than 4.75 mm do not have significant capillarity in unsaturated conditions.  It was inferred that material passing this sieve size actively contributes to the mobile and immobile domains hosting matrix flow paths and some preferential flow paths.  Preferential flow related to non-capillary or film flow may be related to pore spaces generated by particles with greater than 4.75 mm.  Final van Genuchten parameters (Table 4.2) used in the flow simulations were manually calibrated by matching simulated to measured outflow.  Initially estimated Class A parameters, from laboratory measured and tables by Carsel and Parrish (1988), are very similar to final model-calibrated values (shown in Appendix C).  Final 𝐾𝑠𝑎𝑡 values (3.0 m·d-1) for Class C were more closely related to the upper end of its soil classification (i.e., sandy loam (1 m·d-1).  This result is likely attributed to the significantly lower Class C component passing the 4.75 mm sieve.  Note that saturated porosity estimates used in model simulations were obtained from values measured in the laboratory, not from typical soil property tables (i.e., 𝜃𝑠𝑎𝑡= 0.39 – 0.45). Dual-porosity models require estimates of immobile porosity 𝜃𝑖𝑚  [L3·L-3] and mass-transfer rates 𝜔𝑚𝑖𝑚  [T-1].  Immobile porosity values are defined as the portion of matrix material porosity where water is CHAPTER 4  149  effectively stagnant.  Initial estimates of immobile porosity were assumed to be comparable to laboratory-measured field capacity values that are used as input values for residual water contents (shown in Table 4.2), but were refined with model calibration.  It is important that residual moisture content values reflect the lowest possible immobile water content values.  The mass-transfer coefficient 𝜔𝑚𝑖𝑚 was estimated by fitting simulated to measured results.  Uniform flow and solute transport simulations were also carried out using the Richards’ equation (Richards, 1931), for the purposes of comparing results from dual-porosity to single porosity model approaches.  Input values for single porosity simulations were identical to those used in dual-porosity simulations (i.e., 𝜃𝑠𝑎𝑡, 𝜃𝑟, 𝜆, 𝐾𝑑).   4.3. Results 4.3.1. Measured Flow and Bromide Breakthrough Flow monitoring for Pile 5 and the laboratory columns began in the dry season.  In Figure 4-3A, the wet season (October – April) is reflected in rapid increases to pile cumulative outflow (i.e., t = 150 – 300 d and 500 – 700 d), with significantly less flow and low-flow plateaus in the dry seasons (i.e., May – September or t = 0 – 150 d, 300 – 500 d, and >700 d).  Successful efforts to mimic the Antamina recharge cycle in the column experiments are supported by the similarity between column and pile-recorded outflows over the total experimental time (Figure 4-3A).  However, unlike the experimental pile that drains year-round, the column drainage occurs almost exclusively in wet season months.   Initial column outflows begin 8 days into the wet season (i.e., t = 129 d) and no drainage reports to the base of the column over the majority of the dry season.  Discrepancies between cumulative outflows from Column 1 and Column 2 are a result of difficulties encountered during experiment construction.  Eaves were installed on the inner diameter of both columns to recover and remove rainfall along the column walls.  These eaves were located at a lower level in Column 2 (i.e., closer to the upper boundary), thereby capturing ~10% more side-flow and contributing to lower cumulative outflow than Column 1. Bromide breakthrough from Pile 5 was bi-modal, with a high single-point peak shortly following application and a long release over the first year and a half at a much lower maximum concentration.  Bromide was first measured 72 hours after tracer application at concentrations ranging from 0.2 mg∙L-1 to 7 mg∙L-1.  A CHAPTER 4  150  single maximum concentration peak of 289 mg·L-1 Br- (C/C0 = 0.09) was observed at 9 days following tracer application (i.e., 245 d of total experimental time).  A non-conservative tracer (uranine) was also applied concurrently with bromide for the purpose of aiding visual observation of rapid tracer breakthrough (Blackmore et al., 2012).  The high, single-point bromide peak (i.e., 289 mg·L-1 Br-; t = 245 d or 9 days following tracer application) is coincident with the highest uranine concentration measured in the outflow from the waste rock pile (Blackmore et al., 2012), confirming that the early high concentration arrival of the tracer is not a measurement artifact.   The rapid arrival of tracer at high concentrations provides strong evidence for preferential flow in the experimental pile; however, the bromide mass released at this peak concentration represents only a small proportion (i.e., 0.1%) of the total applied bromide mass (Figure 4-3B).  The majority of bromide is released from the test pile over the first year, with peak concentration values of 130 mg·L-1.  The subsequent breakthrough follows a Gaussian distribution from day 250 to 650 of the experimental duration (see Appendix C).  Measured tracer concentrations decrease (or tail) to detection limit values (i.e., ~ 1 – 2 mg·L-1) to the end of the experiment.   Tracer tailing is generally attributed to mass exchange between regions of different mobility (Kissel et al., 1973). For the laboratory columns, bromide arrival occurred at 17 hours (Column 1) and 21 hours (Column 2) following tracer application at initially low concentrations (i.e., 4 mg∙L-1 Br-; Figure 4-3C).  Maximum concentrations in Column 1 were 333 mg·L-1 Br- (C/C0 = 0.11) at 6 days and 307 mg·L-1 Br- (C/C0 = 0.10) in Column 2 at 9 days following tracer application (i.e., day-246 (Col.1) and day-249 (Col.2) of total experimental time).  In both Column 1 and Column 2, column breakthrough curves show trends of rapid release and long tails, similar to tailing observed in the pile experiment.   Cumulative tracer recoveries for Pile 5 and laboratory columns, as a function of area-normalized flow and flow path length, are presented in Figure 4-4.  This method, as opposed to the commonly applied pore volume approach (i.e., cumulative area-normalized out flow multiplied by water content), was used due to uncertainties associated with measured absolute water content values (due to the requirement of placing sensors in relatively fine-grained material; Section 4.2.3) and to avoid the need of selecting a single effective water content value for highly heterogeneous materials.   CHAPTER 4  151  Results shown in Figure 4-4 show the experimental pile and two columns released a similar amount of the applied bromide mass at the end of each experiment, with Pile 5 at ~ 80% and Column 1 and Column 2 at 72% and 75% (respectively).  This result, in addition to similar area-normalized flow between scales (Figure 4-3A) and the conservative nature of bromide, suggests solute transport is retarded or influenced in all experiments by similar processes and characteristics.   Although both experiment scales present similar cumulative tracer recoveries, it is recognized that in terms of flow-path normalized time the pile experiment released half of the applied tracer solute mass (i.e., M/Mo = 0.5) quicker than the column experiments (Figure 4-4).  This slightly faster response may be a result of the larger grain size distribution of materials used in the pile experiment, relative to the column scales, which contributes to increased material heterogeneities and therefore a wider variety of transport regimes (Coppola et al., 2011).  However, this difference did not significantly impact the dominant flow characteristics observed, which were similar in both experiments. 4.3.2. Temporal Moment Analysis Figure 4-5 presents bromide breakthrough curves from all tracer experiments in terms of normalized mass flow rate 𝑆/𝑀𝑜 [T-1] and flow corrected time (τ) [T].  For the former, 𝑀𝑜 [M] is the total tracer mass applied to the system and 𝑆(𝜏) was defined previously in Equation 4-2.  Flow-corrected breakthrough curves stretch and compress results during periods of high and low flow, respectively, and are calculated for time periods following tracer application. The pile flow-corrected breakthrough curve (Figure 4-5A) shows the majority of bromide release has occurred by 300 (flow-corrected) days (or day-545 in experimental time), followed by a significantly lower bromide release.  The jagged line observed beyond 300 (flow-corrected) days in Figure 4-5A is reflective of decreasing bromide concentrations and increasing flow rates during the second wet season (2010 – 2011). Column flow-corrected breakthrough results (Figure 4-5B) are similar between Column 1 and Column 2.  At the column-scale, tracer concentrations are only measured in the wet season as drainage does not report to the column base during the dry season.  As such, column flow-corrected curves are relevant for CHAPTER 4  152  the wet season only and the absence of periods of low flow and shorter flow paths (i.e., residence time) results in smoother curves relative to those observed at the experimental pile-scale.   Data shown in Figure 4-5 is used to calculate temporal moments (Table 4.3). For conservative tracer experiments, only the first and second temporal moments are used to estimate parameters such as the pore water velocity and dispersivity coefficient (Aris, 1958).  The pile experiment has a first temporal moment or mean residence time (𝜏̅) of 295 days and a moderate normalized second temporal moment or spreading (𝐶𝑉𝜏) of 0.20.  Both column experiments (Column 1 and Column 2) have similar mean residence times and spreading (i.e., 𝜏̅  = 14 d and 20 d, 𝐶𝑉𝜏= 0.53 and 0.49; respectively).  Calculated apparent mobile water contents are similar for all experiments and 𝑣, the ratio of these values to in situ moisture content (Equation 4-4), are lower in the pile experiment.  Lower calculated 𝑣 values suggest a higher likelihood of preferential flow at the pile-scale, relative to the column scale.  Low 𝑣 values suggest a dual-domain approach, or immobile-mobile model conceptualization, may be applicable to all experiments. 4.3.3. Numerical Modeling 4.3.3.1. Estimation of dispersivities using the flow-corrected ADE Dispersivity values estimated by fitting Equation 4-7 to normalized mass flow rates are shown in Figure 4-6.  Dispersivity could not be estimated at the larger pile-scale due to a poor agreement between theoretical (ADE-modeled) and measured breakthrough curves (Figure 4-6A).  For column experiments, a moderate fit is observed between flow-corrected-time breakthrough curves and modeled ADE curves and dispersivity values are estimated to be 0.12 – 0.15 m (Column 1) and 0.08 – 0.12 m (Column 2) (Figure 4-6B and C).  However, modeled ADE curves could not simulate peak concentrations (at τ = 10 d) for Column 1 or breakthrough tails (between τ = 19 d – 40 d) for Column 2 (Figure 4-6B and C, respectively).   This result suggests a single porosity ADE may be able to partially simulate solute transport at the smaller scale, which is not possible in an increasingly complex and larger scale system.   The ADE model assumes the flow regime (advection and dispersion) are temporally invariant, which is applicable to the column scale due to its smaller size and one-half wet season was required to flush the CHAPTER 4  153  majority of the tracer mass.  At the pile-scale, several wet-dry seasons were required to flush a similar tracer mass and the change in flow regimes (between seasons) cannot be accommodated well by the ADE model.   4.3.3.2. Flow and solute transport modeling using the MIM approach Numerical simulations implemented the mobile-immobile (MIM) dual-porosity approach, following the inability to simulate transport at the pile-scale and inadequacy to simulate breakthrough peaks (Column 1) and tails (Column 2) using a uniform ADE.  Simulations held the immobile water contents constant and only solute transfer occurs between the mobile and immobile domains (Figure 4-7 and 4-8). A good agreement was observed between measured and simulated solute breakthrough at the pile- and column- scale (Figure 4-7 and 4-8).  Specifically, measured breakthrough peaks and tails from column experiments are captured with the MIM approach.  In regards to Pile 5, the dominant flow and solute transport is captured with the MIM approach however simulations did not capture the single-point peak concentration in pile breakthrough (at 9 days after tracer release corresponding to day 245 since experiment initiation).  Although Pile 5 transport may be better described by a model that includes a small fraction of open void or film flow, more complex simulations were not carried out as this peak represents a small flow component in relation to total flow (i.e., 0.1%). Secondary simulations that assume uniform flow and solute transport (i.e., uniform model) were performed to validate the efficacy and level of simplicity required to simulate observed breakthrough curves from all experiments.   At the pile-scale, the uniform model predicted tracer arrival only after a near complete flushing of the tracer from the pile.  This result confirms that a uniform flow/solute model is inadequate at this larger scale and supports the previous inabilities to model Pile 5 breakthrough with a single-porosity flow-corrected ADE.   Results for the column simulations show that the uniform model is able to reproduce tracer arrival in general terms (Figure 4-6B and4-6C; fine dashed lines); however, this model predicted late arrival and reduced peak concentrations. Differences between simulated breakthrough curves obtained from uniform ADE versus MIM approaches are likely attributed to the use of a time-averaged flow velocity versus CHAPTER 4  154  transient flow (respectively).  Nevertheless, these results suggest the column experiments can be simulated, in a general sense, with a single domain ADE or uniform model approach, however a dual-porosity MIM model better simulates the observed peak concentrations and breakthrough tails.  For Pile 5, only the dual porosity MIM approach yields a fit for measured breakthrough curves. Model simulations for column experiments used identical flow parameters to simulate Column 1 and Column 2 flow and solute transport, which further supports the previous observations (from Figure 4-4 and Figure 4-5B) that both columns contain material with comparable characteristics and transport regimes.  Final dispersivity estimates, immobile porosity and mass-transfer coefficient values used in Pile 5 and column simulations are given in Table 4.4.  Mass-transfer coefficients of Class A material differed by a factor of 3 to 30 between experimental scales, whereas immobile porosity values were similar. Higher dispersivity values were applied in the MIM model simulations, in comparison to those estimated using a single-porosity flow-corrected ADE (shown in Figure 4-6).  A similar observation was noted by Sassner et al (1994), who suggested estimated dispersivity values from ADE-modeled curves may indeed be larger.  Weighting of dispersivity values shown in Table 4.4 to flow path length (e.g., Pile 5: Class A 75%: Class C 25%) produces a pile-specific dispersivity value that is approximately two times higher than those used in column simulations (i.e., Pile 5: 0.25 m versus Column 1 and 2: 0.14 m), which agrees with well-established concepts of scale-dependency of dispersivity (Coppola et al., 2011; Vanderborght and Vereecken, 2007).  Changes to final dispersivity values were applied in several sensitivity analyses and did not substantially impact the results.  Model simulations of Pile 5 solute transport could not capture the initial tracer peak (i.e., 289 mg·L-1 at day-245 or 9 days following tracer application); however, this single-point peak represents only ~ 0.1% of the captured breakthrough mass.   This minor component likely reflects fast preferential or film flow as a result of more heterogeneous, coarser-grained material present at the larger scale.  Although the dual-porosity MIM approach was not able to reproduce the fast preferential flow component at the pile-scale, the model results adequately captured fast matrix flow and solute mass-transfer to slower matrix flow and immobile waters at both scales and should be considered non-unique. CHAPTER 4  155  4.4. Discussion This study compares flow and tracer results at two scales, differing by an order of magnitude in height (i.e., 10 m versus 0.8 m).  The objective of the experimental design/construction was to ensure similar materials and conditions were used in all experiments and reduce the likelihood of discrepancies between scales.  PSD curve similarities (refer to Figure 4-2) support the comparability of study materials used, per a ≤10% difference between the percent passing the 2 mm sieve diameter for Pile 5 and column experiments.  The conclusion that these efforts were adequate in replicating one system at two scales were supported by 3 observations; similar cumulative outflow (Figure 4-3) and measured tracer mass release (Figure 4-4) at both scales, and the application of identical or similar hydrologic parameters (i.e., 𝛼, 𝑛, 𝐾𝑠𝑎𝑡, 𝜃𝑠, 𝜃𝑟, 𝜃𝑖𝑚) in model simulations, with the exception of dispersivities and in particular mass transfer coefficients.   Pile 5 breakthrough curves show the majority of bromide is released over the first year (following tracer application) at a mean residence time of 295 days and preferential flow was evidenced as a minor (~ 0.1%) flow component due to a single maximum concentration bromide peak within days of application.  Breakthrough curves from Column 1 and Column 2 were similar, with both showing a rapid rise in drainage concentrations to maximum values, a decrease in measured concentrations with increasing time, but no indications of preferential flow.   Results depicted in Figure 4-4 show cumulative tracer recoveries (normalized to cross sectional area and flow path length) from all experiments are comparable.  This comparability suggests both experiments are representative of a similar transport scale, or an averaging of observation (or smaller) scale heterogeneities within the porous media (Roth et al., 1991).  This result indicates that this study was successful in mimicking processes active at the larger scale in the laboratory and demonstrates the applicability of estimated solute parameters from column experiments to field scale experiments (Coppola et al., 2011) – with the exception of dispersivities and mass transfer coefficients.   The temporal scale of tracer breakthrough differed between the two experiments.  Two water years (or wet-dry seasons) were required for the experimental pile to release approximately 81% of the total CHAPTER 4  156  applied bromide mass.  At the column scale, one-half wet season resulted in the release of 74% (Column 1) and 72% (Column 2) of the solute mass.  Although temporal differences are noted, bromide continued to be detected in all drainage waters at concentrations near detection limit values (i.e., 1 mg·L-1 – 2 mg·L-1) until the end of the experiment duration.  These long tails suggest extremely slow matrix flow in zones of lower conductivity and/or mass-transfer from stagnant or immobile domains is present in both experiments.   The validity of the flow-corrected time method to calculate temporal moments was evaluated by comparing flow-corrected breakthrough curves to modeled curves using a single-domain advection dispersion equation with homogeneous parameters (i.e., Eriksson et al., 1997; modified from Kreft and Zuber (1978)).  A moderate fit between column-scale flow-corrected data (Column 1 and Column 2) and the ADE-modeled curves indicates temporal moments may be estimated as single effective values and implicitly account for the existence of variable flow paths and macrodispersive solute spreading (Eriksson et al., 1997).  This agreement suggests a single-domain uniform model may be applicable at the column scale.  However, only transient flow models that applied the dual porosity MIM approach were able to capture the breakthrough curve and tailing observed from the column experiments.  These results indicate at least one of the assumptions associated with the flow-corrected time method (i.e., effective first temporal moments or macrodispersion) is not valid.   The same single-domain ADE could not simulate flow-corrected breakthrough at the larger pile-scale.  This result was not surprising as aquifer heterogeneity is typically proportional to experimental scale and the ability to calculate parameters as single or effective values is increasingly difficult in highly heterogeneous systems.  Additionally, the Pile 5 breakthrough curve extends over two water years (Figure 4-3) that alternates between dominantly drain-down matrix flow (dry seasons) and event flow including preferential and fast matrix flow (wet seasons).  This large temporal variation in flow regimes may not be adequately captured by the flow-corrected time compression and add to the difficulty in resolving effective parameters using this method.   Breakthrough curves that present long tails and tracer mass recoveries less than unity (i.e., Pile 5 = 0.80; Column 1 = 0.75; Column 2 = 0.72) suggest a rate-limited mass-transfer model may better explain solute CHAPTER 4  157  transport at both experimental scales and support the application of a MIM approach.  The application of this approach does not infer the absence of macrodispersive processes at either scale, but points to its suitability in representing observed bromide breakthroughs.  In this approach, waste rock can be conceptualized as containing measurement-scale mobile and immobile domains with variable conductivities that are associated with pore-scale mass-transfer processes (Harvey and Gorelick, 2000).  Solute mass is partitioned between either the moving (or mobile) fluid and a less permeable (or immobile) domain (Haggerty and Gorelick, 1995; Carrera et al., 1998; Haggerty et al., 2000) and solute movement between domains is defined by a mass-transfer coefficient (𝜔𝑚𝑖𝑚; d-1 (Eq. 4-13c)) and occurs by physical or chemical processes (Haggerty et al., 2004).   In this study, mass-transfer coefficient values were estimated by fitting the simulated dual-porosity MIM curves to observed breakthrough tails (Table 4.4) and final values present the most significant identifiable difference between column and pile experiments.  Specifically, Class A material mass-transfer coefficients were approximately an order of magnitude smaller at the pile-scale, relative to values estimated for the columns (i.e., 𝜔𝑐𝑜𝑙𝑢𝑚𝑛 = 0.14 -1.4 x 10-3 d-1; 𝜔𝑝𝑖𝑙𝑒 = 4.5 x 10-5 d-1).  The application of the same conservative, non-sorbing tracer in all experiments indicates chemical processes are ruled out as a contributing factor to the difference in mass-transfer coefficient values.  Additionally, the similarity of immobile and mobile porosities, (area-normalized) outflow, and (flow-path normalized) tracer recovery suggests physical processes do not differ considerably between the two scales.   Temporal and/or spatial dependencies are two possible explanations for the mass-transfer coefficient difference.  For the former, temporal dependencies may be related to differences in experiment duration.  This concept is described by Haggerty et al. (2004), which showed mass-transfer coefficients, estimated by fitting the breakthrough curve tail, are representative of mass-transfer processes contributing to the observed tail and operating over the experimental duration.  In this scenario, the timescale of mass transfer estimated from a single-rate model will increase with the experiment duration and produce lower mass-transfer coefficients (Haggerty et al., 2004).  Moreover, breakthrough curve tails from highly heterogeneous systems, such as materials used in this study, may not be captured adeq