UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Hydrological and hydrogeochemical characteristics of neutral drainage from a waste rock test pile Bay, Daniel S. 2009

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_2009_fall_bay_daniel.pdf [ 8.58MB ]
Metadata
JSON: 1.0052724.json
JSON-LD: 1.0052724+ld.json
RDF/XML (Pretty): 1.0052724.xml
RDF/JSON: 1.0052724+rdf.json
Turtle: 1.0052724+rdf-turtle.txt
N-Triples: 1.0052724+rdf-ntriples.txt
Original Record: 1.0052724 +original-record.json
Full Text
1.0052724.txt
Citation
1.0052724.ris

Full Text

HYDROLOGICAL AND HYDROGEOCHEMICAL CHARACERISTICS OF NEUTRAL DRAINAGE FROM A WASTE ROCK TEST PILE by  DANIEL S. BAY B.Sc. Queen’s University, 2004  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in  THE FACULTY OF GRADUATE STUDIES (Geological Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2009 © Daniel S. Bay, 2009  ABSTRACT In 2005 the University of British Columbia, Teck’s Applied Research & Technology (ART) group and Compañía Minera Antamina S.A. initiated a collaborative study to improve the hydrological, geochemical, mineralogical and biological understanding of mine waste rock. The site of the investigation is the Antamina mine, Peru, which consists of a polymetallic Cu-Znskarn deposit that produces high carbonate-content waste rock and typically leads to pH-neutral effluent waters. One component of this study is the construction of five 36 m x 36 m x 10 m (high) waste rock piles upon an impermeable HDPE liner placed to capture and monitor the quality and quantity of all infiltrating water. The first test pile was constructed in 2006 and consists of relatively course, heterogenous marble and hornfels waste rock material that may produce moderately- to poor-quality drainage according to the classification scheme used at the mine. A comprehensive network of instrumentation measures the in-situ temperature and moisture content, and allows the extraction of fluid samples from within the pile. The objective of this thesis is to analyze the hydrological and geochemical data collected from test pile 1 during the initial 21 months of pile operation between January-2007 and October-2008. Results indicate that flow and chemistry exhibit strong spatial and seasonal patterns. Seasonal patterns are driven by the wet season rains that deposit a majority of the annual rainfall over 7 months, and spatial patterns are likely controlled by rock-type heterogeneity and variability in flow path length and residence time that result from the end-dump construction process. Test pile 1 took 3 – 4 months to wet up, and the maximum pile outflow measured during the wet season is nearly 15 m3/day. Water balance calculations indicate that 41 % of precipitation reports as outflow while 59 % evaporates over the course of one water year. Effluent water is pH-neutral (pH = 7 8.5), with seasonally varying SO4 (500 – 2000 mg/L) and Zn (0.5 – 2.5 mg/L) concentrations. Based on the maximum calculated SO4 and Zn loading rates of 4 – 6 and 0.003 – 0.005 mg/kg/week, respectively, nearly 0.4 % of the total initial solid phase sulphur has been depleted, whereas only 0.01 % of the total initial solid phase Zn has been depleted. Waste rock reactivity appears to be declining with time.  ii  TABLE OF CONTENTS ABSTRACT………………………………………………………………………………….…..ii TABLE OF CONTENTS……………………………………………………………………….iii LIST OF TABLES……………………………………………………………………………..viii LIST OF FIGURES……………………………………………………………………………..ix ACKNOWLEDGEMENTS…………………………………………………………………….xi DEDICATION………………………………………………………………………………….xii 1  INTRODUCTION ............................................................................................................... 1  1.1 1.2 1.3 1.4 1.5 1.6  BACKGROUND AND SIGNIFICANCE ....................................................................... 1 THE ANTAMINA MINE ................................................................................................ 2 PROJECT OVERVIEW .................................................................................................. 2 RELEVANCE TO THE ANTAMINA MINE ................................................................. 3 THESIS OBJECTIVES ................................................................................................... 4 LITERATURE REVIEW ................................................................................................ 5 1.6.1 What is Waste Rock? ................................................................................................ 5 1.6.2 End-Dumping and Internal Structure ........................................................................ 6 1.6.3 Waste Rock: Physical Properties .............................................................................. 7 1.6.4 Waste Rock: Flow Processes .................................................................................... 8 1.6.4.1 Soil Water Characteristic Curve............................................................................ 8 1.6.4.2 Hydraulic Conductivity ......................................................................................... 9 1.6.4.3 Matrix and Preferential Flow .............................................................................. 10 1.6.4.3.1 Preferential Flow .......................................................................................... 11 1.6.4.3.2 Capillary Barrier ........................................................................................... 12 1.6.4.3.3 Macropore Flow ........................................................................................... 12 1.6.5 Waste Rock: Chemistry .......................................................................................... 12 1.6.5.1 Residence Time: Impact of Flow on Chemistry.................................................. 13 1.6.6 Challenges in Predicting Weathering from Particle Size and Flow Indices ........... 14 1.6.7 Waste Rock: Neutral Drainage ............................................................................... 14 1.6.8 Challenges in Waste Rock Investigations: Scaling Issues ...................................... 15 1.6.9 Waste Rock Test Piles ............................................................................................ 16 1.7 ORGANIZATION OF THESIS .................................................................................... 17 2 SITE DESCRIPTION, INSTRUMENTATION, SAMPLING AND MATERIAL CHARACTERIZATION ........................................................................................................... 21 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9  LOCATION ................................................................................................................... 21 CLIMATE ...................................................................................................................... 21 REGIONAL GEOLOGY ............................................................................................... 21 DEPOSIT GEOLOGY ................................................................................................... 22 HYDROGEOLOGY ...................................................................................................... 23 DEPOSIT DEVELOPMENT......................................................................................... 24 WASTE ROCK GENERATION AT ANTAMINA ...................................................... 25 WASTE ROCK CLASSIFICATION ............................................................................ 26 TEST-PILE DESIGN..................................................................................................... 26 2.9.1 Site .......................................................................................................................... 27 iii  Base Construction Methods .................................................................................... 27 2.9.2 2.9.3 Pile Construction Methods ...................................................................................... 28 2.10 INSTRUMENTATION ................................................................................................. 29 2.10.1 Pile Instrumentation ................................................................................................ 29 2.10.1.1 Time Domain Reflectometry Probes ................................................................... 30 2.10.1.2 Thermistors.......................................................................................................... 30 2.10.1.3 Soil Water Solution Samplers ............................................................................. 31 2.10.1.4 Gas Ports ............................................................................................................. 31 2.10.1.5 Tipping Buckets .................................................................................................. 32 2.10.1.6 Electrical Conductivity Probes ............................................................................ 32 2.10.1.7 Composite Tank and Flow Splitter ..................................................................... 33 2.10.2 Meteorological Instrumentation .............................................................................. 34 2.10.2.1 Precipitation ........................................................................................................ 34 2.10.2.2 Evaporation ......................................................................................................... 34 2.10.2.3 Air Temperature, Relative Humidity, Wind Speed and Direction. ..................... 34 2.11 WATER SAMPLING .................................................................................................... 35 2.11.1 Water Quality Parameters ....................................................................................... 35 2.11.1.1 Field Parameters .................................................................................................. 35 2.11.1.2 Lab Parameters .................................................................................................... 35 2.11.1.2.1 Dissolved Metals ........................................................................................ 36 2.11.1.2.2 Total Metals ................................................................................................ 36 2.11.1.2.3 Other parameters ......................................................................................... 36 2.12 GAS SAMPLING .......................................................................................................... 37 2.13 WASTE ROCK PROPERTIES ..................................................................................... 37 2.13.1 Solid Phase Mineralogical Composition................................................................. 37 2.13.2 Solid Phase Elemental Composition ....................................................................... 38 2.13.3 Acid Base Accounting ............................................................................................ 38 2.13.4 Particle Size Distribution ........................................................................................ 39 2.13.5 Hydraulic Properties ............................................................................................... 40 2.13.6 Dry Bulk Density .................................................................................................... 40 2.13.7 Porosity ................................................................................................................... 41 2.14 SUMMARY ................................................................................................................... 42 3  RESULTS: TEST PILE HYDROLOGY ........................................................................ 56  3.1  PRECIPITATION .......................................................................................................... 57 3.1.1 Estimation of Missing Precipitation Data ............................................................... 57 3.1.1.1 Local Gauges and Linear Least Square Regression Analysis ............................. 57 3.1.2 Estimation Validity ................................................................................................. 58 3.1.3 Uncertainty in Precipitation Measurements ............................................................ 60 3.1.4 Final Site Precipitation and Reported Trends ......................................................... 61 3.2 EVAPORATION ........................................................................................................... 62 3.2.1 Evaporation: Study Period ...................................................................................... 63 3.2.2 Evaporation: 2007-2008 Water Year ...................................................................... 63 3.2.2.1 Pan Evaporation and Penman-Monteith Estimates ............................................. 64 3.2.2.2 Modeled Evaporative Fluxes: SoilCover ............................................................ 65 3.2.2.3 Lysimetry Water Balance Estimate ..................................................................... 66 3.3 OUTFLOW .................................................................................................................... 66 iv  Correcting Tipping Bucket Data: RANSAC........................................................... 67 3.3.1 3.3.2 Total Pile Discharge ................................................................................................ 68 3.3.3 Outflow Hydrographs ............................................................................................. 69 3.3.4 Hydrograph Variability ........................................................................................... 70 3.3.5 Lysimeter Net Infiltration ....................................................................................... 72 3.3.6 Effect of Antecedent Moisture Content .................................................................. 72 3.3.7 Wet Season Responses ............................................................................................ 74 3.3.7.1 Preferential and Macropore Flow ........................................................................ 74 3.3.7.2 Pressure-Wave Displacement of Water through the Pile .................................... 77 3.3.7.3 Non-Vertical Flow............................................................................................... 77 3.3.8 Difficulty Identifying Different Flow Pathways and Multiple Arrivals ................. 78 3.3.9 Uncertainty.............................................................................................................. 79 3.4 MOISTURE CONTENT ............................................................................................... 81 3.4.1 Time Domain Reflectometry Data .......................................................................... 81 3.4.2 TDR Probe Response .............................................................................................. 82 3.4.2.1 Protective Layer .................................................................................................. 82 3.4.2.2 Compacted Surface ............................................................................................. 83 3.4.2.3 Batter ................................................................................................................... 85 3.4.2.4 TDR Variability................................................................................................... 86 3.4.3 Wetting Front Propagation and Evidence of Preferential or Non-Vertical Flow ... 87 3.4.4 Change in Storage ................................................................................................... 88 3.5 DISCUSSION ................................................................................................................ 89 3.5.1 Water Balance ......................................................................................................... 89 3.5.2 Pore Water Displacement and Macropore Flow Velocity ...................................... 90 3.5.3 Effects of the Batter on Pile Hydrology.................................................................. 93 3.6 SUMMARY ................................................................................................................... 93 4  RESULTS: TEST PILE GEOCHEMISTRY ............................................................... 108  4.1  ELECTRICAL CONDUCTIVITY .............................................................................. 109 4.1.1 Lysimeters ............................................................................................................. 109 4.1.2 Composite Tank .................................................................................................... 111 4.1.3 Pile Interior ........................................................................................................... 111 4.1.3.1 Changes at Depth .............................................................................................. 112 4.2 TEMPERATURE ........................................................................................................ 113 4.2.1 Lysimeters ............................................................................................................. 113 4.2.2 Pile Interior .......................................................................................................... 114 4.2.2.1 Internal vs. External Temperatures ................................................................... 115 4.2.2.2 Spatial Variability in Temperatures .................................................................. 116 4.2.2.3 Evidence of Advection and/or Evaporation from Front Batter ......................... 116 4.3 PH ................................................................................................................................ 117 4.3.1 Lysimeters ............................................................................................................. 117 4.3.2 Composite Tank .................................................................................................... 118 4.3.3 Sump ..................................................................................................................... 118 4.3.4 Pile Interior ........................................................................................................... 119 4.4 SOLUTES .................................................................................................................... 119 4.4.1 Metals .................................................................................................................... 120 4.4.1.1 Zinc.................................................................................................................... 120 v  4.5 4.6 4.7  4.8 5  4.4.1.1.1 Lysimeters, Composite Tank and Sump ..................................................... 120 4.4.1.1.2 Pile Interior ................................................................................................. 121 4.4.2 Cations .................................................................................................................. 122 4.4.2.1 Lysimeters, Composite Tank and Sump ........................................................... 122 4.4.2.2 Pile Interior ........................................................................................................ 123 4.4.3 Trace Metals.......................................................................................................... 123 4.4.4 Major Ions ............................................................................................................. 124 4.4.4.1 Dissolved Sulphate ............................................................................................ 124 4.4.4.2 Alkalinity........................................................................................................... 125 4.4.4.3 Nitrate ................................................................................................................ 125 PORE GASES.............................................................................................................. 126 GEOCHEMICAL MODELING .................................................................................. 127 DISCUSSION .............................................................................................................. 128 4.7.1 Gypsum Precipitation and the SO42-/Ca2+ Ratio ................................................... 128 4.7.2 Zinc: Source and Attenuation Mechanisms .......................................................... 129 4.7.3 Spatial Variability ................................................................................................. 130 4.7.4 Comparison to Field Cell Chemistry .................................................................... 131 SUMMARY ................................................................................................................. 133 RESULTS: TEST PILE MASS LOADINGS ............................................................... 154  5.1 5.2  INITIAL PILE COMPOSITION ................................................................................. 154 SOLUTE MASS LOADINGS ..................................................................................... 155 5.2.1 Low-Resolution Mass Loadings ........................................................................... 156 5.2.1.1 Cumulative Loadings ........................................................................................ 156 5.2.1.2 Time-Averaged Loading Rates ......................................................................... 156 5.2.1.3 Flow vs. Chemistry ........................................................................................... 157 5.2.1.4 Specific Conductance-Solute Correlations........................................................ 158 5.2.2 High-Resolution Mass Loadings........................................................................... 159 5.2.2.1 Basal Lysimeter SO4 Loading ........................................................................... 159 5.2.2.2 Lysimeter SO4, Ca and Zn Loadings ................................................................. 160 5.2.3 Comparison Between Low- and High-Resolution Loadings ................................ 160 5.2.4 Comparison to Field Cells .................................................................................... 161 5.3 ELEMENT DEPLETION ESTIMATES ..................................................................... 162 5.3.1 Mineral Depletion Uncertainties ........................................................................... 163 5.4 SUMMARY ................................................................................................................. 164 6  DATA INTEGRATION ................................................................................................. 174  6.1 6.2 6.3 6.4  EVIDENCE OF DIFFERENT FLOW MECHANISMS ............................................. 174 PYRITE OXIDATION: ESTIMATING SO4 .............................................................. 175 OXYGEN DEPLETION ESTIMATE ......................................................................... 176 SPATIAL VARIABILITY IN PILE BEHAVIOUR ................................................... 177 6.4.1 Protective Layer .................................................................................................... 178 6.4.2 Tipping Phase I - Sub-lysimeter B ........................................................................ 179 6.4.3 Tipping Phase II - Sub-lysimeter A ...................................................................... 180 6.4.4 Tipping Phase III - Sub-lysimeter C ..................................................................... 181 6.4.5 Entire Pile / Basal Lysimeter D ............................................................................ 183 6.5 SUMMARY ................................................................................................................. 183 vi  7  CONCLUSIONS ............................................................................................................. 187  7.1 7.2  PROJECT OVERVIEW AND SIGNIFICANCE ........................................................ 187 SUMMARY OF RESULTS ........................................................................................ 187 7.2.1 Material Properties ................................................................................................ 187 7.2.2 Hydrology ............................................................................................................. 188 7.2.3 Chemistry .............................................................................................................. 188 7.2.4 Solute Loadings .................................................................................................... 189 7.2.5 Data Integration .................................................................................................... 190 7.3 RELEVANCE TO THE ANTAMINA MINE ............................................................. 191 7.3.1 Tucush Dump ........................................................................................................ 191 7.3.2 Re-Classification of Class B Waste Rock ............................................................. 192 7.3.3 Possibility to Reduce Solute Concentrations Based on Dump Modifications ...... 192 7.4 RECOMMENDATIONS FOR CURRENT PILE MAINTENANCE AND FUTURE PILE CONSTRUCTION ............................................................................................................ 193 7.4.1 Instrumentation ..................................................................................................... 193 7.4.2 Modeling ............................................................................................................... 193 7.4.3 Data Management ................................................................................................. 193 7.4.4 Further Tests ......................................................................................................... 194 7.4.5 Personnel ............................................................................................................... 195 7.5 CONCLUDING REMARKS ....................................................................................... 195 8  REFERENCES ................................................................................................................ 196  APPENDIX A: APPENDIX B: APPENDIX C: APPENDIX D: APPENDIX E: APPENDIX F: APPENDIX G: APPENDIX H: APPENDIX I:  PILE DESIGN AND CONSTRUCTION .................................................. 206 PILE 2 & 3 CONSTRUCTION MODIFICATIONS FROM PILE 1 ..... 217 PRECIPITATION ....................................................................................... 225 EVAPORATION ......................................................................................... 243 OUTFLOW .................................................................................................. 262 TIME DOMAIN REFLECTOMETRY PROBES.................................... 287 COMPOSITE TANK, FLOW SPLITTER AND SUMP ......................... 296 DATALOGGER PROGRAM .................................................................... 300 MATLAB M.FILES .................................................................................... 306  vii  LIST OF TABLES Table 1.1. Comparison of recent waste rock test piles................................................................................ 18 Table 2.1. General classification of waste rock types from the Antamina mine......................................... 44 Table 2.2. Test pile 1 construction timeline. ............................................................................................... 45 Table 2.3. Solid phase mineral composition of waste rock......................................................................... 46 Table 2.4. Solid phase elemental composition of waste rock. .................................................................... 46 Table 2.5. Summary of instrumentation...................................................................................................... 47 Table 2.6. Final mass calculation. ............................................................................................................... 47 Table 3.1. Total estimated volume, weekly rate, and daily flux of precipitation and lysimeter outflow.... 95 Table 3.2. Summary of evaporation estimates throughout the study period and water year. ..................... 96 Table 3.3. Cumulative interpreted flow for each lysimeter. ....................................................................... 97 Table 4.1. Parameter comparison between basal-lysimeter and composite tank. ..................................... 135 Table 4.2. Temperature comparison between top and bottom of pile. ...................................................... 135 Table 4.3. Solute discharging and receiving limits ................................................................................... 136 Table 4.4. Minerals modeled to be near saturation in basal lysimeter effluent......................................... 136 Table 5.1. Cumulative and average wet and dry season loading rates. ..................................................... 165 Table 5.2. Specific conductance-solute lysimeters and composite tank regression correlations. ............. 165 Table 5.3. High- and low-resolution basal lysimeter loading comparison; 2007. .................................... 165 Table 5.4. High- and low-resolution lysimeter loading comparison; 11- to 19-April-2007. .................... 166 Table 5.5. Average field cell and lysimeter loading rates. ........................................................................ 167 Table 6.1. Integration of waste rock material and lysimeter effluent properties. ..................................... 185  viii  LIST OF FIGURES Figure 1.1. Neutral drainage waste rock project. Components relevant to this thesis are shaded in blue. . 19 Figure 1.2. Waste rock pile schematic. ....................................................................................................... 19 Figure 1.3. Flow processes in waste rock. .................................................................................................. 20 Figure 2.1. Map of mine location................................................................................................................ 48 Figure 2.2. Plan view of schematic lithology of the Antamina deposit. ..................................................... 49 Figure 2.3. Schematic lithology and metal zonation of Antamina .............................................................. 49 Figure 2.4. Map of the Antamina mine. ...................................................................................................... 50 Figure 2.5. South view of completed test pile 1 and platforms................................................................... 51 Figure 2.6. Test pile 1 base design and instrumentation schematic. ........................................................... 52 Figure 2.7. Test pile 1 side profile and instrumentation schematic............................................................. 52 Figure 2.8. Plan view location of instrumentation points in relation to pile exterior.................................. 53 Figure 2.9. Inside the instrumentation hut. ................................................................................................. 54 Figure 2.10. Particle size distribution of waste rock material placed in the pile. ....................................... 55 Figure 3.1. Antamina rain gauge locations with respect to the study site. .................................................. 98 Figure 3.2. Reported and estimated cumulative rainfall compared to the Antamina precipitation............. 98 Figure 3.3. Rainfall comparison between the two site rain gauges. ............................................................ 99 Figure 3.4. Daily and cumulative estimated precipitation. ......................................................................... 99 Figure 3.5. Variability of rain as measured by site rain gauge during Jan-2008 ...................................... 100 Figure 3.6. Class A Pan evaporation schematic and Yanacancha station, Antamina. .............................. 100 Figure 3.7. Daily precipitation and evaporation ....................................................................................... 101 Figure 3.8. Comparison between top and batter of pile. ........................................................................... 101 Figure 3.9. Outflow hydrograph for the sub-lysimeters and basal lysimeter. .......................................... 102 Figure 3.10. Precipitation and lysimeter fluxes. ....................................................................................... 104 Figure 3.11. Upper, lower and expected flow for basal lysimeter D and sub-lysimeter C. ...................... 105 Figure 3.12. Shallow TDR moisture content and precipitation in mm/day. ............................................. 106 Figure 3.13. Line 4 TDR moisture content response. ............................................................................... 106 Figure 3.14. Line 1 TDR moisture content and precipitation. .................................................................. 107 Figure 3.15. Line 2 TDR moisture content and precipitation. .................................................................. 107 Figure 4.1. Continual and weekly-obtained specific conductivity vs. and pile discharge. ....................... 137 Figure 4.2. Parameter comparison between basal lysimeter and composite tank. .................................... 138 Figure 4.3. Temporal and spatial distribution of specific conductance. ................................................... 139 Figure 4.4. Cross-sectional diagram of four construction stages and four lysimeters .............................. 140 Figure 4.5. Internal and external pile temperature variation. .................................................................... 141 Figure 4.6. Average temperatures in lines 1, 2 and 4 over the water year. ............................................... 142 Figure 4.7. Temporal and spatial distribution of pH. ................................................................................ 143 Figure 4.8. Temporal and spatial distribution of Zn. ................................................................................ 144 Figure 4.9. Sump comparison to basal lysimeter and composite tank. ..................................................... 145 Figure 4.10. Temporal and spatial distribution of Ca. .............................................................................. 146 Figure 4.11. Temporal and spatial distribution of Na. .............................................................................. 147 Figure 4.12. Temporal and spatial distribution of Mg. ............................................................................. 148 Figure 4.13. Temporal and spatial distribution of K. ................................................................................ 149 Figure 4.14. Temporal and spatial distribution of SO4. ............................................................................ 150 Figure 4.15. Temporal and spatial distribution of alkalinity..................................................................... 150 Figure 4.16. Percent charge balance error of PHREEQC model of basal lysimeter data. ........................ 151 Figure 4.17. Gypsum SI for the basal lysimeter D and composite tank E. ............................................... 151 Figure 4.18. Calcite SI for the basal lysimeter D and composite tank E. ................................................. 151 Figure 4.19. A trilinear Piper Plot of basal lysimeter D effluent .............................................................. 152 Figure 4.20. Regression analysis of Ca and SO4 at basal lysimeter D. ..................................................... 152 Figure 4.21. Correlation between SO4 and Zn at lysimeters and composite tank. .................................... 153  ix  Figure 5.1. Best estimated cumulative pile loadings................................................................................. 168 Figure 5.2. Best estimated weekly SO4, Ca and Zn mass loadings for all four lysimeters. ...................... 169 Figure 5.3. Weekly SO4, Ca and Zn solute concentration and mass loading. ........................................... 170 Figure 5.4. Specific conductance correlation to SO4, Ca and Zn as measured at D. ................................ 171 Figure 5.5. Basal lysimeter loading comparison to concentration and flow rate; 2007. ........................... 172 Figure 5.6. Lysimeter loading variation, 11- to 19-Apr-2007................................................................... 173 Figure 6.1. Integration of physical properties and observations throughout the pile. ............................... 186  x  ACKNOWLEDGEMENTS  It was a privilege to work with, and learn from, my supervisors Roger Beckie, Leslie Smith and Uli Mayer. Thank you for your guidance and support, and for keeping the atmosphere light and humorous. Many thanks go to Bern Klein, Ward Wilson and Stephane Brienne for their support on the Antamina project.  This thesis could not have been accomplished without the financial support of the Natural Science and Engineering Research Council of Canada, Teck’s Advanced Research and Training group, and Compañia Minera Antamina.  I owe a great deal to all of my UBC colleagues, especially Juan Carlos Corazao Gallegos, Matt Neuner, and Craig Nichol. I exceptionally enjoyed sharing an office with several great people and benefitting from the combined wealth of knowledge within it. Specifically, this includes: Sharon Blackmore, Mike Conlan, John Dockrey, Katie Jones, Cassandra Koenig, Holly Peterson and Cindy Starzyk. Special thanks also goes to Celedonio Aranda, Charlene Haupt and Joe Marcoline, and to Andre Wild, David Jones, Karim Damani, Sukhi Hundal and Joern Unger for other logistical and technical help. I would also like to express deep gratitude to Bae Yong Joon for his spiritual support along the way.  At Antamina, I owe a great deal to Humberto Valdivia, Bartolome Vargas and Raul Jamanca. Several others in the environmental area, and people from the contractors Hijos de Ayash were also integral in the process. A very special thanks goes to Fabiola Sifuentes.  xi  To my loving parents Eli and Masza,  Without your endless love and support I could have never accomplished this endeavor. This thesis, my proudest achievement, is dedicated to you both.  xii  Press on. Nothing in the world can take the place of perseverance. Talent will not; Nothing is more common than unsuccessful men with talent. Genius will not; Unrewarded genius is almost a proverb. Education will not; The world is full of educated derelicts. Persistence and determination are omnipotent. -  Calvin Coolidge  The important thing is not to stop questioning. -  Albert Einstein  xiii  1 1.1  INTRODUCTION BACKGROUND AND SIGNIFICANCE  The mining of base metals and other resources removes and transports more material than any other industry (ICOLD, 1996; Blowes et al., 2007). Every year, millions of tonnes of earth is exhumed, transported and processed, which consequently impacts the local environment. As a result of the increasing prevalence of mining operations, these local impacts have received global attention. In recent decades, mining practices have departed from the former doctrine of profit generation without concern for socio-environmental issues. Today, efficient and more sustainable practices are being used in the mining industry, and mine waste management continues to draw more attention. Mining activities focus on the extraction of metals and other resources from geological deposits, which may release contaminants to the environment during extraction and processing operations. Since most metal mines contain sulphide minerals, the exposure of mine wastes such as tailings and waste rock to the atmosphere often generates acidic effluent water that typically contains high dissolved solute concentrations. When this occurs, the effluent is referred to as Acid Rock Drainage (ARD). While mine wastes are typically stationary and remain in a given location indefinitely, their effluent can potentially traverse very large distances. On United States Forest Service Lands alone, there are an estimated 20,000 – 50,000 small mines releasing acidic drainage and an estimated 6,400 km of rivers and streams in the eastern United States that have been affected by this process (USDA, 1993; Blowes et al., 2007; Kleinmann et al., 1991). In Canada an estimated $ 2 – 5 billion are required to address all of the costs of poor-quality water effluent from mine environments (Feasby and Tremblay, 1995; Price, 2003). Based on these figures, the magnitude of current and future environmental impacts from the mining industry is substantial. When a significant acid-buffering capacity is found in the host rock of an ore deposit, it is possible for the mine waste effluent to have a neutral pH. This is referred to as Neutral Rock Drainage (NRD). While metals such as aluminum, lead, iron and copper are relatively immobile under neutral conditions, other metals such as zinc, molybdenum, arsenic, antimony, selenium, 1  cadmium and chromium may be more mobile than under lower pH conditions (the term metal is broadened to include metalloid elements such as arsenic; Price, 2003). The pH-neutral effluent water may still contain a high dissolved load and lead to environmental impacts. There is relatively little literature concerning neutral mine drainage, particularly originating from waste rock. It is for this reason that a comprehensive investigation examining the physical, hydrological, chemical, and biological mechanisms involved in metal release, transport and attenuation in neutral-pH mine effluent is underway at the Antamina mine, Peru.  1.2  THE ANTAMINA MINE  The Antamina mining facilities are operated by Compañia Minera Antamina S.A. (CMA), which is a privately held company owned by BHP Billiton (33.75 %), Xtrata (33.75 %), Teck (22.5 %) and Mitsubishi (10 %). CMA is a multi-billion dollar copper and zinc mining project located in the Peruvian Andes, approximately 270 km north of Lima. Project facilities extend from 4,700 m a.s.l where the conventional open pit mine, milling, and concentrator flotation operations are located down to the sea-level port facilities. The mine is connected to the port via a 302-km concentrate pipeline. Official mining operations began in 2001 and are expected to continue until at least 2020. Processing of the low-/marginal-grade stockpile is expected to occur between 2021 and 2024, and mine closure is scheduled to be completed by at least 2029 (Golder, 2007a). Throughout the life of the mine, an estimated 600 million tonnes of ore will be unearthed, resulting in an additional 1,500 million tonnes of waste rock. The ‘waste rock-to-ore’ ratio is approximately 2.5 (Golder, 2007a).  With several neighbouring village communities relying on the seeps, springs and rivers of the region, Antamina has a responsibility to find suitable practices that minimize the socioenvironmental impact on the surroundings. Financing is provided from several international lenders and requires rigorous compliance with environmental and social guidelines as adhered to by domestic and international regulations.  1.3  PROJECT OVERVIEW  The research in this thesis is part of an extensive collaborative research project among the University of British Columbia, Compañia Minera Antamina S.A., Teck, and the Natural 2  Sciences and Engineering Research Council of Canada. The primary objective of the project is to characterize and understand the processes that control the release, mobilization and transport of metals in NRD from waste rock. The intent of the project is to provide a comprehensive characterization of neutral drainage conditions at Antamina and to understand how to use shorttime and small-scale testing to predict long-time, large-scale field responses. Improved understanding of NRD leaching will allow Antamina to better manage their waste rock and identify, transport and dispose of it in the most cost-effective fashion with minimal environmental impact in short- and long-term timeframes.  The project involves a suite of experiments ranging from mineralogical analyses and microbiological and geochemical laboratory studies to field-scale experimental waste rock piles that are employed to address the processes governing metal and other solute release, transport and attenuation in this neutral-pH drainage environment (Figure 1.1).  1.4  RELEVANCE TO THE ANTAMINA MINE  The responsibility of monitoring, controlling and mitigating the hydrological vulnerability of its waste rock management practices is crticial to Antamina’s sustainability. By the end of 2006, nearly one-third of the 1.5 billion tonnes of waste rock estimated to be generated throughout the life of the mine from 2001 to (at least) 2020 was already removed (Golder, 2007a). Antamina’s waste rock is sorted into three categories of expected leaching behavior on site based upon elemental and mineral assays from drill cores. Classes A, B and C represent waste rock that will typically produce leachate with solute loads that will require treatment, may require treatment and won’t require treatment, respectively.  There are two principal waste rock disposal sites at Antamina, the East Dump and the Tucush Dump. The East Dump is by far the largest dump at Antamina and receives all of the Class A material, and some Class B and C material. The Tucush Dump receives only Class B and Class C waste rock material because it is situated within a partially-karstic valley that has a large potential for rapid solute transport. Both the East and Tucush Dumps are located within different drainage basins.  3  Several methods are utilized by the mine to re-route, capture and treat a majority of mine effluent water. The East Dump is located upon lower permeability strata that directs most of the poor quality drainage water to a tailings pond. Drainage from the Tucush Dump is expected to be more benign, and is therefore routed through a constructed wetland to attenuate dissolved solids and solutes. Due to a significant variability in Class B waste rock, there is a level of uncertainty surrounding the drainage quality of Tucush effluent and wetland performance. To address this issue, the first test pile in this neutral-pH drainage study was constructed using Class B waste rock.  With better understanding of the metal release and attenuation mechanisms in the Class B material, the mine will be able to manage its waste rock in a more effective manner. This may involve directing more or less Class B waste rock to the Tucush Dump to ensure that the effluent compliance levels are met and the potential impacts of waste water are mitigated.  1.5  THESIS OBJECTIVES  This thesis characterizes the initial 2 years of hydrological and geochemical data obtained from the first experimental waste rock pile (test pile 1) that was built (Figure 1.1). This pile was constructed with ‘run-of-the-mine’ Class B marble and hornfels waste rock using the same conventional end-dumping method that the mine employs to dispose of the waste rock. Constructed in 2006 by Corazao Gallegos (2007), this 36 x 36 x 10 m (high) pile was designed at an intermediate scale between the laboratory and field cell experiments, and the existent mine waste rock piles. Aranda (2009) examines the leaching characteristics of eight field cells constructed with material from test pile 1. Pile dimensions were decided based on the tradeoff between the largest size to ensure results that are representative of full-scale piles and the most cost-effective way in which to construct the pile. The piles represent the largest sized-experiment within the context of this comprehensive project. Test pile 1 results will be used to address the following research questions:  1. What are the principal mechanisms that control the hydrology, chemistry and solute mass loadings in this Class B waste rock test-pile?  4  2. How can results from this experiment be related to those from other experiments at different scales and related to the existing mine waste rock piles? 3. How can results be used to improve management of the Tucush Dump? 4. How can results be used to improve short-term mitigation and long-term mine closure strategies?  Although results from the first two years of pile operation do not provide long-term trends in pile behaviour, they are useful in characterizing the initial pile flushing and hydrogeochemistry, and will help in understanding which flow processes are occurring within the pile under certain conditions. This study will be useful in subsequent scale-up investigation and modeling efforts, and can be combined with results from several previously published reports related to this comprehensive project (Corazao Gallegos, 2007; Corazao Gallegos et al., 2007; Bay et al., 2009; Aranda, 2009; Aranda et al., 2009; Conlan, 2009).  1.6  LITERATURE REVIEW  Previous studies have investigated waste rock-water interactions, reactions and weathering rates that control acid generation, sulphate and metal release from mine waste environments. Smith and Beckie (2003) contains a comprehensive summary of hydrological and geochemical transport processes in waste rock, whereas Nichol (2002), Wagner (2004), Corazao-Gallegos (2007), Marcoline (2008) and Neuner (2009) contain detailed reviews of the processes that lead to poor-quality drainage from waste rock environments. The following section examines several important processes that contribute to solute mass loadings in waste rock, with particular attention given to the physical properties of waste rock that determine the nature of and chemical composition of the flow.  1.6.1  What is Waste Rock?  Two principal forms of waste are generated from mining activities: mine tailings and waste rock. Tailings are the fine-grained material left over after mineral processing, whereas waste rock is the low-grade overburden material that surrounds an ore deposit and must be removed in order to access it. There has been a considerable amount of research pertaining to mine tailings (Wunderly et al., 1996; Blowes et al., 1998; Moncur et al., 2005; Petrunic et al., 2005) and heap 5  leach processes (Brierley and Brierley, 2000; Olson and Clark, 2008; Webb et al., 2008), yet relatively less is known about mine waste rock and the mechanisms responsible for metal release, attenuation and transportation.  Waste rock is produced by the crushing force of blasting, and results in a wide variety of particles that range in size from clay to large boulders. Most waste rock is placed in large disposal sites located near the open pit where they typically remain indefinitely. As Nichol et al. (2005) describe, these waste rock piles are generally thick unsaturated zones tens to hundreds of meters high, and are composed of physically and chemically heterogeneous materials containing a large variability in size, texture, permeability, macropores, fractures and capillary breaks. The internal structure of a waste rock pile is largely dependent on the method in which the material is placed, which itself is dependent on site conditions and equipment availability (Bellehumeur, 2001; Fala et al., 2003).  1.6.2  End-Dumping and Internal Structure  A comparison of different waste rock disposal methods is found in Corazao (2007), the two most conventional being push- and end-dumping. Push-dumping involves dumping material in lifts at the top of a pile and a bulldozer pushing it level whereas end-dumping involves dumping material directly across the crest. Waste rock is segregated in both methods and leads to different internal structures which leads to different flow and chemical transport pathways throughout the pile. Push-dumping generally results in a coarse lower zone and a non-uniform upper zone with horizontal traffic surfaces between lifts. In end-dumping, the conventional waste rock disposal strategy employed at Antamina, material segregation occurs upon the tipping of waste rock from the truck as cobbles and boulders accumulate near the bottom of the slope and silts and sands tend to settle near the top of the pile (Fala et al., 2005). As illustrated in Figure 1.2, this results in a heterogeneous pile that is highly anisotropic with spatially variable hydraulic conductivity (Saretzky, 1998).  An intermediate region is often present between the upper finer zone and the lower coarser zone, and characteristically includes evenly distributed material (Morin et al., 1991). Discontinuous inter-bedded inclining layers of fine- and coarse-grained material, approximately dipping at the 6  angle of repose (~37°), were observed in several waste rock piles as stated by Fala et al. (2005), Stockwell et al. (2006) and Azam et al. (2007). These layers result from the preferential settling (or deposition) of coarser materials underneath finer-grained materials that settle at a slower rate and may lead to flow focusing (Figure 1.2).  The compaction of the top of the pile under the traffic of heavy equipment causes a finer grainsized, lower-permeability traffic layer in which infiltration is much slower compared to the coarse-grained, unconsolidated pile batters around the edge of the pile (Fala et al., 2005; Bellehumeur, 2001). At the Cluff Lake mine, Saskatchewan, Canada, Bellehumeur (2001) found that an untrafficked waste rock surface consisting of loose gravel, cobble-sized waste rock and a few fines encountered almost immediate infiltration at the surface under the most extreme rainfall events, whereas a finer, compacted traffic surface led to ponding and overland flow. These led to dramatic differences in the hydrological and geochemical regime of the pile and to stratification or channeling of the flow of water.  1.6.3  Waste Rock: Physical Properties  When waste rock is exposed to atmospheric oxygen and precipitation, it may undergo a series of (bio)geochemical reactions that can potentially release contaminants such as heavy metals to the environment. Physical heterogeneities in waste rock (e.g., particle size, texture, lithology, chemical composition, stratification, channeling, segregation, sorption, and permeability) impact the rate, direction and uniformity of flow through waste rock. These flow differences will directly affect the chemical composition and solute mass loadings of the effluent water. In general, the two main factors that control the fluid flow and chemical transport in waste rock are the particle size distribution and texture (Smith and Beckie, 2003).  Waste rock typically exists within a range of grain sizes that spans almost six orders of magnitude, between 1 μm and 1 m, depending on the different processing and depositional methods (Fala et al, 2005). This physical heterogeneity makes the understanding and prediction of flow through unsaturated waste rock material highly complex. To define the relative proportion of fine and coarse fractions in a granular matrix, particle size distributions (PSD) are often used. PSD’s may help indicate the tendency of certain types of flow to occur in waste rock, 7  as water retention generally increases as the particle size and porosity decrease (Aubertin et al., 2003; Fala et al., 2005).  Waste rock texture is a primary control on the infiltration capacity, sorptivity and geometry of fluid pathways. It is typically granular in texture due to the blasting process and often ranges from zones with a fine-grained matrix enveloping large cobbles and boulders to matrix-free clast-supported zones (Nichol et al., 2005). If waste rock consists of a relatively large proportion of fine material (e.g., has greater than 20 % fines less than 2 mm in diameter), it creates a matrix in which the larger cobbles and boulders are likely embedded (Dawson and Morgenstern, 1995; Smith and Beckie, 2003). The waste rock pile is referred to as acting in a ‘soil-like’ manner (Smith and Beckie, 2003). Conversely, if the waste rock consists of a relatively small proportion of fine material (e.g., less than 20 % fines), the fine-grained material in combination with an assortment of larger particles may creates large voids and connected macropores. This type of waste rock is referred to as ‘rock-like’ (Smith and Beckie, 2003). The flow of water is more channeled in rock-like material and dominated by matrix flow in soil-like material, with implications for discharge quality and quantity.  1.6.4  Waste Rock: Flow Processes  The hydraulic properties of waste rock are a function of the physical properties of the waste rock material. In the finer-grained matrix where capillarity controls flow, this relationship can be expressed by the soil water characteristic curve and the hydraulic conductivity, which may indicate the tendency of one type of flow to occur over another, which has implications for solute mobilization. To understand the processes that lead to solute mass loadings, waste rock flow processes are examined.  1.6.4.1 Soil Water Characteristic Curve A soil water characteristic curve describes the relationship between particle size, soil moisture content and/or soil water suction (i.e., matric or capillary forces). Yazdani et al. (2000) describe that capillarity is related to the size of the particles within a structure, and that particles with a diameter > 4.75 mm (#4 sieve) exhibit little capillarity regardless of whether they are under loose or dense arrangement. Material < 4.75 mm in diameter will typically exhibit slow and diffuse 8  matrix flow whereas material with a grain size distribution containing less than 35 % fines (< 4.75mm diameter), rapidly loses its ability to retain water under tension and would typically exhibit macropore-like flow. Herasymuik et al. (1995) use the same distinction to characterize between fine- and coarse-grained materials and serves as a suitable reference to distinguish between these material classes. Yazdani (1995) presents a detailed account of how moisture content (including antecedent moisture and water retention), pore geometry and particle size changes can affect the hydraulic properties of the pile and thus the hydraulic conductivity.  1.6.4.2 Hydraulic Conductivity In unsaturated waste rock conditions, hydraulic properties are highly dependent on the physical waste rock properties which dictate the amount and connectivity of water-filled pore spaces (Fala et al., 2005). Hydraulic conductivity varies with location, due to depositional processes, and in time, due to variations in saturation and mineral precipitation-dissolution (Morin et al., 1991). Water retention generally increases as particle size and porosity decrease, yet the saturated hydraulic conductivity tends to increase with the average size of the particles (Aubertin et al., 2003; Mbonimpa et al., 2002). However, very large boulders (i.e., > 1 – 2 m in diameter) do not have a bulk porosity and do not contribute to the overall bulk porosity of a waste rock pile (Yazdani et al., 2000; Fala et al., 2003; Neuner, 2009). Due to a large variability in particle size, the infiltration capacity can accordingly vary by several orders of magnitude as explained by Bellehumeur (2001), who found that steady state infiltration was quite different from rapid infiltration at different locations in a waste rock pile, suggesting that a bi-modal permeability was present.  Much like particle size is highly variable in waste rock piles, so too is hydraulic conductivity, which is a function of the particle size and orientation. Morin et al. (1991) state that a single conductivity value cannot be used to accurately represent an entire waste rock pile, and that a range of values should be considered to account for the variations in particle size, rock characteristics, and construction techniques. It is for this reason that highly variable systems are often attributed multi-modal properties (Gerke and van Genuchten, 1993; Diodato and Parizek, 1994; Flury et al., 1994; Tuller and Or, 2002). For simplicity, the hydraulic conductivities for waste rock can conceptually be divided into two types of flow: 9  1. matrix flow, and 2. preferential flow.  1.6.4.3 Matrix and Preferential Flow In most unsaturated waste rock, flow is either in the form of matrix or preferential flow (Figure 1.3). In a soil-like waste rock in which a significant proportion of finer-grained particles are present, the fine-grained matrix is generally able to retain and transport water under tension or capillary forces. This is referred to as matrix flow. Conversely, in a rock-like waste rock in which a large fraction of coarse material is present, the dominant flow mechanism may be via non-capillary or gravity-driven preferential flow. Each type of flow can be described using different physical and chemical properties such as porosity, particle size, hydraulic conductivity, matric suction, etc.  Matrix flow typically experiences slow and diffuse, matric (and gravity) force-driven capillary flow throughout both wet and dry periods whereas the coarser grained waste rock will encounter rapid, gravity-driven preferential flow that is generally activated under high moisture conditions. Therefore, matrix flow is typically the dominant flow mechanism in waste rock (Wagner et al., 2006; Neuner, 2009). The fine-grained matrix material has the ability to retain a larger volume of water under matric forces whereas the coarser-grained materials typically retain a lower volume of water.  When the matrix is fully saturated, the specific discharge of water q [m/s] through the material can be described by Darcy’s Law, which relates the hydraulic conductivity K [m/s] of a saturated porous medium to the hydraulic gradient, 𝜕𝜕ℎ�𝜕𝜕𝜕𝜕 [unitless] in order to determine the specific discharge as described in the following equation:  𝜕𝜕ℎ  𝑞𝑞 = −𝐾𝐾 � 𝜕𝜕𝑙𝑙 �  [1-1]  This equation describes the one-dimensional specific discharge of water and assumes that the material is fully saturated and that the flow is laminar. Under unsaturated conditions of waste 10  rock, the vertical, gravity-driven infiltration flux through the unsaturated zone is a function of matric pressure ψ [kPa] or volumetric water content ϴ [unitless] and can be expressed as K(ψ) or K(ϴ) as summarized in Richard’s Equation (1931): 𝑑𝑑𝑑𝑑  𝑞𝑞𝑧𝑧 = −𝐾𝐾(𝜓𝜓) � 𝑑𝑑𝑑𝑑 + 1�  [1-2]  While matrix flow is subject to Darcy’s Law, preferential flow may or may not be, depending on the type of flow (Figure 1.3).  1.6.4.3.1 Preferential Flow Under certain conditions, clast-supported regions of the pile can often create large voids that may be filled under heavy infiltration events. This can lead to water by-passing the matrix under high saturation conditions and can result in rapid slugs of relatively fresh water penetrating much quicker than flow through the matrix (Nichol, 2002; Smith and Beckie, 2003). Preferential flowpaths generally occur due to collections of connected pores that concentrate flow to certain pathways, and typically result in large variations in the timing, magnitude and volume of the flow response at the base of a waste rock impoundment (Corazao Gallegos, 2007). Since matrix flow is much slower, preferential flow will undoubtedly lead to differences in residence time and can greatly impact water quality. According to Smith et al. (1995), water can infiltrate the entire depth of an experimental waste rock pile in times from within only a few hours to several years, indicating that most waste rock piles have spatially variable hydraulic properties due to variations in moisture content, grain size distribution and/or mineralogy (Fala et al., 2005; Smith and Beckie, 2003). Based on the occurrence of frequent experimental observations of this preferential flow in unsaturated porous media, Flury et al. (1994) explain that this phenomenon is commonly believed to be the rule rather than the exception.  Preferential flow consists of macropore flow, which describes the rapid flow of water through large connected voids, and capillary barrier-flow, which describes the slower movement of water around fine-grained material via capillary contact. Preferential flow likely most often occurs due to capillary barriers that cause flow focusing. Capillary barrier-flow is subject to Darcy’s Law whereas macropore flow is not (Figure 1.3). 11  1.6.4.3.2 Capillary Barrier Evidence of the preferential flow of water along fine-grained lenses or around large boulders through capillary contact has been observed in several investigations (Diodato and Parizek, 1994; Bellehumeur, 2001; Bussiere et al., 2003; Fala et al., 2005; Stockwell et al., 2006). Under conditions where interbedded fine- and coarse-grained layers are present (i.e., as created from the end-dumping of waste rock), a capillary barrier may be formed in which water may be preferentially retained in the finer-grained material due to capillary forces. This may result in the preferential flow of water that is focused to other regions, likely having implications on solute loadings.  1.6.4.3.3 Macropore Flow The flow of water through large connected void spaces that is typically activated under high infiltration events and responsible for rapidly channeling of water much faster than the capillarydominated matrix-flow velocities (up to 5 m over a few hours) is referred to as macropore flow (Germann and Beven, 1981; Eriksson et al., 1997; Newman, 1999; Smith and Beckie, 2003; Nichol et al., 2005; Stockwell, 2006; Marcoline, 2008; Neuner, 2009). Macropore flow is also referred to as non-capillary flow. Due to a higher infiltration rate and subsequent lower residence time, this type of flow typically consists of a lower dissolved load than matrix flow.  1.6.5  Waste Rock: Chemistry  Fala et al. (2005) conclude that the flow of water, transport of aqueous components, and dissolution/precipitation of solid minerals are closely coupled processes in mine waste environments. In the presence of reactive waste rock, these processes often interact in complex ways to generate poor quality drainage. Based on this, waste rock effluent chemistry is ultimately determined by the physical waste rock properties (PSD, texture, structure, mineralogy) and flow through this material.  The leaching of metals and generation of acidic drainage are both common features in waste rock where sulphide minerals are present. Pyrite is the most common sulphide mineral in most mine environments, the oxidation of which has been very well documented (Morin et al., 1991; Pantelis and Ritchie, 1992; Ritchie, 1994; Jambor and Blowes, 1998). The amount of, and time 12  in which, water is in contact with a sulphide mineral surface (e.g., pyrite) is a significant controlling factor that affects water quality. The release of metals is a complex process that is primarily dependent on the waste rock mineralogy and presence of water and oxygen. Precipitation events lead to the highly variable vertical flow of water through a waste rock pile. Just as the presence of water is a critical component in characterizing pile behaviour, so is the distribution and movement of pile gases such as oxygen. The movement of oxygen in waste rock piles may be influenced by temperature gradients between the pile surface (subject to ambient temperature) and pile interior (mainly due to heat released from pyrite oxidation). While gas transport typically occurs through diffusion, it may be enhanced through thermally-driven gas convection if the material is sufficiently permeable (Lefebvre et al, 2001). Therefore, both oxygen transport and temperature have strong influences on the rate at which geochemical processes proceed.  1.6.5.1 Residence Time: Impact of Flow on Chemistry There is a general relationship between outflow volume and chemistry in waste rock (Wagner et al., 2006). Slow-moving water through the fine-grained matrix will typically have a higher dissolved load because it has a higher contact time with a greater surface area compared to faster-flowing water through macropores, which is generally associated with rapid slugs of fresher water (Smith and Beckie, 2003). Assuming no distinct chemical differences between material, the finer-grained material will typically undergo the most significant weathering. For example, Strömbererg and Banwart (1999) found that particles < 0.25 mm contributed to nearly 80 % of the sulphide and silicate dissolution at the Aitik Copper mine, Sweden.  Under relatively wet conditions, larger macropores can preferentially flush and channel water rapidly through the unsaturated pile, the shorter residence times generally result in lower dissolved loads. During dry periods, water transport is slow and predominantly exists as matrixflow and results in a higher dissolved load (Wagner et al., 2006). Effluent water draining from the base of the pile typically represents both types of flow, but can be dominated by one flow or another depending on site conditions, the antecedent moisture content and time of year (Smith and Beckie, 2003; Nichol et al., 2005).  13  1.6.6  Challenges in Predicting Weathering from Particle Size and Flow Indices  During the deconstruction of a 12-m high waste rock dump,  Stockwell et al. (2006)  characterized the physical properties that control the flow of water by measuring soil water suction, volumetric water content, grain size distribution and mineral weathering at various locations within the pile. The goal was to relate these physical and chemical characteristics to the tendency for preferential flow and mineral weathering to occur based on the relative fraction of fine-to-coarse grained material. A large fine-grained fraction by weight may indicate a tendency for matrix flow to occur within the pile (more weathering), whereas a relatively coarse distribution may indicate a tendency for macropore-dominated flow to occur more frequently (less weathering). However, Stockwell et al. (2006) was unable to establish a relationship between physical flow indices and chemical weathering characteristics.  1.6.7  Waste Rock: Neutral Drainage  Whereas several investigations have examined the acidic rock drainage of mine environments, relatively less is known about the neutral rock drainage. Blowes et al. (2007) report that neutralpH conditions exist at several tailings impoundments and open pits. Neutral-pH drainage can result provided there is a sufficient source of acid buffering potential in the gangue minerals to counter the acid generated by sulfide mineral oxidation. Neutralization often occurs in three distinct pH-buffering plateaus that correspond to the dissolution of carbonates (circum-neutral), hydroxides (acidic) and silicates (very acidic) (Blowes and Ptacek, 1994; Moncur et al., 2005). Several studies have examined these pH-dependent acid buffering tendencies in mine tailings, with particular emphasis on laboratory investigations pertaining to the secondary mineralization that forms coatings around mineral surfaces (Petrunic et al., 2006; Moncur et al. 2005; Sherlock et al., 1995; Jambor and Blowes, 1998; Blowes et al., 1998; Lin and Herbert, 1997; Davis and Ritchie, 1986, Ritchie, 1994; Jambor and Blowes, 1998). Other laboratory investigations have focused on the neutralization potential of acidic mine waste water (Sobek et al., 1978; Sherlock et al., 1994, Strömberg and Banwart, 1998). There is limited work pertaining to neutral drainage from waste rock.  Several waste rock experiments across different scales have investigated the mechanisms that lead to acid mine drainage. Only a limited number of experiments at the laboratory scale have 14  examined the mechanisms that lead to neutral drainage from waste rock. No comprehensive hydrological and geochemical investigation into neutral-pH waste rock drainage from a fieldscale test-pile has yet been carried out.  1.6.8  Challenges in Waste Rock Investigations: Scaling Issues  The mechanisms that lead to mineral weathering and acid generation are highly dependent on scale. Since it is never feasible to have detailed characteristics of a given waste rock, various waste rock leaching methodologies spanning several orders of magnitude from μm to 10’s of m have been established. Results of these experiments consistently indicate a discrepancy between the mineral weathering rates at the different scales (Malmström et al., 2000). Corazao Gallegos (2007) describes waste rock investigations spanning several scales from the laboratory to the field-scale. The intention is to scale-up or extrapolate from lab to field scale in order to generate reliable correction factors that will enable the prediction of loadings from existing waste rock (Otwinowski, 1995; Eriksson et al., 1997; Malmström et al., 2000; Frostad et al., 2005).  Laboratory settings provide the most manageable, cost-effective may in which to conduct physical, hydrological, chemical, mineralogical and biological experiments on waste rock in a controlled setting. Laboratory experiments are somewhat biased because they do not subject the material to actual environmental conditions. On the other hand, field cells are larger than most laboratory tests and are constructed in the field so that they are exposed to the same atmospheric conditions of the existent waste rock pile. These experiments still demonstrate limitations in predicting metal leaching from the exiting waste rock piles based on their limited particle size distribution (e.g., less than 10 cm) and loadings rates, but are useful for validating leaching release rates obtained from laboratory tests (Frostad et al., 2005). Consequently, field-scale test piles are typically constructed with ‘run of the mine’ waste rock and exhibit trends most similar to the existing waste rock piles.  There is a large discrepancy of several orders of magnitude between the weathering rates reported in the lab and field (Malmström et al., 2000). Empirical, site-specific correction factors can possibly be determined but cannot be applied at different sites. These differences are most likely due to the variation in particle sizes between the different experiment scales as lab 15  experiments may only range over 2 - 3 orders of magnitude but field experiments may vary by orders of 5 or more. Since surface area decreases with larger particle sizes, the relative abundance of surface site reactions will likely be higher in laboratory experiments.  Due to the complexities involved in characterizing the hydrological and geochemical mechanisms involved in mine waste rock, various 1-, 2- and 3-D flow, transport, and coupled reactive transport models have been generated to better understand waste rock environments (e.g., Lefebvre, 1994; Lefebvre and Gelinas, 1995; Mayer et al., 1999; Sracek et al., 2004; Linklater et al., 2005). Wagner (2004) contains a detailed review of the various models. However, because it is difficult to obtain accurate predictions based solely on numerical models, field and lab experiments that take place over several years are required to make advances towards a more complete conceptual framework.  1.6.9  Waste Rock Test Piles  Numerous investigations of operational waste rock piles at various mines demonstrate that the physical, hydrological and chemical characterization using conventional subsurface methods is often highly complex and leads to large uncertainties in estimating and predicting pile behaviour (Smith et al., 1995; Herasymuik, 1995; Eriksson et al., 1997; Price, 2003; Lefebvre et al., 2001; Sracek et al., 2004; Linklater et al., 2005; Milczarek et al., 2009). The exact flow paths and locations of obstacles encountered along the way, such as a large boulder or a low-permeability layer, are largely unknown and only hypothesized based on inferences from the data. This will also have strong implications on the porewater chemistry. Consequently, these challenges have led to the construction of several field-scale test-pile experiments in various climates throughout the world (e.g., Table 1.1).  Field-scale waste rock piles provide a necessary intermediate link in scale between laboratory and field cell experiments and the full-scale waste rock piles. There are no well-established criteria for waste rock test pile construction. In general, the design of each pile is determined by a balance between practicality and cost. Ideally, each pile is large enough to sufficiently capture all of the field-scale processes while meeting reasonable budget constraints. For example, while the differences between matrix and macropore flow may be evident in a smaller test pile 16  (perhaps 2 – 3 m high), gas transport dynamics that typically manifest in larger-scale operational waste rock dumps may not be. In these large dumps, oxygen may not be able to penetrate to the centre of the pile whereas a smaller pile may be well-aerated and not entirely representative of the operational dump (Pantelis and Ritchie, 1992). Therefore, all of these considerations are required when designing a waste rock test pile.  1.7  ORGANIZATION OF THESIS  This thesis examines the first two years of data collected from waste rock test pile 1 at the Antamina mine. Relevant data from other parts of the project will be referred to and, if appropriate, compared to the test pile data. Final integration of the test pile data and data from other studies is a future task in the Antamina research project.  This chapter summarized the project scope and motivation and reviewed the pertinent literature about waste rock hydrology and chemistry. Chapter 2 examines the site conditions of the Antamina mine, project design and test pile construction, and physical and chemical properties of the waste rock used in test pile 1. Chapter 3 examines the hydrological characterization of test pile 1, with emphasis on the outflow rates and flow mechanisms. Chapter 4 examines the waste rock effluent chemistry and Chapter 5 integrates hydrology and chemistry with particular emphasis placed on the evolution of the solute mass loadings. Chapter 6 synthesizes selected data from Chapters 2 to 5 to highlight the main mechanisms that contribute to solute mass loadings of test pile 1. Chapter 7 summarizes the major findings and conclusions from each chapter. Future recommendations to improve this investigation and significance for the Antamina mine are also discussed.  17  TABLES  Table 1.1. Comparison of recent waste rock test piles. Test Pile Site  Location  Cluff Lake Mine  Saskatchewan, Canada  Key Lake Mine  Saskatchewan, Canada  Antamina Mine  Ancash Department, Peru Northwest Territories, Canada New South Wales, Australia Papua Province, Indonesia  Diavik Mine Cadia Hill Mine Grasberg Mine  Authors  Pile Dimensions [m] (length x width x height) 8x8x5  Bellehumeur, 2001 Nichol et al., 2005 Wagner et al., 2006 Marcoline, 2008 Carey et al., 2005 25 x 25 x 12 Stockwell et al., 2006 Corazao Gallegos et al., 2007 36 x 36 x 10 Neuner et al., 2009  60 x 50 x 14  Rhode & Williams, 2009  ~ 200 x 35 x 15  Andrina et al., 2003  480 x 80 x 20  18  FIGURES  Figure 1.1. Neutral drainage waste rock project. Components relevant to this thesis are shaded in blue.  Figure 1.2. Waste rock pile schematic. Modified from Fala et al. (2005).  19  Flow in Waste Rock  Preferential Flow  Matrix  Darcian  Non-Darcian  Capillary Barrier  Macropore  Figure 1.3. Flow processes in waste rock.  20  Darcian  2  2.1  SITE DESCRIPTION, INSTRUMENTATION, SAMPLING AND MATERIAL CHARACTERIZATION LOCATION  The Antamina mine is located, specifically, at 9°32’ south and 77°03’ west (approx. UTM 274 000 east and 8 945 000 north) at over 4,200 m above sea level, and roughly 270 km north of Lima, Peru (Redwood, 1999; Lipten and Smith, 2004). Situated near the mine is the Huascarán National Park, which is designated as a United Nations Education, Scientific and Cultural Organisation (UNESCO) World Biosphere Reserve and World Heritage site (Figure 2.1).  2.2  CLIMATE  The mine does not reflect a typical tropical or high altitude climate. Mean annual temperature is between 5.5 and 6.0 °C and mean annual precipitation is between 1200 and 1300 mm. The area surrounding the mine is unique in that it constitutes the highest elevation tropical mountain setting in the world. The climate is typically cool and moist throughout the year and bears similarity to an alpine tundra ‘ecozone’, which is characteristic of shrub and grassland vegetations generally found above tree line. Weather at Antamina is governed by the tropical rain belt, which oscillates about the northern and southern tropics creating distinct ‘wet’ and ‘dry’ periods that respectively fall between October – April and May – September. Since meteorological data was first recorded at Antamina in 2000, the maximum reported daily precipitation is 36 mm and the hourly temperatures range between -4 and 23 °C.  2.3  REGIONAL GEOLOGY  On the western coast of South America, the oceanic Nazca Plate is being subducted towards the east under the continental South American Plate. This process of subduction has led to significant folding, faulting, uplift, volcanic and plutonic processes that are responsible for the mineral-rich Andean mountain belt that extends along the western coast of the continent from the southern tip of Chile and Argentina up to Colombia and Venezuela (Figure 2.1).  The Central Peruvian Andes consist of the Western and Eastern Cordillera ranges. The Western Cordillera consists of two smaller ranges: the Cordillera Negra (to the west) and the Cordillera Blanca (to the east). The Antamina deposit is situated along the eastern edge of the Cordillera 21  Blanca, which is part of a polymetallic belt transecting Peru from 6° to 14° south (Figure 2.1). The area consists of high grade Zn-Pb-Ag-Cu-Au mineralizations with significant Cd, Bi, Mo and As associations (Lipten and Smith, 2004; Love et al., 2004).  The Cordillera Negra and the coastal zone to the west form a magmatic arc that was active from the Late Jurassic to the Tertiary. Further east, thick marine sediments consisting of slates, quartzites, sandstones, shales and coal with marine limestone were deposited from the Late Jurassic to the Late Cretaceous to form the Western Trough (Lipten and Smith, 2004). In the mid-Cretaceous, a marine transgression deposited thick marine carbonates followed by marine shales (113 – 88 Ma and 88 – 84 Ma, respectively; Lipten and Smith, 2004). The Jumasha and Celendin Formations developed during the carbonate and shale depositions, respectively. Later, a multi-phase quartz monzonite porphyry batholith intruded into the Cretaceous-aged sedimentary limestone deposits of the Jumasha Formation, creating the Antamina skarn deposit 5 – 16.5 Ma (Lipten and Smith, 2004; Love et al., 2004). Quartz monzonite is a granitic rock typically comprised of a quartz, plagioclase and potassium-feldspar assemblage that is resistant to weathering.  2.4  DEPOSIT GEOLOGY  The Antamina skarn deposit is rich in copper and zinc and is the source of the Antamina mineral reserve. Subsequent hydrothermal alteration to the Cu-Zn skarn body introduced copper orebodies at the limestone contact, with zinc, lead and bismuth also found in contact with limestone, in addition to metamorphic marble and hornfels (Evans et al., 2005). Ore formation may have been limited due to the fine-grained composition of the hornfels, which results in a low apparent porosity and permeability and, hence trace sulphides consistently below grade level (Lipten and Smith, 2004; Love et al., 2004).  Skarn deposits usually form by a chemical metasomatism of rocks at the contact zone between magmatic granite intrusions into carbonate-rich rocks (e.g., limestones). Figure 2.2 illustrates that the Antamina skarn orebody is elongated (3 km x 1 km x 1 km) in a northeast direction and is situated in the Antamina valley (Redwood, 1999). Approximately 90 % of the skarn deposit is mineralized and contains considerable amounts of copper, zinc, silver and molybdenum, existing 22  primarily as sulphide minerals. Sulphide zones generally exist at the skarn-limestone contacts and are physically easy to separate. Since the upper reach of the ore body was recently unearthed by Quaternary glaciation, it has not resulted in significant oxidation or enrichment.  Pyrite, chalcopyrite and magnetite are ubiquitous throughout the orebody, with a lesser presence of pyrrhotite and sphalerite. The deposit consists of several mineralization or metal zonations going from the outer layers to the outer rims (Figure 2.3). Specifically, the outer layers consist of intermediate skarn (Cu-Zn-±Mo-±Bi), green endoskarn (Zn-Cu-Ag-Bi-Pb) and mixed brown (Cu-Zn-Ag-Bi) endoskarn, whereas the outer rim consists of marble (±Zn-±Pb-±Ag) and hornfels (Zn-Pb-Ag-BI) (Lipten and Smith, 2004). In addition, there is a pocket of wollastonitebornite exoskarn (Cu-Zn-Ag-Bi), and the whole deposit is hosted by limestone. Lipten and Smith (2004) and Love et al. (2004) both contain detailed geological descriptions of the history and composition of the lithographic and metallic composition of the deposit.  2.5  HYDROGEOLOGY  The Antamina mine is located on the eastern edge of the continental divide that separates water from draining to the Pacific or Atlantic Oceans. The mine is located within the Amazon basin, upon the upper catchment of the Maranon River which is a tributary of the Amazon River. It is situated between two steeply dipping limestone ridges that act as geographic and hydrologic boundaries. Nearly 10 % of the surface area of the Peruvian Andes consists of Cretaceous-aged karstic limestone in which a highly inter-connected system of seeps, springs, streams and rivers exists throughout the region. Karstic terrain often contains large conduits that can rapidly transport groundwater over a short time period (Evans et al., 2005).  The bedrock geology of the Antamina region consists of:  1. a lower sequence of clasitc sediments consisting of sandstones, quartzites, shales and various carbonates, and 2. an upper sequence of calcareous facies such as limestones and various shales.  23  The calcareous upper sequence is where the karstic terrain is located, and typically results in low runoff and high infiltration situations. The northeast side of the Tucush Valley is underlain by karstic limestone of the Middle Jumasha that diverts water to the adjacent Ayash watershed. The southern flank of the valley consists of Celendin Formation and Upper Jumasha material, both of which are expected to act as hydraulic barriers (Evans et al., 2005).  Satellite image and air photo interpretation, field mapping, facture trace analysis, karst surface and spring mapping, and dye-tracing studies have all been carried out to characterize and present a conceptual model of the terrain surrounding Antamina (Evans et al., 2005). At Antamina there are two main waste rock dumps, the East and Tucush Dumps, and one low/marginal-grade stockpile, all of which generate poor quality drainage (Figure 2.4). The East Dump is located in the Huincush Valley, to the east of the open pit and to the west of the concentrator and tailings pond. It is composed of both reactive and non-reactive material, and resides upon less-permeable clayey soils that direct most of the drainage water to the tailings pond. The Tucush Dump is located within the Tucush Valley (4.5 km long, 400 – 600 m deep) and receives non-reactive (Class C) and possibly reactive (Class B) waste rock, which is placed on the southern, lesspermeable Celendin Formation and Upper Jumasha regions of the valley.  2.6  DEPOSIT DEVELOPMENT  The terrain surrounding Antamina is composed of rugged, steeply dipping limestone peaks, glacial valleys and lakes and deep canyons. The region is littered with fossils and ruins, showing a rich history of previous inhabitants. The word Antamina stems from the Quechua words of ‘anta’ and ‘mina’, which when translated to English mean ‘copper’ and ‘mine’, respectively. It is believed that copper was mined in the vicinity of Antamina in pre-Colonial times, but the first records of small-scale mining in the area date back to the 1850’s. The first significant exploration effort was carried out nearly a century later by the Cerro de Pasco Corporation in the early 1950’s, who estimated the Antamina grade to be 3 % or approximately one million tons of copper (Redwood, 1999; Love et al., 2004).  Property rights of the mine changed hands several times between the 1950’s and the 1990’s until 1996 when they were finally privatized, leading to the formation of Compañia Minera Antamina 24  S.A. Feasibility studies increased the deposit’s original reserve estimates almost 4-fold. In 2008 the most recent plan indicated that Antamina has a total of 617 million tons of estimated reserves, with 1.23% Cu, 1.03% Zn, 0.029% Mo and 13.7g/t Ag. The amount of ore processed daily is expected to increase from the current 87,000 to a sustained 104,000 tonnes per day. With the newest facilities, the maximum daily capacity of processed ore will be as high as 130,000 tonnes per day (Golder, 2007a).  2.7  WASTE ROCK GENERATION AT ANTAMINA  Mineral ore deposits such as Antamina’s quartz monzonite porphyry intrusion exist in transitional zonations of varying ore grades that decrease with distance from the contact zone (Figure 2.2; Figure 2.3). A significant amount of overburden material must be removed in order to access the ore deposit. High-grade ore is sent for processing, whereas material with low- to marginal-grade ore proportions is stockpiled for processing at a later date. The uneconomical overburden material is considered waste rock and is commonly placed near the open pit in large, multi-layered end – dumped waste rock piles.  All of the excavated material at Antamina originates at the open pit mine. Material is blasted apart to allow the mining equipment to gain access to the ore. Based on previously assayed core samples, the blasted material is either processed, stockpiled or disposed of as waste rock. High grade ore material is sent to the primary crusher and SAG (Semi-Autogenous Grinder) mill. At the concentrator, the material is pulverized by a series of ball-mills and then subjected to a series of hydrochemical flotation processes that separate most of the mineral from the host rock. Mineral concentrate is then sent along a 302 km-long pipeline to the coastal port of Huarmey, whereas the remaining effluent and tailings are piped to the tailings pond where it is retained by a 300-m + high tailings dam. Marginal- and low-grade ore is stockpiled for later processing (Golder, 2007a).  Waste rock represents an obstacle in mineral extraction and is costly to transport away from the point of extraction. It also has a significant capacity to release a large amount of metals and other elements to the surrounding environment. More reactive waste rock is sent to the East Dump,  25  whereas less reactive waste rock with a lower metal content can either be sent to the Tucush dump or to the tailings dam where it will be used to fortify and extend the tailings dam.  2.8  WASTE ROCK CLASSIFICATION  Waste rock at Antamina is categorized into 3 general classes (A, B and C) based on rock type, chemical composition and visual attributes (Table 2.1). The most reactive material is Class A, whereas the least reactive is Class C. Class B represents an intermediate reactive material (Golder, 2006).  Of a total projected 1.5 billion tonnes of waste rock estimated to be produced during the life of the mine from 2001 to (at least) 2020, it is estimated that 44 % will consist of Class A, 42 % will consist of Class C and 14 % will consist of Class B.  2.9  TEST-PILE DESIGN  As part of the neutral waste rock drainage study at Antamina, five 36 x 36 x 10 m (high) waste rock test-piles were built between 2006 and 2009, all of which were based on the same design. The construction of the first pile (2006) using Class B material is outlined in detail by Corazao Gallegos (2007). Appendix A includes design plans and photographs of pile construction and instrumentation. The second and third test piles (2007) were built using Class A material with a few modifications from the first test pile (Appendix B). The fourth and fifth test piles (2008) were also built in the same manner. This thesis is based on the data collected from the first two years of operation of the first test pile, referred to as test pile 1.  Each pile consists of a large basal lysimeter with three sub-lysimeters within the basal lysimeter area. The lysimeters collect all drainage from the pile and convey it through piles to an instrument hut where the flow rate is metered, samples can be collected and the temperature and electrical conductivity are continuously monitored. A number of instruments have also been installed within each pile along a six so-called instrument lines. The detailed description of pile lysimeters and instrumentation follows next.  26  2.9.1  Site  The experiment site was prepared by constructing three platforms from waste rock. The upper, intermediate and lower platforms have approximate elevations of 4385, 4373 and 4371 m above sea level, respectively. The pile base was constructed on the 4373-platform, and waste rock was end-dumped from the 4385-platform (Figure 2.5). All of the instrumentation lines and drainage pipes placed within the pile are connected to the instrumentation hut located on the 4371platform. Between 2-February and 9-July-2006, over 26,000 metric tonnes of waste rock material was delivered to site for the construction of test pile 1.  2.9.2  Base Construction Methods  On the 4373-platform, the base of the pile was built with a 3 % gradient, dipping to the southeast corner. A 0.8 m-high protection berm was conformed around the square perimeter, making the pad 36 (l) x 36 m (w) from apex to apex. A 60-mil High Density Polyethylene (HDPE) geomembrane was installed upon the base to capture and prevent any water from penetrating deeper. This catchment area is referred to as the basal lysimeter. Three (3) smaller 4 x 4 x 0.5 m (high) catchments, referred to as sub-lysimeters, were constructed within the basal lysimeter and also lined with HDPE geomembrane. The sub-lysimeters are located along the centre (east-west) axis of the pile (Figure 2.6). A 4” HDPE pipe conveys water from each of the zero-tension lysimeters to an instrumentation hut for continuous water quality/quantity monitoring. Corazao Gallegos (2007) contains more details about the design, construction and instrumentation of the pile base.  A 0.35 m protective layer of 2B rejected material (< 1.5 “), referred to by the mine as ‘lastre’, was placed directly above the geomembrane and compacted. This material consists of a crushed Class C waste rock (almost pure calcium carbonate) that the mine typically uses for road construction. This geomembrane protective layer is referred to as the 2B-protective layer (Appendix A). Within the basal lysimeter, a series of perforated HDPE drainage pipes were placed upon the protective layer to collect basal drainage water and convey it to the monitoring equipment in the instrument hut. The vertical hydraulic conductivity of the compacted 2Bprotective layer is roughly two orders of magnitude lower than the anticipated vertical  27  conductivity rates in the waste rock layer and is believed to represent a relatively impermeable boundary.  To address the possibility that the compacted lastre material would compromise the chemistry of the waste rock effluent, a sump was installed at the lowest (southeast) corner of the pile so that water samples representative of the 2B-protective layer could be collected and compared with lysimeter data.  To protect the geomembrane, instrumentation and drainage network from the force of waste rock that will be end-dumped from 10 m above, approximately 5,560 metric tonnes of Class B black marble waste rock, stockpiled on site since 2-, 3-February-2006, was used to construct a 1.5 mhigh protective layer upon the 2B-protective layer (Table 2.2). Since the 2B-protective layer is not constructed with waste rock, it is not included in the protective layer discussion. Throughout the remainder of this thesis, the protective layer will only be in reference to the waste rock protective layer unless otherwise stipulated (Appendix A).  In a parallel field-cell study, Aranda (2009) characterized the mineralogical and elemental properties of materials from the protective layer and three subsequent tipping phases of this pile. The results from Aranda (2009) are summarized in Table 2.3 and Table 2.4.  2.9.3  Pile Construction Methods  The waste rock for the pile was end-dumped from the 4385-platform in 3 separate phases, with a pause after each phase for installation of instrumentation lines 1 - 4 along the progressing pile slope (Appendix A). On 10-May-2006, the first tipping phase (Tipping Phase I) took place with the delivery 6,200 metric tonnes of Class B marble diopside waste rock (Table 2.2). With each subsequent payload discharge, the pile progressed outwards from the anterior slope. Along the centre-line of this slope, a 1 - 2 m-deep trench was excavated for the instrumentation installation (discussed below). All instrumentation was manually protected by fine material or small gravel, depending on the type of instrumentation, and then covered with more material for protection.  28  The second tipping phase (Tipping Phase II) occurred on June 27 - 28 and consisted of over 6,000 metric tonnes of Class B marble diopside waste rock (Table 2.2). Two 1 - 2 m-deep trenches were excavated at the front of the pile, one located along the centre and line 3 beginning from the same location and extending laterally to the front-corner (northeast) of the base of the pile. Instrumentation placed within these trenches is discussed below.  The third and final tipping phase (Tipping Phase III) took place on 7-, 8-July-2006 and consisted of almost 8,400 metric tonnes of Class B marble diopside and grey hornfels waste rock (Table 2.2). A 1-2 m-deep trench was excavated along the centre of the pile slope and instrumentation was installed and protected in similar fashion to the previous trenches (discussed below).  2.10 INSTRUMENTATION A total of six instrumentation lines were placed within the pile to characterize the internal pile behaviour (Figure 2.6; Figure 2.7; Figure 2.8). Instrumentation lines 1 - 4 are located within the slope trenches whereas lines 5 and 6 are located within the protective layer. A summary of instrumentation placed in each line is found in Table 2.5, and a discussion of each instrumentation type follows in the following section. GPS coordinates are available for all instrumentation ports except for lines 5 and 6 (Appendix A).  Additional instrumentation was installed in the instrumentation hut to monitor the quality and quantity of water discharging from the base of the pile. Meteorological data was measured on site and throughout the Antamina mine property.  2.10.1 Pile Instrumentation Instruments used in the pile include thermistors to measure temperature, and Time Domain Reflectometry (TDR) probes to measure moisture content. In the instrument hut, tipping buckets and electrical conductivity (EC) probes continuously measure flow and solution EC from the basal lysimeter and three sub-lysimeters. A rain gauge measures site precipitation, although several other rain gauges are located at different places at the mine. All instruments are all connected to a CR1000 datalogger (www.campbellsci.ca). Energy is supplied by a 56 W solar  29  panel. Details of pile instrumentation locations within the pile and installation methods can be found in Corazao Gallegos (2007) and Appendix A.  2.10.1.1 Time Domain Reflectometry Probes Twenty-two (22) Time Domain Reflectometry (TDR) probes were custom-built and installed in the test pile to provide continuous, automated in-situ volumetric water content estimates (Corazao Gallegos, 2007). The TDR probes, capable of accurately measuring moisture content in the high dissolved solid porewater environment expected in the test pile, were built using the design of Young (1998) that was elaborated by Nichol el al. (2002, 2003). These probes are unique because they include a high resistance coating along the centre conductor rod that enables them to function better in the conductive waste rock porewaters (Nichol et al., 2002, 2003). Similar TDR probes have been employed in other waste rock investigations (e.g. Neuner, 2009).  Five TDR probes each were installed in instrumentation lines 1 – 4 (vertical) and one TDR probe was installed in each instrumentation line 5 and line 6 (base) (Table 2.5). Moisture content data provided by the TDR measurements are automated and obtained at 30-minute intervals, and allow for estimates of average wetting front arrival times, variable moisture content and storage fluctuations in response to seasonal variations.  No soil-specific TDR calibration correlating the raw TDR travel time to the in-situ volumetric water content of waste rock in the test pile 1 is available. A soil-specific relationship between similarly-designed TDR probes calibrated in waste rock from the Cluff Lake mine (Nichol et al., 2002, 2005) is used to calibrate the TDR probes at the Antamina test pile. Without a soil-specific calibration, relative changes in moisture content are emphasized over absolute changes. Appendix F includes more information about the nature of the calibration, data processing steps, and other calibration equations considered.  2.10.1.2 Thermistors Thirty-six (36) thermistors (RST Instruments, TH0002 2252 ohm 0.01 °C, www.rstinstruments. com) were installed in test pile 1, 32 of which are located within the pile and an additional 4 are located in the instrumentation hut (one per each lysimeter’s flow-through conveyance system). 30  Six thermistors are located in each of instrumentation lines 1 to 4 (pile) and a total of eight thermistors are located in lines 5 and 6 (base) (Table 2.5). Each thermistor was factory calibrated based on their custom-ordered length (pile: 60 – 21 m; outflow drainage conveyance system: 5 m). Thermistor measurements are automated and obtained at 30-minute intervals.  Temperature data provided by the thermistors is principally used to assess whether heat is emitted during sulphide mineral oxidation and to estimate the extent of temperature-driven gas convection. However, several secondary reasons exist, namely: for geochemical speciation modeling and other calculations (i.e., mineral solubility), neutralization capacity (Sherlock et al., 1995), and to affect temperature corrections for TDR and electrical conductivity.  2.10.1.3 Soil Water Solution Samplers Fifteen (15) soil water solution samplers (Soil Moisture Equipment Corp., www.soilmoisture. com, model #: 1920F1L12B02M2, 12”, 2 bar cup) were installed in lines 1, 2 and 4 (five samplers in each line) to provide point-source water samples at various horizontal and vertical locations throughout the pile (Table 2.5). These samplers have a maximum capacity of 500 ml. Samples are extracted from the water sampling panel located in the instrumentation hut.  2.10.1.4 Gas Ports A total of sixty-one (61) ports provide gas samples from within the pile. The original pile design included sixty-four (64) gas sampling ports but several (~ 5 - 10) gas tubes were compromised during installation and resulted in the installation of duplicate tubes where possible. 10 ports are located in line 5 and 14 in line 6 (base), and an additional 10 in each of lines 1 to 4 (vertical) (Table 2.5; Appendix A). The gas sampling ports consist of 1.6 mm ID polyethylene tubes (www.new-line.com). One endpoint was covered with a porous geotextile, and placed facing downwards to prevent dust and water from entering the line. It was then covered with pebblesized material to allow for unobstructed gas flow. The other end extends to the instrumentation hut where it is connected to the gas panel where samples can be extracted.  31  2.10.1.5 Tipping Buckets Infiltrating rain water that penetrates the full depth of the pile eventually reaches the base of test pile 1 test pile where the basal lysimeter and three sub-lysimeters are located. Water from the lysimeters then flows to the instrumentation hut via 4” HDPE pipes. Inside the instrumentation hut, water passes through one of four flow-through conveyance cells installed for water quality measurements, and then into one of the four custom-built tipping bucket flow gauges (Corazao Gallegos, 2007). The flow-through conveyance system allows for continuous electrical conductivity and temperature data to be obtained, and a sampling port facilitates sample extraction (Figure 2.9). Tipping buckets, used to measure discharge, are based on designs that have been used in similar experiments to measure the flow rate of water from test piles (Nichol et al., 2005; Andrina et al., 2006; Neuner, 2009).  Water flowing from the basal lysimeter flows to the ‘large tipping bucket’, which was designed to operate under a maximum flow rate of 60 L/min and a maximum volume-per-tip of nearly 3 L. Water flowing from each of the three sub-lysimeters separately drains to one of three ‘smalltipping buckets’ which were designed to operate under a maximum flow rate of 16 ml/s and a maximum volume-per-tip of nearly 25 ml. Each tipping bucket is connected to the CR1000 datalogger in the instrumentation hut, which records the time of each tip.  Each tipping bucket was individually calibrated in January-2007 and again in August-2008. The calibration equation converts the raw tip-time (time between consecutive tips) to a corresponding flow rate (Appendix E). The equivalent volume of water during each tip is calculated by multiplying the calibrated flow rate by the reported tip-time. Results are totalled on a daily basis and reported as m3/day. For further information regarding the background, design, installation and initial calibration of these flow tipping buckets, refer to Corazao Gallegos (2007) and Appendix E.  2.10.1.6 Electrical Conductivity Probes Electrical conductivity is indicative of the concentration of charged species in a porewater solution, whereby higher values correspond to higher ionic strengths and electrical conductivity. A custom-built electrical conductivity (EC) probe was installed within each of the four 32  lysimeters’ flow-through conveyance system (Figure 2.9) and connected to the CR1000 datalogger to provide automated EC measurements at 30-minute intervals. Probes were designed and built at the University of British Columbia (Vancouver, BC, Canada). Corazao Gallegos (2007) and Neuner (2009) describe the theory, design and installation of these inexpensive sensors.  The raw EC data was temperature-corrected to account for temperature variations (Hayashi, 2004): EC 25 = SC = ECT [1 + a (t − 25)]  [2-1]  where ECt is the measured EC [µS/cm] at a given temperature t [°C], EC25 is the EC at a standard reference temperature of t = 25 °C and referred to as specific conductance (SC). The temperature compensation factor a [°C-1] is an arbitrary value (dependent on the TDS-EC relationship) and accounts for increases to EC with temperature increase. A commonly used a value is 2 %, which describes a ‘2 % increase of EC per 1°C increase of temperature’. Other, more site-specific temperature compensation values have been employed in previous experiments (e.g., 1 – 3 %; Hayashi, 2004).  2.10.1.7 Composite Tank and Flow Splitter A composite tank was installed to permit weekly mass balance (water) and mass loading (solute) calculations for the entire pile. Serving as a secondary reservoir that retains a proportion of water draining from all four lysimeters, the composite tank allows for a comparison of the entire pile flow and chemistry averaged between sampling intervals with the flow and chemistry from instantaneous samples collected from individual lysimeters.  All of the water draining from the bottom of each of the four tipping buckets converges to a single pipe that directs water to a flow splitter. The flow splitter, 0.8 m long x 0.33 m wide x 0.75 m tall, is designed to separate approximately 6.25 % of the combined effluent water and convey it (under gravity) to the 2.5 m3 composite tank. Corazao Gallegos (2007) contains details about the design, construction, installation and calibration of the flow splitter.  33  The composite tank consists of a multi-layered plastic that prevents solar rays from penetrating the plastic (so as not to affect the geochemistry of the water) and a high-resistance outer layer that prevents degradation over time. The 2.5 m3 tank (1.55 m high x 1.60 m wide) was installed at a 3 % gradient towards the direction of the outlet valve roughly located 2 cm above the base of the tank. It is assumed that there is a negligible amount of water or sediment remaining in the tank after it is drained. For precaution, a roof is located above the tank to prevent any damage from occurring (Corazao Gallegos, 2007). Further detail can be found in Appendix G.  2.10.2 Meteorological Instrumentation 2.10.2.1 Precipitation Precipitation at the test pile site is measured using a Rainnew 111 Tipping Bucket Wired Rain Gauge (+/-1% accuracy) from RainWise Inc. (www.rainwise.com) connected to a Campbell Scientific CR1000 datalogger. The rain gauge is located on the roof of the instrumentation hut. Another six tipping bucket rain gauges (Campbell Scientific TR-525 rain gauge; ± 1% accuracy; www.campbellsci.ca) are located within 5 km of the study site. Each tip is equivalent to 0.254 mm of rainfall. Assuming that the rain is spatially uniform, the rainfall can be multiplied by the horizontal surface area of the pile in order to estimate the total volume of water falling on the pile surface. Due to a mean annual temperature of 5.5 – 6.0 °C, almost all precipitation is in the form of rain. When it does snow, it typically melts within hours to days.  2.10.2.2 Evaporation A ‘Class A’ pan located less than 4 km from the study site at the Yanacancha meteorological station provides pan evaporation (Epan) measurements. Free-water Epan measurements represent an integrated proxy of the actual evaporation that occurs due to transient climatological conditions such as net radiation, air temperature, humidity and wind speed. Allen et al. (1998) describes the pan evaporation set up in detail. This is further explained in Appendix D.  2.10.2.3 Air Temperature, Relative Humidity, Wind Speed and Direction. See Appendix C.  34  2.11 WATER SAMPLING Twenty-one (21) water quality monitoring locations exist in test pile 1, 15 sampling ports and 6 sampling stations. The 6 sampling stations are located outside the pile (pile exterior) and consist of the lysimeters, composite tank and sump. The 15 sampling ports are located inside the pile (pile interior) where the soil water samplers (SWS) are located along instrumentation lines 1, 2 and 4 (Appendix A).  The 6 sampling stations are located outside the pile and monitor drainage from each the basal lysimeter, three sub-lysimeters, composite tank and sump. The sampling stations from each of the three sub-lysimeters (A, B and C) and the basal lysimeter (D) are located in the flow-through lysimeter conveyance cell in the instrumentation hut (Figure 2.9). The composite tank (E) is located adjacent to the instrumentation hut and is sampled at the outflow valve. Water is sampled from the sump (F) via a hose protruding from the pile’s south eastern corner.  2.11.1 Water Quality Parameters The samples were collected at weekly, bi-weekly, monthly, bi-monthly or semi-annual intervals. The frequency depends on the station, parameter, time of year, and pile maturity. Field parameters, described next, are obtained for each sample.  2.11.1.1 Field Parameters Field parameters include temperature (± 0.1 °C), pH (± 0.03 units), specific conductance (± 0.05 % µS/cm) and dissolved oxygen (± 0.05 % mg/L). These parameters are measured with a handheld multiparameter instrument (WTW Multi 340i). Field parameters are measured on a weekly basis for the sampling stations due to their inexpensive cost and east of measurement. The 15 sampling ports (i.e., SWS) were analyzed weekly during the first year and bi-weekly thereafter, depending on the time of year.  2.11.1.2 Lab Parameters Lab samples were analyzed weekly for all 4 lysimeters and composite tank, with samples analyzed bi-weekly for the sub-lysimeters after the first year. It should be noted that the sub-  35  lysimeters do not always flow in the dry season at Antamina in which case sampling can be intermittent or non-existent. Samples from the sump are extracted and analyzed every 6 months.  Samples from the 15 SWS were obtained weekly during the first year, and every 2 to 4 months thereafter, depending on the time of year. Suction lysimeter samples can only be collected when a sufficient volume of water is available (discussed below). In situations when insufficient solution volume for complete analysis is available, samples are sent to the lab in order of priority; dissolved metals, total metals, anions, alkalinity, and nitrogen species.  All samples are refrigerated to an optimal storage temperature of 4 °C during transport from the field to the laboratory in Lima. Samples are analyzed by Envirolab Peru S.A.C. by inductively coupled plasma mass-spectrophotometry (ICP-MS) for pre-determined solutes. The suite of analyzed constituents is described below. Corazao Gallegos (2007) describes the analytical method and parameter detection limits.  2.11.1.2.1 Dissolved Metals Dissolved metals have the highest priority for analysis. When 250 ml of a sample are obtained, the following dissolved metals will be tested for : Al, Sb, As, Ba, Be, Bi, Bo, Cd, Ca, Co, Cr, Cu, Sr, Sn, P, Fe, Li, Mg, Mn, Hg, Mo, Ni, Po, Ag, Pb, Se, Na, Ta, Ti, V, Zn. At the time of collection, samples intended for dissolved metal analysis are filtered and preserved with nitric acid to reduce the pH to less than 2.  2.11.1.2.2 Total Metals When 500 ml of a sample are obtained, the same constituents that are tested for dissolved metals are tested for total metals. Total metal samples are not filtered, but are preserved with nitric acid, similar to dissolved metals.  2.11.1.2.3 Other parameters Other parameters include non-metals and nutrients, etc. When 1000 ml of a sample are obtained, total alkalinity, chloride, fluoride, pH, sulphate, ammonium, nitrate and nitrite are all analyzed  36  (recall that the soil water samplers have a maximum capacity of 500 ml). Anions are not preserved.  2.12 GAS SAMPLING Gas samples were obtained every six months: April-2007, November-2007 and August-2008. Samples were analyzed using an OmniStar Quadrapole Mass Spectrometer and a Varion 4200 Gas Chromatograph.  2.13 WASTE ROCK PROPERTIES A detailed characterization of the physical waste rock properties is critical in understanding the hydrological and chemical processes occurring within the waste rock pile. The solid phase mineralogical and elemental composition, particle density and acid-base accounting of the waste rock is presented based on field cell analyses by Aranda (2009). (Note, for the purpose of this investigation, it is assumed that the mineralogical proportions measured in the field cells represent the same proportions in the test pile). A particle size distribution was performed on selected material from end-dumped pile material. Saturated hydraulic conductivity is measured on the compacted, top surface of the pile. The bulk porosity of the waste rock is estimated based on the mass and volume of test pile 1.  2.13.1 Solid Phase Mineralogical Composition The black marble, marble diopside and grey hornfels waste rock used in the construction of this test pile originated from the Jumasha Formation (Section 2.3). This material was classified as Class B waste rock based on the element-based assays employed by the mine (Table 2.1). Mineralogical composition analyzed using Mineral Liberation Analysis (MLA). The MLA uses a Scanning Electron Microscope equipped with sophisticated software that performs an automated, systematic, quantitative evaluation of the abundance, association, size and shape of minerals within a given sample. Results of this analysis show that average pile composition consists of 48.4 % silicate, 47.5 % carbonate, and 2.2 % sulphide minerals (Aranda, 2009). The remaining 1.9 % consists of phosphate, oxide, sulphate and other minerals. Nearly 56 % all sulphide minerals consists of pyrite (1.2 % of the total pile), and 95 % of all sulphide minerals when combined with chalcopyrite (20.2 % of sulphides), pyrrhotite (10.2 % of sulphides) and 37  sphalerite (8.2 % of sulphides). Calcite is the dominant carbonate mineral. Tipping Phase II material has the highest sulphide and silicate and the lowest carbonate proportions. Tipping Phase II has the highest carbonate content. Table 2.3 summarizes MLA results from Aranda (2009).  2.13.2 Solid Phase Elemental Composition Whole-rock elemental analysis was conducted on material from the protective layer and three tipping phases by the ALS Chemex Laboratory Group using the ICP-MS61 methodology. Average pile sulphur (S) content is approximately 0.61 wt. %, with 27.1 % wt calcium (Ca), 1.7 % wt. iron (Fe), 0.09 % wt. zinc (Zn) and 0.06 % wt. copper (Cu). The high Ca content indicates a significant buffering capacity of the waste rock (Table 2.4). Tipping Phase II has the highest Fe and S proportions and Tipping Phase III has the highest Ca and Zn proportions. Refer to Aranda (2009) for more detailed attributes of this material.  2.13.3 Acid Base Accounting Sobek et al. (1978) provided the earliest methodology for acid base accounting (ABA), which the mining industry uses to obtain an initial characterization for predicting acid rock drainage. ABA is used to quantify the constituents of the material under study to determine how much acid it can produce when oxidized or how much acid it can neutralize. The acid potential (AP) is determined in a static test by calculating the theoretical amount of acid that can be produced if all of the total sulfur is oxidized to sulfuric acid and is expressed as kg CaCO3/tonne. The neutralization potential (NP) is determined experimentally by digestion of hydrocholoric acid, followed by a titration to determine the amount of acid consumed, expressed in kg CaCO3/tonne (Sobek et al., 1978; Lawrence and Wang, 1996). Results are typically presented in the form of a Neutralization Potential Ratio (NPR) which is the NP-to-AP ratio. A higher ratio value means a larger the neutralization potential. Any value less than 2 most likely ‘acid-generating’ and higher than 4 is most likely ‘non-acid generating’ (Price, 1997).  Acid base accounting was conducted on waste rock from the protective layer and all three tipping phases to provide a preliminary assessment of how strongly the waste rock can buffer the generated acidity when exposed to weathering under natural conditions. The NPR ratios range 38  between 5 and 291, with material from Tipping Phase II having the lowest ratio (NPR ~ 5) and Tipping Phase III having the highest ratio (NPR ~ 5). By this definition, all of the waste rock placed in the pile is non-acid generating. Further details can be found in (Aranda et al., 2009).  2.13.4 Particle Size Distribution Waste rock particle size distributions (PSD) generally encompass size fractions that range over several orders of magnitude. Regardless of the coarse proportions, the waste rock PSD curves typically exhibit a long tail at the fine size fractions (Smith and Beckie, 2003). The most common way to perform a grain size analysis is by sieve separation which physically separates the material based on different sieve sizes.  Five particle size distribution tests were performed on the test-pile waste rock by Golder Associates using the ASTM D 5519 methodology. The first three tests were performed on material randomly selected from 26 truckloads of stockpiled material and the additional two tests were performed on one individual truckload dumped adjacent to the pile during Tipping Phase I and II. Nearly 450 tonnes of waste rock was used for the PSD on material from Tipping Phases I and II, and was removed from the total mass of material placed in the pile (Table 2.6). The average PSD curve for the protective layer (stockpiled material) is shown in Figure 2.10 along with those from Tipping Phase I and II.  Very large boulders in the waste rock necessitated the use of excavators and other mechanical equipment to separate larger particle sizes (Corazao Gallegos, 2007). It is important that PSD analyses are performed shortly after material deposition at site to avoid any compaction, settling, or dissolution in rain. Material for the concurrent field cell investigation by Aranda (2009) was randomly selected from these stockpiles using 25-kg ‘grab-bags’.  The material is relatively coarse and heterogeneous, with sizes extending across more than four orders of magnitude. There is a long tail of fine material, and 15 – 25 % of the material is larger than 1 m in diameter. Another way to characterize PSD curves is through the uniformity coefficient CU which describes the ratio between the diameter of the size fraction that represents nearly 60 % of the total sample (D60) over the 10 % size fraction (D10). Morin et al. (1991) report 39  that although typical waste rock has a widely distributed grain size, the CU (D60/D10) is always at least 20. The average uniformity coefficient for the five PSD curves measured on this waste rock is CU = 330 mm / 10 mm = 33. 2.13.5 Hydraulic Properties The hydraulic properties of waste rock are heavily determined by the PSD (Chapter 1). Yazdani et al. (2000) determined that particles with a diameter > 4.75 mm no longer have significant capillarity under unsaturated conditions. Based on the PSD (Figure 2.10), approximately 5 - 10 % of the pile is made up of material < 4.75 mm in diameter, indicating that the pile is relatively coarse and that preferential (i.e., macropore) flow may be present along with matrix flow.  Eight fixed-ring infiltrometer tests were performed at several locations of the top, compacted layer of the pile to measure the one-dimensional (vertical) saturated hydraulic conductivity. Methods were similar to those outlined in Bouwer (1986). Results varied by two orders of magnitude from 1.2 x 10-4 (10 m/day) to 3.7 x 10-6 m/s (0.3 m/day), and an average saturated hydraulic conductivity is 2.7 x 10-5 m/s (2.3 m/day). No direct hydraulic conductivity tests were conducted along the less consolidated, coarser batter although empirical methods are used to provide indirect estimates of hydraulic conductivity in Chapter 3.  2.13.6 Dry Bulk Density The dry bulk density of the stockpiled protective layer was measured (using a water replacement method) before it was placed in the pile (Corazao Gallegos, 2007). The stockpiled material was compacted in a similar manner to its placement in the protective layer of the pile and tested. Results indicate that the stockpiled material had a dry bulk density of ρb = 1845 kg/m3. Dry bulk density test was not measured on the end-dumped material from Tipping Phases I, II and III.  The amount of material compaction can affect porosity and the overall flow regime. As Nichol (2002) indicates, flow from a 5 m-high waste rock pile can vary by several orders of magnitude with a density increase of 40% due to compaction. It is for this reason that it is important to consider the way in which the waste rock material was handled and minimize handling to simulate actual waste dump conditions. The stockpiled material was flattened and likely 40  compacted to a degree resulting in a slightly higher than expected density. Nearly three months after the waste rock was stockpiled and the density measured, the material was finally placed in the protective layer. During this time the material could have settled, reacted with atmospheric conditions of the wet season, and some finer material may have been washed away. Based on these factors, results of the bulk loose density test on the stockpiled protective layer could not be used as a proxy bulk density of the pile and another method was utilized. An uncertainty of up to 30 % was attributed to this estimate, resulting in a dry bulk density of ρb = 1845 ± 550 kg/m3. A second way bulk density estimated was performed based on the mass and volume of test pile 1. Using GPS data, the volume of waste rock that was delivered to site is estimated to be 11,600 m3. However, both mass and volume estimates have to account for the amount of material that was dumped at the back slope of the pile (i.e., the first tips of Tipping Phase I) but was not included in the 36 x 36 m pile footprint (1,600 m3 or 3,200 tonnes). In addition, approximately 2,600 tonnes of waste rock was removed from the bottom perimeter of the pile after completion to conform to the initial pile design dimensions and to allow access to the three exposed sides of the pile. Accounting for the material removed for the PSD (Section 2.13.4) and material removed from the protective berm during the last tipping phase (Corazao Gallegos, 2007), the mass of waste rock placed in the pile is estimated to be 20,000 ± 4,000, with a total estimated volume of 10,000 ± 2,000 m3 (Table 2.6). Based on the mass and volume of test pile 1, the resulting bulk density is estimated at 2,000 ± 400 kg/m3. Based on the shape of the pile, the top compacted layer represents 24 % of the total pile surface area and the batters make up the remaining 76 %.  2.13.7 Porosity The bulk porosity n of a porous medium can be difficult to estimate accurately without in-situ measurements. Due to a wide variety of material sizes ranging from fine sand and silt to large boulders in waste rock, it is difficult to determine an accurate, representative porosity value without more specific information and intimate knowledge of the complex relationships between grain sizes in the existing sample. Obtaining actual porosity measurements would significantly disturb the pile so instead porosity was estimated.  41  Bulk porosity is estimated based on the void volume Vv and total volume VT, or the dry bulk density ρb [kg/m3] and specific gravity/particle density GS [kg/m3] of a porous medium: n = VV/VT = 1 – ρb/GS  [2-2]  where ρb = 2000 ± 400 kg/m3. The specific gravity of a material can be measured experimentally and compared to an equal volume of pure water. Particle density was measured for the kinetic field cell waste rock (Aranda, 2009). An average particle density value of 2894 ± 60 kg/m3 was obtained for all field cells. Waste rock from Tipping Phase II (FC-2) has the highest average particle density of 3140 kg/m3, and material from Tipping Phase III (FC-4) has the lowest particle density of 2747 kg/m3. The experimental results are rather consistent with results of whole-core density analyses for limestone/marble and hornfels performed by AMEC (AMEC, 2008). Using the two bulk density ranges of 1845 ± 550 kg/m3 (empirical) and 2000 ± 400 kg/m3 (calculated), the calculated porosity ranges between 0.15 < n < 0.42. However, the presence of large boulders may decrease bulk porosity because while they contribute to the total pile mass but not the total void volume (Neuner, 2009; Fala et al., 2003). Therefore, it is assumed that any boulders with a diameter larger than 0.5 - 1 m would not contribute to the overall porosity. Based on the PSD (Figure 2.10), approximately 15 to 25 % of the waste rock contributes to the final mass and volume, but not the void volume. The resulting total volume VTn that contributes to porosity is estimated to be 75 – 85 % of VT. The final bulk porosity is therefore estimated to be 0.45 ± 0.15 based on the bulk density of the protective layer and 0.4 ± 0.1 based on the total pile mass and volume.  2.14 SUMMARY This chapter described the climate, geography, geology and hydrogeology of the Antamina mine. A brief history of the mine history and the mining activities was described in order to provide a history of waste rock used in this study. Different waste rock types are classified based on their chemical content and the Class B (possibly-reactive) waste rock material was used in the pile. The pile design and construction was also described, including the dimensions, instrumentation 42  and timeline for construction. The physical properties of the waste rock placed in the pile were examined and will serve as the foundation for understanding the hydrological and solute release behaviour of the pile.  43  TABLES  Table 2.1. General classification of waste rock types from the Antamina mine. Metal Limits/Restriction Class Rock Type Zn As Sulphides Visible Oxides (ppm) (ppm) (%) (%) Hornfels, Limestone, A >1500 >400 >3 >10 Marble, Skarn, Intrusive Hornfels, Limestone, B 700 – 1500 <400 <2-3 <10 Marble Hornfels, Limestone, C <700 <400 <2-3 minimal Marble  44  Table 2.2. Test pile 1 construction timeline. Initial Bulk Moisture Content (%)1  Observations (qualitative)  End  2-Feb  3-Feb  3-Mar  10-Mar  Wet and rainy  11-Mar  3-Apr  Wet and rainy  5-Apr  17-Ap  Wet and rainy  Protective Layer  4-May  10,11May  Tipping Phase I  10-May  11-May  12-Jun  26-Jun  27-Jun  27-Jun  29-Jun  6-Jul  Tipping Phase III a3  7,8-Jul  9-Jul  Tipping Phase III b3  7,8-Jul  9-Jul  20-Jul  24-Jul  5-Aug  6-Aug  Waste Rock Protective Layer Stockpile Geomembrane Installation 2B-Protective Layer Sub-Lysimeter (material, geomembrane and 2B Protective Layer)  Instrumentation Line 1 Installation and Covering Tipping Phase II Instrumentation Line 2 & 3 Installation and Covering  Instrumentation Line 4 Installation and Covering Perimeter Adjustment  Type  Initial Mass [metric tonnes]  Beginning  Construction Stage  Black Marble Marble Diopside  5,740 6,400  3.1%  -  2.1%  Sunny, dry Mainly sunny and dry, some clouds  Marble Diopside  Marble Diopside Grey Hornfels  6,120  5,775 2,765  1.7%  Sunny, dry  1.9% 2  Mainly sunny and dry, some clouds Sunny, dry Sunny, dry Mainly sunny and dry, some clouds  * based on Corazao Gallegos (2007) 1 as measured during particle size distribution 2 estimated based on water content of Tipping Phase I and II. 3 two different tipping phases that are views as one.  45  Table 2.3. Solid phase mineral composition of waste rock. Phase Type  Protective Layer Black Marble  Mass [t] Mineral Sulphide Carbonate Silicate Phosphate Oxides Sulphates Others Total  5,560  1 2  0.5 55.6 41 1 1.2 0.3 0.4 100  Tipping Phase I Marble Diopside  Tipping Phase II Marble Diopside  Tipping Phase III Marble Diopside/ Grey Hornfels 2,835 5,385 5,780 % Material Placement Phase by weight1 2 4.5 1.9 47.5 11 73.8 49.6 82.2 23.6 0.3 0.4 0.24 0.4 1.3 0.30 0.1 0.3 0.14 0.1 0.1 0.17 100 100 100  2  19,560 2.2 47.5 48.4 0.5 0.8 0.2 0.2 100  based on field cell material 10-840 mm in diameter as determined by MLA on (Aranda, 2009) weighted average  Table 2.4. Solid phase elemental composition of waste rock. Tipping Tipping Solute Unit Protective Layer Phase I Phase II Mass tonnes 5,560 2,835 5,385 As ppm 134.0 67.9 44.0 Ca % 30.0 26.7 18.2 Cu ppm 588.0 819.4 715.1 Fe % 0.87 1.27 3.5 Mg % 0.88 1.23 1.7 Mo ppm 5.65 92.6 44.2 Na % 0.12 0.09 0.27 Pb ppm 553.5 309.2 822.0 Zn ppm 1267.5 645.0 498.0 S % 0.24 0.29 1.21 1  Pile  Tipping Phase III 5,780 30.5 32.7 410.4 1.14 0.99 46.8 0.09 470.0 983.9 0.57  based on field cell material 10-840 mm in diameter as determined by ICP-MS (Aranda, 2009) weighted average of whole pile  46  Pile2 19,560 69.1 27.1 604.0 1.72 1.18 41.0 0.15 567.3 882.2 0.61  Table 2.5. Summary of instrumentation. Line Line Instrumentation 1 2 TDR 5 5 Thermistor 6 6 Gas 10 10 SWSS 5 5 Electrical Conductivity 0 0 Probe Tipping Bucket 0 0  Line 3 5 6 10 0  Line 4 5 6 10 5  Line 5 1 5 10 0  Line 6 1 3 14 0  Instrumentation Total Hut 0 22 4 36 0 64 0 15  0  0  0  0  4  4  0  0  0  0  4  4  Table 2.6. Final mass calculation. Mass [metric tonnes] 2  Stage Protective Layer Tipping Phase I Tipping Phase II Tipping Phase III  Density [kg/m3] 2770 2910 3140 2850  Initial Mass1 5,560 6,260 6,010 8,380  Particle Size Distribution -225 -225  Exterior  2  Pile Adjustment  -3,200 -400 -2,600  Final Pile Mass 1  Berm  Final 5,560 2,835 5,385 5,780 19,560  initial mass dumped on site according to Antamina Mining Operations Aranda (2009)  47  FIGURES  CORDILLERA NEGRA  CORDILLERA BLANCA  LIMA  Figure 2.1. Map of mine location. Modified from Aranda et al. (2009).  48  Figure 2.2. Plan view of schematic lithology of the Antamina deposit. (Antamina, 2003).  Figure 2.3. Schematic lithology and metal zonation of Antamina (Antamina, 2003).  49  Figure 2.4. Map of the Antamina mine.  50  4385 m-platform  4373 m-platform  Test Pile 1 Instrumentation Hut  4385 m-platform 4373 m-platform  4371 m-platform Figure 2.5. South view of completed test pile 1 and platforms.  51  N  Sump  Figure 2.6. Test pile 1 base design and instrumentation schematic (plan view).  Figure 2.7. Test pile 1 side profile and instrumentation schematic (cross section). 52  a) Vertical Relief  b) Location of Instrumentation Lines 1 - 4  8945570  Line 1  8945560  Line 2 Line 3  8945550 North  Line 4  8945540  2007 Top 2007 Berm  8945530  2007 Bottom  8945520 8945510 275400  Instrumentation Hut  275450 East  275500  Figure 2.8. Plan view location of instrumentation points in relation to pile exterior. 53  Figure 2.9. Inside the instrumentation hut.  54  Figure 2.10. Particle size distribution of waste rock material placed in the pile.  55  3  RESULTS: TEST PILE HYDROLOGY  This chapter characterizes the water balance of the 36 m x 36 m x 10 m-high waste rock test pile 1 during the first 21 months of operation from 18-Jan-2007 to 30-Sept-2008. By examining each component of the water balance equation in test pile 1, this chapter will identify:  1. the percentage of precipitation that infiltrates through the test pile, 2. spatial and temporal flow differences, 3. what flow mechanisms are prevalent, 4. average linear groundwater and wetting front velocities, and 5. moisture content changes.  The water balance equation for test pile 1 is: P – R – ΔS – ET – G – Q = 0  [3-1]  where precipitation P, runoff R, change in storage ΔS, evapotranspiration ET, groundwater seepage G, and basal outflow Q are typically expressed in m3/day or a time averaged flux of mm/day. Since test pile 1 is free of any vegetation, there is no transpiration and evaporation E need only be considered. When ponding is observed on the compacted traffic surface at the top of the pile, the ponded water does not run off and eventually evaporates or infiltrates, or may generate small-scale runoff that infiltrates in areas of higher permeability. Runoff is further prevented by a 0.5 m berm that is placed around the top surface of the pile. A French drain installed on the natural slope that abuts test pile 1 intercepts groundwater flow and prevents subsurface flow from entering the pile (Corazao Gallegos, 2007). The only source of water into test pile 1 is precipitation falling on the top and side of the pile. The water balance equation 3-1 is then rearranged: P – E = ΔS + Q  56  [3-2]  3.1  PRECIPITATION  A total of 1526 mm of precipitation P is reported by the site rain gauge during the 21 month period between 18-January-2007 and 30-September-2008. This amount corresponds to 1978 m3 of precipitation across the horizontal surface area of the test pile (1296 m2). A clear seasonal variability is apparent, as a significant proportion of the annual rainfall occurs within nearly one half of the year (wet season), leaving the other half relatively dry (dry season). Due to a poor rain gauge-datalogger connection from February- to June-2008, rainfall at the study site during this period is estimated based on measurements made at Antamina rain gauges located within 5 km of the site (Appendix C).  3.1.1  Estimation of Missing Precipitation Data  Using meteorological data from several weather stations located throughout the Antamina property (Figure 3.1), 375 mm of precipitation is estimated to have fallen at test pile 1 during the period of missing rain data between February- and June-2008. Weekly precipitation data is used for this estimation because it provides the best compromise between monthly (low resolution, low standard error) and daily (higher resolution, higher standard error) analyses. The 123-day long gap is reduced to 119 days (17 weeks) because the partial week after the gap in which rain data is obtained registers a higher amount of precipitation than estimated, and is used because it represents rain recorded by the rain gauge. Conversely, 10.2 mm of rain is registered during the partial day in which the rain gauge stopped working. When this amount of rain is accounted for, the total reported site precipitation becomes 1516 mm (1526 – 10.2 mm).  The following section summarizes the steps used to estimate the missing site precipitation. Appendix C outlines these steps in greater detail.  3.1.1.1 Local Gauges and Linear Least Square Regression Analysis Precipitation reported by the site rain gauge is compared to precipitation reported by six Antamina rain gauges (Figure 3.1). Based on data from the entire study period, the site rain gauge is best correlated with the Yanacancha rain gauge (r2 = 0.85) situated approximately 4 km away and 200 m lower in elevation, and Quebrada Antamina (r2 = 0.84) situated less than 5 km  57  away and 300 m higher. Based on precipitation data from these two stations, the missing site precipitation (uN) is estimated using a least squares analysis: 𝑓𝑓(𝑢𝑢𝑁𝑁 ) = 𝑎𝑎1 �𝑀𝑀𝑥𝑥 𝑁𝑁 � + 𝑎𝑎2 �𝑀𝑀𝑦𝑦 𝑁𝑁 � + 𝑑𝑑  [3-3]  where N is the number of data points in each basis function (i.e., the number of weeks; N = 1, 2, 3….17); f(uN) is the estimated mm of rainfall at the test site at a given week N; Mx and My are the Yanacancha and Quebrada Antamina basis functions during at a given week N; a 1 and a 2 are the Yanacancha and Quebrada Antamina parameters (i.e., correlation factors to site data); and d is the offset term. Based on calculations outlined in Appendix C, the a 1 and a 2 parameters are calculated to be 0.510 ± 0.017 and 0.354 ± 0.015, respectively, and the d offset is calculated to be 2.76 ± 0.17. Accounting for error, a total of 375 ± 15 mm of precipitation is estimated to have fallen in the period during which the datalogger failed to record rainfall at the test site.  3.1.2  Estimation Validity  An examination of the precipitation, wind direction, temperature and relative humidity for the period of missing site precipitation shows no clear departure from normal variability dating back to 2001 (Appendix C). The precipitation estimate proved to be a suitable substitute for the missing site precipitation because:  1. results from applying equation 3-3 to estimate precipitation at the test site over the entire study period (and not just during the period of missing precipitation) are similar to the measured precipitation over the same time period (see below), with only a 6 % difference between cumulative totals (Figure 3.2), and 2. cumulative precipitation from a second rain gauge (Campbell Scientific TR-525, same as the Antamina gauges) installed on site (28-Sept-2008) has a similar shape to the cumulative precipitation of the first site rain gauge during the same period 135-day period (see below), with a 4 % difference between cumulative totals (Figure 3.3).  In the first case, the linear regression analysis is extended to estimate rainfall at the test pile over the entire study period from 18-Jan-2007 to 30-Sept-2008 based on the weekly Yanacancha and 58  Puente Juprog rain data (Figure 3.2). The cumulative measured site precipitation (rain gauge) and cumulative estimated precipitation (equation 3-3) over the entire study period shows that they are quite similar in shape. The total measured precipitation during this period (1516 mm) is approximately 6 % lower than the total estimated precipitation during the same period (1607 mm), indicating that the least squares analysis provides a close approximation to the reported site rainfall. For comparison, cumulative rainfall reported at the Yanacancha and Quebrada Antamina stations is also presented.  In the second case, a comparison of overlapping daily rainfall data is carried out between the site rain gauge and a backup rain gauge installed on top of the test pile on 28-Sept-2008, at the end of the study period. For the purpose of this analysis, performance of the two gauges extends beyond the study period and 135 days of data (28-Sept-2008 to 12-Feb-2009) is compared between both gauges (Figure 3.3). Precipitation from this period is not included in the water balance calculations referred to throughout the remainder of this investigation. The total measured precipitation during this period for the primary site gauge (780 mm) is approximately 4 % higher than the backup gauge (747 mm). These differences may be the result of:  1. natural variability in total weekly precipitation i.e., spatial heterogeneity of precipitation particularly in mountainous regions (Buytaert et al., 2006). 2. slight manufacturing differences between the primary rain gauge (RainWise) and the backup rain gauge (Campbell Scientific).  Nevertheless, results indicate that there is an expected variability in precipitation even over a small area (i.e., 10’s of m), regardless of whether they are based on weekly or daily data. Since cumulative site precipitation (measured and estimated) is based on weekly data and is within the same range of variability as the two site rain gauges which are based on daily data (± 5 %), this indicates that the result of the linear regression model is acceptable for the purpose of this investigation and can be used in the succeeding water balance calculations (Section 3.2.2; Section 3.5.1)  59  3.1.3  Uncertainty in Precipitation Measurements  It is possible that a number of rain gauge tips were not recorded by the datalogger due to the way in which the datalogger prioritizes the data being recorded (Appendix E). This would result in an underestimation of precipitation. Due to the nature of the precipitation data, it is nearly impossible to estimate the extent of unrecorded or ‘missed’ tips at site. That the reported site precipitation fell within the range of precipitation reported at other Antamina sites indicates that the amount of missed tips can be assumed to be insignificant for the purpose of this study.  Appendix C discusses the differences between the site rain gauge and the Antamina rain gauges, but does not consider the uncertainty involved in using rain gauges in general. In a study of local errors resulting from tipping bucket rain gauges, Ciach (2003) reports that while tipping bucket rain gauges are convenient and reliable tools they are subject to systematic errors (calibration, wetting-evaporation losses, etc.) and random instrumentation errors (heterogeneous distribution, time sampling, hydrodynamic flow instabilities in the pivot bucket mechanism, turbulent wind, etc.). These errors are typically unknown and largely ignored.  Furthermore, Ciach (2003) states that the magnitude of rain gauge errors is highly dependent on local rainfall intensity and the time interval over which the rainfall is examined. In examining moderate rates (10 mm/hr) over 5- to 60-minute time intervals, Ciach (2003) attributes errors between 3 – 5 % based on small time-scale intervals, and the errors continue to decrease as the time interval in which rainfall is recorded increases. It is likely that these errors are inherent to some degree in the site rain gauge data.  Error is also introduced in the linear calibration equation relating tips to precipitation rates or equivalent volumes of water. Experimental observations by Rain Wise Inc. demonstrate that while the theoretical volume of water to trigger a tip is 8.2 mL, the actual volume that triggers a tip is approximately 10 % less than theoretical. The empirical calibration of 7.4 mL/tip is thus recommended by the manufacturer because it accounts for splashing effects and residual water droplets that remain after a tip due to surficial tension, despite the application of certain Teflonbased sprays meant to reduce this. The relationship, however, is not linear because variable precipitation intensities will result in variable rainfall fluxes. Experimental observations by the 60  manufacturer indicate that when the flow rate does not exceed 75 mm/hour, the varying intensity can be corrected for by subtracting 3 % from the total precipitation. Since the hourly precipitation never exceeds this rate (Section 3.1.4, below), the 3 % correction can be applied to this rain gauge data. However, because there is no empirical evidence in the current literature that supports this 3 % correction, it cannot be applied to the data with sufficient reliability. Therefore, while most rain gauge manuals attribute a ± 1 % accuracy error arising from instrumentation errors, an uncertainty range of ± 5 % is attributed to the reported site precipitation due to all of the above-mentioned factors.  3.1.4  Final Site Precipitation and Reported Trends  In the 499 days in which site rain gauge data is available, a total of 6008 tips are reported by the gauge, which corresponds to a total of 1526 ± 76 mm of reported rain over the 21-month period. Accounting for the 375 ± 15 mm of precipitation estimated to have occurred during the 119-day period of missing site data, approximately 1891 ± 91 mm of rain is estimated to have fallen on site (after removing 10.2 mm described in Section 3.1.1). Site precipitation that is based on both measured and estimated precipitation is referred to as the final site precipitation. Figure 3.4 shows the cumulative estimated precipitation [mm] over the entire period along with the daily reported precipitation. Since the study period does not include two full years of data, the annual precipitation was calculated based on the first and last 365 days of the 21 month-long study period. For example, annual precipitation is 1224 mm/year between 18-Jan-2007 and 17-Jan2008, whereas it is 1370 mm/year during between 1-Oct-2007 and 30-Sept-2008. Both values are within the range of annual precipitation reported at other Antamina rain gauges.  There is a significant difference between the amount of rain during the wet and dry seasons, with the wet season typically extending from October through April and the dry season from May through September (Figure 3.4). The period of missing rain data extends across both of these periods with approximately 70 and 53 days in the wet and dry season periods, respectively. Based on measured precipitation throughout the entire 21 month-long study period (except for Feb- to Jun-2008), the average precipitation flux is 1.3 mm/day during the dry season and 4.6 mm/day during the wet season. Total dry season precipitation is 25 % of the total wet season  61  precipitation. There is nearly 25 % more precipitation in 2008 than during the same number of days in 2007 (Table 3.1).  Of the 499 reported days in which site rain data is available, 343 of these days (69% of the total days) report rain, with a maximum flux of 27.9 mm/day on 25-Apr-2007. This value does not account for varying intensities and durations of precipitation throughout the day. For example, it does not display whether all of this precipitation fell over 30 minutes or 12 hours, or whether most of it occurred during the morning, afternoon, or night. To better understand the variability in precipitation, this 499-day period is examined for 12-, 6-, and 1-hour intervals.  In general, the variability in precipitation increases with shorter time-intervals (Figure 3.5). Precipitation is reported in 52 % of the 12-hour periods (maximum flux of 18.3 mm/12 hr on 2Jan-2008), 38 % of the 6-hour periods (maximum flux of 16.3 mm/6 hr on 27-Nov-2007) and 14 % of the 1-hour intervals (maximum flux of 8.9 mm/hr on 24-Jun-2008). The highest precipitation flux is 1.02 mm/min on 25-Apr-2007 (not surprisingly the same day as the maximum daily precipitation). While the maximum recorded precipitation at 24-hour and minute-intervals are coincident with the same rain event, the highest recorded precipitation rates measured at 12-, 6- and 1-hour intervals are observed at other times of the year, regardless of whether it is the wet or dry season. This indicates that precipitation is highly variable at this site.  3.2  EVAPORATION  Evaporation E that is used in the water balance (equation 3-2) is estimated using the pan evaporation (semi-empirical), Penman-Monteith (theoretical), SoilCover (model), and large-scale pan lysimetry (empirical water budget) approaches during two periods:  1. the study period (i.e., the first 21-months of pile operation) from 18-Jan-2007 to 30Sept-2008, and 2. one complete water year from 1-Oct-2007 to 30-Sept-2008.  Pan evaporation measurements (EPAN) from the Yanacancha meteorological station ( Figure 3.6) are converted to the site-specific bare soil conditions of test pile 1 by using a 0.65 bare soil pan 62  correction factor Kpan-BS (Appendix D). The results are referred to as corrected pan evaporation. The Penman-Moneith approach requires a calculated bare soil correction of KBS = 0.8 to convert ETref to an estimated bare soil evaporation EBS (Appendix D). Results from the SoilCover model and large-scale pan-lysimetry water balance methods are only examined in the context of one water year (Table 3.2). Further details can be found in Appendix D.  Evaporation during the entire study period is examined first, followed by evaporation during the water year.  3.2.1  Evaporation: Study Period  Over the entire study period, the pan evaporation approach (EPAN) estimates total evaporation at test pile 1 to be 1423 mm (75 % of 1891 mm of total precipitation) whereas the PenmanMonteith approach (EPM) estimates total evaporation at 1617 mm (86 % of total precipitation). That the Penman-Moneith estimate is higher than the corrected pan evaporation estimate is consistent with other comparisons between the two models (Wilson et al., 1997; Allen et al., 1998; Swanson et al., 2003).  The study period (Jan-2007 through Sept-2008) does not span a representative time period. It is likely that the pile had not fully adjusted to the disturbances experienced during pile construction, which likely results in a moisture content less than residual saturation due to evaporation from the blasting process and during end-dumping. Since the 21 month-long study period consists of two complete dry seasons but one full and one partial wet season, it could not be easily compared to other results. Therefore a more detailed examination of the data during one water year, which includes one complete wet and dry season, is more representative of pile behaviour and can be more easily compared to results from other investigations.  3.2.2  Evaporation: 2007-2008 Water Year  The water year period extends from 1-Oct-2007 (beginning of the 2007-08 wet season) through 30-Sept-2008 (end of the 2008 dry season). This period is examined because:  63  1. nearly one year has passed since the test pile construction was completed and the pile is no longer affected by post-construction disturbances (i.e., waste rock has retained additional moisture content that it lost through evaporation during the end-dumping process), 2. flow at the end of the dry season is at a minimum compared to the rest of the year (further explained in Section 3.3.4), and 3. the net change in storage can be assumed to be ΔS ~ 0, which is important for water balance calculations (Section 3.4.4).  Evaporation estimates during the water year are presented below.  3.2.2.1 Pan Evaporation and Penman-Monteith Estimates Table 3.2 indicates that evaporation based on EPAN measurements yields an estimate of 887 mm (68 % of the 1297 mm of precipitation during the water year), whereas the EPM approach yields an estimate of 972 mm (75 %) during the water year. The average daily EPAN and EPM fluxes during the water year are relatively similar, with values of 2.4 and 2.7 mm/day, respectively. However, EPM is much more variable than EPAN (Figure 3.7) and average EPAN and EPM estimates are 5 to 10 % higher during the dry season compared to the wet season. For example, average wet season evaporation fluxes are 2.4 and 2.6 and mm/day for EPAN and EPM, respectively, and the corresponding dry season averages are 2.5 and 2.8 mm/day, respectively. A higher relative humidity during the wet season (Appendix C) likely results in more similar vapour pressures between the waste rock surface and the adjacent air mass, whereas the vapour pressure gradient increases during the dry season (which has a lower average relative humidity) and promotes higher evaporation rates.  The average daily precipitation rate during the water year is 3.5 mm/day, with an average of 5.1 mm/day during the wet season and 1.4 mm/day during the dry season (Table 3.1). Average precipitation-to-evaporation ratios (P :E) approximate 2 and 0.5 during the wet and dry season, respectively, for both EPAN and EPM. These ratio differences indicate that precipitation is greater than evaporation during the wet season and precipitation is less than evaporation during the dry season, and that there is a shift in dominance depending on the season (Figure 3.7). 64  These two approaches have limitations. Accurately characterizing evaporation at any given site can be highly complex. Allen et al. (1998) explain that no weather-based calculations can be expected to perfectly predict evapotranspiration under every climatic situation due to simplification in formulation and errors in data measurement. While pan evaporation data provides the closest proxy for daily potential evaporation, correcting the data from an open pool of water to a bare soil surface creates uncertainty in achieving accurate results. Pan evaporation methods are susceptible to the microclimatic conditions under which the pans are operating and the rigour of station maintenance among other factors (e.g., location, design specifications, material and measurement frequency). Although pan evaporation measurements are recorded daily at the Yanacancha weather station, they do not occur at the same time each day and the height of water in the pan is only refilled once per week. Allen et al. (1998) determined that, in general, a ± 15 % error should be attributed to pan evaporation data. This error range is applied to the pan evaporation data for test pile 1.  The Penman-Monteith approach is also useful in predicting evaporation based on meteorological conditions, but typically overestimates evaporation under unsaturated conditions and under low evaporation conditions, even with precision instruments and excellent environmental conditions (Wilson et al., 1997; Allen et al., 1998; Swanson et al., 2003). These estimates are also further complicated by several factors such as the soil, particle size distribution, moisture content, surficial permeability, infiltration rate, occurrence of ponding, aeration, etc. Allen et al (1998) reported error estimates as high as 20 % when using this approach. This error range will also be applied to the Penman-Monteith estimates for test pile 1.  3.2.2.2 Modeled Evaporative Fluxes: SoilCover The SoilCover evaporative flux model estimates evaporation to range between 32 and 41 % of the total precipitation during the water year, depending on the material properties (Table 3.2). This is approximately half of the evaporation estimated using the previous two approaches.  65  Estimates produced from this model proved too unreliable due to the lack of several soil property parameters that are required in order to achieve accurate results. This model is thus too unconstrained and could not be calibrated.  3.2.2.3 Lysimetry Water Balance Estimate The final approach used to estimate evaporation is based on the pan-lysimetry water balance method. Conducting a water balance calculation that includes P (Section 3.1, above) and outflow Q (Section 3.3, below), and assuming no net change in storage (ΔS ~ 0) over the course of one water year, water balance equation 3-2 can be rewritten to calculate evaporation E:  E=P–Q  [3-4]  Using equation 3-4 and assuming that there is no net change in storage (ΔS ~ 0), a total of EWB = 766 mm (59 % of site precipitation) is calculated to have evaporated from the pile during the water year (Table 3.2). This approach integrates several factors that the other approaches fail to address, one of the most important factors being the difference in evaporative flux between the rubble-based batters and fine-grained, compacted top of the pile (Figure 3.8). The lysimetry water balance approach provides the most reliable integration of climatic variables and material properties, particularly with regards to the differences between the top surface layer and the batter. However, it does not quantify these differences (i.e., separate evaporation from the batter vs. top of pile), and cannot be used to provide a higher-resolution evaporation estimate (i.e., daily, weekly, monthly). In addition, this approach heavily relies on the accuracy of the precipitation and outflow measurements and that no net change in storage occurred over the water year. All three of these components are based on a combination of reported data, interpolations and assumptions. Nevertheless, this approach is more reliable than the other quasiempirical and mathematical approaches in estimating evaporation at test pile 1.  3.3  OUTFLOW  Pile construction was complete by September-2006 but the automated data collection system did not become operational until 18-Jan-2007, mid-way into the wet season. Chapter 2 introduced the 36 x 36 m basal lysimeter D, and the three 4 x 4 m sub-lysimeters A, B and C located within 66  the basal lysimeter along the centre line (Figure 2.6). The combined flow from these four lysimeters is the outflow component Q in the water balance equation 3-2.  Based on qualitative observations of when water first began flowing from the base of the pile, it took approximately three to four months for the pile to ‘wet up’ after construction was complete (Corazao Gallegos, 2007; Bay et al., 2008). Discharging water is first observed from basal lysimeter D nearly three months after pile construction was complete, whereas flow did not start draining from the interior sub-lysimeters A and B until nearly four months after pile construction. In general, water penetrates the batters more rapidly than it does the full 10 mdepth of the pile. Once the pile began draining during this initial wetting up period, the rate of drainage clearly became affected by the rate and intensity of precipitation. The term ‘wet-up’ is used to describe the initial period after pile construction in which the waste rock moisture content (likely below residual saturation) increased to attain a state of dynamic equilibrium over a one-year timescale (e.g., a water year).  3.3.1  Correcting Tipping Bucket Data: RANSAC  Four tipping buckets are used to measure the rate of discharging water from each of the four lysimeters (Chapter 2). Each tipping bucket was calibrated to convert the ‘tip-time’ to a corresponding flow rate (Appendix E). Shortly after the tipping bucket flow gauges were calibrated and began collecting data in January of 2007, it became clear that there was a problem with the tipping bucket data. Due to a software problem, the datalogger randomly failed to record tips from each of the four tipping buckets. These missed tips are accounted for by using the Random Sample Consensus (RANSAC) model, which is typically applied to datasets in which significant outliers are present (Fishler and Bolles, 1981). This model identifies the outliers in the data that result from unrecorded tipping bucket tips, and allows for the subsequent correction. Appendix E contains a detailed description of how this was accomplished, a summary of which is provided next.  Based on a set of user specified parameters, the RASNAC model assigns each data point (i.e., tip-time) an ‘inlier’ or ‘outlier’ value. Inliers represent data whose distribution can be explained by a set of model parameters, i.e., can be fitted to a line, and represent the ‘base case’ or ‘field67  observed’ flow rate. Once identified, inlier points are placed in the inlier ‘bin’. Outliers represent data points that do not fit the model and can consist of both noise and erroneous measurements. Missed tips are associated with higher tip-times than base case tips. They are subsequently assigned m-integer values depending on how many missed tips the model estimates that they are associated with, and are then placed in the corresponding ‘bin’ level. For example, if the base case tip-time tt is 30 seconds, the tip-time may be around 60, 90 or 120 s at m = 1, 2 or 3 missed tips, respectively. Since the ‘tip-time to flow-rate’ calibration equation is non-linear, the resulting flow rates do not change by the same corresponding factor e.g., at tt = 30 s, the flow rate q = 5.9 L/min and at tt = 120 s, the flow rate q = 1.4 L/min. See Appendix E for further explanation.  Results of the RANSAC model indicate that the proportion of base case data points within the total interpreted dataset range between 65 and 80 %. This means that 20 to 35 % of the data in the interpreted datasets originate from missed tips, depending on the tipping bucket and the flow rate (i.e., there are more missed tips at higher flow rates). Since the hydrological characterization is based on this interpolated data, flow data is attributed an error of ± 20 to 35 %, depending on the lysimeter. Appendix E includes further discussion.  Despite some relatively large error values in the resulting dataset, results of the RANSAC model are determined to be applicable for the purpose of this investigation. Unless otherwise stated, interpolated outflow is considered as outflow data for the remainder of this discussion.  3.3.2  Total Pile Discharge  A total of 1175 m3 of water drained from the base of the pile throughout the study period from 18-Jan-2007 through 30-Sept-2008. An additional 59 m3 of water is accounted for based on data points removed due to maintenance or whole periods in which the datalogger did not record data from a given tipping bucket but field observations confirm that the tipping bucket was flowing (Appendix E). A total outflow of Q = 1234 m3 of water is estimated to have discharged from the base of the pile during the entire 21 month study period. A majority of this outflow is reported by basal lysimeter D, but the sub-lysimeters allow for the investigation of smaller-scale (i.e., higher resolution) processes that occur within the context of a larger waste rock volume. A summary of  68  lysimeter contribution and accounting for any removed or estimated flow at each lysimeter is given in Table 3.3. Of an estimated total of 1234 m3 of pile outflow over the entire study period, basal lysimeter D contributes QD = 1202 ± 231 m3, whereas sub-lysimeters C, B and A contribute QC = 23.4 ± 7.1 m3, QB = 2.8 ± 0.7 m3 and QA = 6.4 ± 1.4 m3, respectively. The source of the errors is described in Appendix E. The basal lysimeter represents 97.4 % of the total pile effluent, whereas sublysimeters C, B and B represent 1.9, 0.2 and 0.5 %, respectively.  3.3.3  Outflow Hydrographs  Outflow from the basal lysimeter and three sub-lysimeters is presented as unit hydrographs that show the evolution of outflow [m3/day] over time (Figure 3.9). Each hydrograph illustrates peak discharges in response to heavy precipitation events, and both the rising and receding limbs that result on either side of a significant outflow peak. Outflow from the base of the pile is nearly 50 % of the total precipitation over the entire study period and 41 % of the precipitation during the 2007-08 water year (Table 3.2).  All four outflow hydrographs demonstrate highly variable flow rates that differ by several orders of magnitude depending on the location and time of year. A strong seasonal response in outflow is apparent, with high- and low-flow periods generally corresponding to the wet and dry seasons (October through April and May through September, respectively). All lysimeters respond to event-driven precipitation inputs, with a variable lag time depending on the lysimeter location and height of material above it. Each lysimeter also illustrates a characteristic drying period during which the pile is draining to near-residual saturation. This is particularly apparent in the basal lysimeter hydrograph (Figure 3.9d). Aside from some similarities between lysimeter hydrographs, each lysimeter exhibits its own distinct characteristic response to precipitation (Figure 3.9a). The average pile flow rate over the entire study period is approximately Q = 2.0 m3/day, with wet and dry season averages of nearly Q = 3.0 and 0.5 m3/day, respectively. Lysimeter flow  69  results exhibit similar characteristics to those in a 5 m high waste rock test pile in northern Saskatchewan presented by Nichol et al. (2005) in which:  1. peak flow rate (magnitude and timing) was different among lysimeters, 2. lysimeters with the highest flow rate showed the most rapid response to precipitation whereas those with the lowest flow rates showed the slowest response, and 3. various flow mechanisms (matrix flow, macropore flow, displacement, etc.) are reflected in the data  These observations are discussed throughout the remainder of this chapter.  3.3.4  Hydrograph Variability  Total wet and dry season outflow Q [m3], weekly outflow rates [m3/week] and average outflow flux [mm/day] for each lysimeter is compared to the total precipitation during the same period (Table 3.1). Since the wet and dry seasons consist of 7 and 5 months respectively, results are presented as time-averaged values. In addition, because the study period begins in the middle of the 2006-7 wet season (Jan-2007), data for that season is incomplete and unrepresentative of the entire wet season. A larger volume of water from precipitation corresponds to a larger volume of discharging water. This is not a direct correlation, as each lysimeter responds differently to precipitation events.  Outflow response to precipitation events is observed anywhere between a few hours and several weeks after a high-intensity precipitation event. The maximum outflow rates during the 21 month study period is QD = 14.7 and QC = 0.44 m3/day for basal lysimeter D and sub-lysimeter C, respectively, on 25-Feb-2008 (Figure 3.9). Unfortunately, there is no site precipitation data available at this time (Section 3.1.1; Appendix C). Nearly 4 km away, 75 mm of rain was measured at the Yanacancha rain gauge during the entire week, which is much higher than the average weekly wet season site precipitation of 45 mm/week (Table 3.1). For the study period, a maximum of QB = 0.038 m3/day is reported for sub-lysimeter B on 27-Feb-2008 and QA = 0.038 m3/day for sub-lysimeter A on 27-Jan-2008.  70  These maximum values represent daily flow rate averages and do not explicitly indicate the maximum observed flow rate that could have been sustained for shorter periods of time within the day. For example, basal lysimeter D reaches an equivalent flow rate as high as QD = 21.6 m3/day that is sustained for a few hours and is 50 % higher than the maximum calculated daily flow rate. This is based on the highest outflow measured by the large tipping bucket, QD = 15 L/min. Similarly, sub-lysimeter C reaches equivalent flow rates as high as QC = 0.8 m3/day (0.54 L/min) for a few hours, which is a 75 % increase over the maximum calculated daily flow rate. Sub-lysimeters A and B both report maximum flow rates of QA, QB = 0.052 m3/day (0.036 ml/s). These maximum flow rates are important when comparing the area-normalized flow behaviour between the lysimeters (below). Additionally, this information will be useful in subsequent hydrological models in which it is important to know the maximum outflow volumes expected from this test pile.  It is difficult to calculate a minimum flow rate for the lysimeters, since all four lysimeters have gaps in the data due to instrumentation maintenance/problems or periods in which no water drained to the lysimeters. Site observations confirm that all three sub-lysimeters stopped flowing during certain periods of the dry season, whereas the large lysimeter could flow continuously throughout the year (if maintenance was not required). The lowest reported flow rate at basal lysimeter D is slightly more than QD = 100 L/day (0.1 m3/day), and is observed at the end of both dry seasons (i.e., September/October of 2007 and 2008). The lowest reported sub-lysimeter flow rate (greater than 0 ml/s) is nearly Q = 0.15 L/day (1.5 x 10-4 m3/day), and occurs near the end of the dry season (Figure 3.9).  When all four lysimeters are flowing, the basal lysimeter outflow response is representative of the entire pile response, and can be conceptualized as a composite average of the spatiallyvariable sub-lysimeter responses. When any or each of the three sub-lysimeters cease to flow, the basal lysimeter response is assumed to be an approximate representation of the entire pile because D contributes a majority of the pile response (Section 3.3.2).  71  3.3.5  Lysimeter Net Infiltration  Lysimeter responses are compared against one another by normalizing the flow rate by the area footprint (q = Q/A = mm/day; sub-lysimeter area = 16 m2, and basal lysimeter area = 1248 m2). The resulting discharge flux represents the cumulative net infiltration at each lysimeter. For example, Table 3.3 presents the cumulative volume of water that is recorded at each lysimeter during the course of the 2007-08 water year. Based on this data, the net cumulative infiltration calculated at the basal lysimeter is qD = 540 mm/year, whereas sub-lysimeter C has a higher value of qC = 890 mm, and B and A have qB = 110 and qA = 280 mm/year, respectively. For comparison, the net precipitation during the water year is qPile = 1297 mm/year (Table 3.2). Similarly, the daily infiltration flux for each lysimeter is also calculated and compared to the daily precipitation flux (Figure 3.10).  In response to a heavy precipitation event, basal lysimeter D and sub-lysimeter C experience larger, more rapid infiltration fluxes than sub-lysimeters A and B, which experience more attenuated flux responses. These differences are likely due to the effect of the batter, which permits rapid, less attenuated flow responses to precipitation. The daily infiltration flux at sublysimeter C is almost an order of magnitude higher than sub-lysimeters B and A, and reaches up to 150 % of the maximum daily infiltration flux reported at basal lysimeter D, particularly in response to two large precipitation events in Jan- and Feb-2008 (Figure 3.10a). During these two events, the daily infiltration flux for sub-lysimeter C is larger than the precipitation flux, possibly indicating that flow is focused from other parts of the pile to C. The flow response of the batters disproportionately affects the spatially- and temporally-variable hydrogeological response of the pile. This behaviour is also observed at a 5 m-high waste rock test pile at the Cluff Lake mine, Saskatchewan, Canada (Nichol et al., 2005).  3.3.6  Effect of Antecedent Moisture Content  In dry soils, matric forces usually dominate over gravitational forces. In response to a precipitation event, a significant proportion of water is drawn into the unsaturated matrix to replenish the moisture content (Baver et al., 1972). Surficial infiltration rates are highest at the start of a rainfall event (or wet season) as high matric forces (capillary, adsorption) combine with gravity to induce strong infiltration rates until moisture content stabilizes and the unsaturated 72  hydraulic conductivity is roughly equal to the infiltration rate (Smith and Beckie, 2003). Therefore, antecedent moisture content is an important consideration when examining the hydrological behaviour of a given material.  Pile construction was completed in Sept-2006 and the data was online in Jan-2007 (Chapter 3). The waste rock was likely extremely dry (i.e., below residual saturation) due to drying from the blasting process and further evaporation during the end-dumping process. (Note. this is not the case for the protective layer which was stockpiled on site for up to 3 months (Chapter 2) and likely had a moisture content above residual saturation). The wet-up period of 3 - 4 months confirms that the pile was initially dry and that a large proportion of infiltration was taken up into storage in the few months after pile construction was complete. It is for this reason that the water year was set to begin almost one year after pile construction was complete and the pile is assumed to have adjusted to any construction disturbances.  Based on site precipitation and the hydrograph response of the basal- and sub-lysimeters in the test pile (Figure 3.4; Figure 3.9d), the pile is at its driest point at the end of the dry season or beginning of the wet season (September/October). This is further confirmed by the Time Domain Reflectometry (TDR) moisture content data discussed below (Section 3.4). When the wet season rains begin, the pile is at its driest and a considerable proportion of the infiltrating water is retained under the high initial matric forces. Since the volume of water entering the pile is greater than the volume of water discharging from the base of the pile, the overall water storage of the pile begins to increase (as indicated by the TDR data discussed below). The extent at which this will occur is dependent on the pile dimensions as well as moisture content, hysteresis, material wilting point and field capacity, relative proportion of fine-grained material in the pile and net infiltration, all of which control the flow of water through the pile and how much moisture the pile material can store (Smith and Beckie, 2003).  Lysimeter discharge rates are accordingly lowest at the end of the dry season/beginning of the wet season. Discharge rates increase days to weeks after the onset of the wet season, depending on the location within test pile 1. For example, beginning on 24-Feb-2007, nearly 280 mm of rain is introduced during 50 near-consecutive days of rain after an uncharacteristically dry 73  February (Figure 3.10b). Nearly 7 days after this high-precipitation period begins, the daily infiltration rate at basal lysimeter D responds on 2-Mar, whereas sub-lysimeter C infiltration responds nearly 3 - 4 days after the basal lysimeter on 6-Mar. Both continue to experience increasing flow rates for the next 2 - 4 weeks, after which they attain relatively high flow rates. Flows to sub-lysimeters A and B are more attenuated and do not respond until almost 19 to 22 days after the onset of the 50-day precipitation period. Nearly one week after, they reach relatively high, sustained flow rates (Figure 3.10).  Similar trends at the onset of the 2007-08 wet season (i.e., Oct-2007) are observed in basal lysimeter D fluxes, which show delays up to 2 weeks after the wet season begins (Figure 3.10a). No sub-lysimeter data are available for this period and it is likely that they were dry during a portion of this period. Flow to sub-lysimeter C indicates higher infiltration rates at the end of October, whereas sub-lysimeters A and B do not experience significant increases until December.  3.3.7  Wet Season Responses  Once the pile has reached a state of dynamic equilibrium over a one year time scale (i.e., has wet-up and water contents always exceed residual saturation) and flow at all four lysimeters exhibit high flows that are representative of the wettest pile conditions over the course of a water year, several flow responses become evident in the data. These include preferential macropore flow, pressure-wave responses and non-vertical flow.  3.3.7.1 Preferential and Macropore Flow The nonspecific descriptive flow that occurs in certain areas at faster rates than normal in response to a rainfall event is called ‘preferential flow’, and is believed to occur due to collections of large connected pores or obstacles that concentrate flow to certain pathways. Preferential flow is typically activated under high-intensity and high-infiltration precipitation events when the matrix water content is high and matric pressures are relatively low (Chapter 1). During these events, water can penetrate to depth at velocities much higher than would be expected for matrix flow alone.  74  Based on lab and field-scale experiments, Fala et al. (2005) observed that preferential flow can be caused by continuous macropores, or by capillary flow through vertical, horizontal or inclined layers of relatively high hydraulic conductivity, and that it has the ability to control the movement of water within a pile. Capillary flow is typically subject to Darcian principles and occurs in the finer-grained matrix, whereas macropore flow is typically referred to as noncapillary flow, is not subject to Darcian principles, and generally occurs within the coarsergrained pile regions. Preferential flow that occurs over timescales of days to weeks is typically associated with capillary flow, whereas non-capillary flow generally occurs over hourly timescales. Macropores have been reported to rapidly channel water much faster than the capillary-dominated matrix-flow velocities (Germann and Beven, 1981; Eriksson et al., 1997; Newman, 1999; Nichol et al., 2005).  While the antecedent moisture condition is shown to be an important consideration in characterizing pile hydrology (Section 3.3.6), the intensity of a precipitation event can also play a critical role in pile response. Based on a qualitative examination of the outflow data from test pile 1, there is a clear response in outflow (particularly near the surface) any time in which the daily precipitation exceeds 8 - 10 mm/day. This is also observed in the moisture content data (Section 3.4), and can sometimes be resolved in the water chemistry as the chemical signature of macropore flow is typically fresher and has a lower dissolved load than the typical matrix pore water (as discussed in Chapter 4). The following section will examine the evidence of preferential flow in the test pile in further detail.  Wagner et al. (2006) observed that matrix flow was the dominant type of flow during periods with low or moderate infiltration, and that the macropores were only episodically wet. Similarly, for the Antamina test pile, only when sub-lysimeters A and B report larger sustained flow rates during the latter half of the wet season did basal lysimeter D and sub-lysimeter C show direct responses to precipitation events (i.e., when precipitation ≥ 8 -10 mm/day). It is hypothesized that when sub-lysimeters A and B encounter high sustained flow rates characteristic of mid- to endof the wet season, this indicates that the pile matrix has reached a state in which the moisture content is relatively high compared to the rest of the year. Throughout the remainder of this discussion, when the pile attains this condition it will be referred to as the ‘wet season wet-up’ 75  that describes the wettest conditions that the pile experiences when under a dynamic equilibrium over a one year time scale. Only when the pile has attained the ‘wet season wet-up’ state and ponded water is available at the surface (to provide free water), can macropores may be activated under high intensity precipitation events. Even if the pile has achieved the ‘wet season wet-up’ state, if there is no ponded water at the surface then macropores will act as barriers to flow.  Sharp responses to precipitation events are apparent in basal lysimeter D and sub-lysimeter C data. Basal lysimeter D consistently shows the first response to a precipitation event, shortly followed by C which typically exhibits a higher flux compared to basal lysimeter D (Figure 3.10). A faster response in the basal lysimeter is likely caused by a rapid infiltration of water at the pile edges, whereas water has to infiltrate (or be displaced by a pressure wave, see below) further to get to sub-lysimeter C (6 - 9 m) at the base. Sub-lysimeter C typically has a higher infiltration rate than basal lysimeter D in response to peak events, which may be explained by non-vertical flow paths that focus water into a preferential pathway.  Between 15- and 23-Mar-2007, sub-lysimeter B also reports a flux rate increase from qB = 0.14 to 1.5 mm/day, and sub-lysimeter A reports a flux increase from qA = 0.22 to 1.7 mm/day between 15- and 27-Mar-2007. Under these conditions, the pile is described as attaining the ‘wet season wet-up’. During the next month, in response to a 32 mm-precipitation event between 8and 10-Apr-2007, basal lysimeter D and sub-lysimeter C fluxes begin to increase rather sharply (Figure 3.10b). The gradient of the rising limbs of both C and D fluxes is nearly identical (which is not usually the case), and both experience very rapid flux rate increases over the two days of heavy sustained precipitation (by as much as 250 % from their respective fluxes a few days prior). As basal lysimeter D and sub-lysimeter C fluxes are increasing, both sub lysimeter A and B fluxes are decreasing, and don’t begin to increase until 4 (sub-lysimeter B) to 7 (sub-lysimeter A) days after the rain event began. During the following 6 - 8 weeks, sub-lysimeters A and B maintain relatively high sustained flow rates, and D and C fluxes experience rapid responses to large precipitation events greater than 8 - 10 mm/day. Similar observations are also reported during the 2008 wet season.  76  3.3.7.2 Pressure-Wave Displacement of Water through the Pile The maximum observed pile outflow flux (i.e., a combination of the basal lysimeter and three sub-lysimeters) is 12 mm/day. Based on this, it would take water over 800 days to travel through the complete 10 m-depth of the pile and over 80 days to travel through a 1 m-deep edge of the pile (i.e., pile toe). These calculations do not account for porosity or water content (Section 3.5.3, below). However, the previous section explained that outflow responses at the base of the pile are observed only a few days after a large precipitation event. While it is likely that a proportion of this outflow water results from macropore flow, it is also likely that the rapid outflow responses to precipitation events (after the pile has attained the ‘wet season wet-up’ state) are due to a ‘pistoning’ or displacement of water through the pile that result from a pressure-wave. This means that water reporting at the base of the pile is likely pre-event water (i.e., water already stored in the pile) and does not include the same water molecules that originated from that precipitation event just a few days prior. This applies more to the pile interior (10 m height) than it does to the more permeable batters (variable height). Similarly, Nichol et al. (2005) observed that despite calculated macropore flows as high as 5 m/day, a majority of outflow from an applied tracer at the Cluff Lake waste rock pile consisted of pre-event water. These displacement mechanisms in response to high-intensity precipitation events after a period of wet conditions may be the result of a pressure-induced wave that is discussed further using soil moisture data from the Time Domain Reflectometry probes (Section 3.5.2, below).  3.3.7.3 Non-Vertical Flow The heterogeneous nature of waste rock leads to non-uniform flow. While the general direction of water movement is downwards, it also includes a horizontal flow component. Under high infiltration events, an increase in the vertical flow component may result in an increase in the magnitude of the horizontal component as well, increasing the collection area (or footprint) of a given flow pathway. It is also possible that flow with a considerable horizontal component can occur due to structural differences such as obstacles (i.e., a large boulder) or interbedded fineand coarse-grained layers. These interbedded layers are typically a result of the way in which the waste rock material was placed (i.e., end-dumped), and create/facilitate capillary barriers and/or transport pathways (Chapter 1). When infiltrating water exhibits a relatively significant horizontal flow component, it can lead to a situation where the outflow of a given collection area 77  is greater than the expected outflow based on the precipitation flux for the same area over the same time period (Neuner, 2009). This suggests evidence of behaviour that is referred to as nonvertical flow. There is evidence of this type of flow in the test pile hydrology, particularly under relatively large precipitation events.  Flow to sub-lysimeter C reports infiltration fluxes that exceed the daily precipitation flux at several points during Mar-2007, whereas basal lysimeter D is consistently below the precipitation flux (Figure 3.10b). During the following wet season, basal lysimeter D continues to remain below the precipitation flux, and sub-lysimeter C only exceeds the precipitation flux under heavy sustained precipitation periods centered on 16-Jan and 27-Feb-2008 (Figure 3.10c). For example, on 16-Jan-2008, the net infiltration flux at C is 27 mm/day, while basal lysimeter D approaches almost 9 mm/day. Precipitation is near 10 mm/day. No daily precipitation is available for the 27-Feb event (Section 3.1), but it is likely that the flow to sub-lysimeter C also exceeds the precipitation flux on this day as well based on the nature of the response of sublysimeter C.  That the daily infiltration flux at sub-lysimeter C exceeds the daily precipitation flux at select times indicates that water may be focused to certain locations of the pile and non-vertical flow mechanisms may be active within the pile under high precipitation events. While there is a possibility that that these large peaks may be overestimations of the actual flow based on the way in which the flow data is interpolated (Appendix E), this analysis assumes that these large peaks are representative of pile outflow.  3.3.8  Difficulty Identifying Different Flow Pathways and Multiple Arrivals  Nichol et al. (2005) found that several spatially distinct flow paths were identifiable in the outflow data from a 5 m-high waste rock pile at the Cluff Lake mine (Chapter 1). The different flow paths were described as occurring due to spatial heterogeneity, non-capillary flow, saturation or presence of boulders. While it is most probable that multiple flow paths are present in the Antamina test pile, they could not be resolved in the outflow hydrographs (even at higher resolution e.g., shorter time intervals). This may be due to the data interpolation and filtering  78  (smoothing) schemes in which the data is processed and the way in which it is presented (Appendix E).  Another possible explanation (though unlikely) is due to the lysimeter design. Outflow that reports to the 36 x 36 m basal lysimeter or one of the three 4 x 4 m sub-lysimeters likely consists of water originating from various matrix and (possibly) macropore flow paths. Matrix flow is generally more uniform and diffuse compared to the sporadic, more rapid, smaller-scale pulses of water associated with macropore flow. Due to a large collection area of the lysimeters, outflow can be conceptualized as spatially averaging the different matrix and macropore (if activated) flow pathways. That various flow pathways cannot be distinguished may also be due to the fact that a 1.5 m-thick waste rock protective layer (slightly compacted by heavy machinery during placement) may have dissipated the macropore slug making it harder to resolve in the lysimeter data.  3.3.9  Uncertainty  It has already been explained that a large range of error is associated with the outflow data due to a considerable amount of missed tips in the tipping bucket data (Section 3.3.1; Appendix E). However, there is another level of uncertainty that results in less reliable data at higher flow rates. This primarily applies to the recording system that measures outflow from sub-lysimeter C, which does not function optimally under very high flow rates (Appendix E). Based on a 3-second scanning resolution, the datalogger reports each tip in 3-second intervals, regardless of when the tip occurs. Based on tipping bucket design (funnel aperture, chamber capacity; Appendix E) it is nearly impossible for two tips to occur within one 3-second interval. Nevertheless, when plotted on a ‘tip-time vs. time’ graph, this creates contiguous, coarse resolution flow data under high flow rates. In reality, the tips do not occur at three second intervals, yet it is nearly impossible to determine at what point within a given interval the tip actually occurred. (Note, this scanning resolution interval has subsequently been changed to 0.125 seconds).  In probability and statistics, the continuous uniform distribution is a family of probability distributions that describes a statistical situation in which it is equally probable to have a tip at any time within a given interval. Applied to the tipping bucket data, it states that there is an equal 79  probability that a tip will occur at any point in the 3-second interval. For a random variable such as a tip, the expected value m is calculated as:  𝑚𝑚 =  𝑎𝑎+𝑏𝑏 2  [3-5]  where the minimum and maximum interval bounds are a and b respectively. When applied to the tipping bucket data, a is always 0 seconds and b is always 3 seconds, resulting in an expected value of m = 1.5 seconds. Therefore, flow data is corrected by adding 1.5 seconds to each interpolated tip-time (interpolated time between consecutive tipping bucket tips) before it is converted to a flow rate (Appendix E). Since this data represents an average between the minimum and maximum bounds, it is referred to as the ‘expected’ flow data. It is possible that the tip-time did not always occur midway between a and b, so the minimum and maximum flow rates are used to represent the lower and upper flow rate bounds. The interpolated data represents the lower bound of a given interval (i.e., a = 0) and the upper bound (estimated maximum) flow rate (recall that, in general, there is an inverse relationship between tip-time and flow rate; Appendix E). Conversely, the upper bound of a given interval (i.e., b = 3) requires that 3 seconds is added to each tip-time and represents the lower bound (estimated minimum) flow rate. All of the results previously presented in this chapter were based on the ‘expected’ flow dataset. A summary of the expected, upper and lower bound estimates at each lysimeter is presented in Table 3.3.  Due to the nature of the calibration equation (Appendix E), the upper and lower bound flow rates (a and b in Equation 3-5) diverge during higher flow rates and converge during lower flow rates, effectively illustrating that there is a larger error bound at very high flow rates. Departure from the expected value is asymmetric due to the nature of the calibration equation, with the upper bound exhibiting a larger departure from the expected value than the lower bound. Figure 3.11 illustrates how the error bounds relate to the expected volumetric flow data for basal lysimeter D and sub-lysimeter C. Note that sub-lysimeter C has a much higher range in values at the highest observed flow rates (i.e., upper bound is twice the value of lower bound flow rate), indicating the lowest reliability for this tipping bucket data under very high flow rates (Figure 3.11b). To note, this means that despite assuming a uniform probability of a tip falling within a given 3-second 80  interval, the resulting flow rate will not exhibit a uniform distribution based on the non-linear nature of the calibration equation (Appendix E). Even still, the flow pattern does not change dramatically between the upper and lower bounds.  Based on the entire study period, the cumulative upper and lower bound estimates for basal lysimeter D are 3.3 and 3.0 % higher and lower than the expected flow, respectively. The upper and lower estimates for sub-lysimeters A and B do not vary by more than 1 %, whereas the lower bound sub-lysimeter C estimate is 7 % less than the expected flow, and the upper bound is 10 % higher. This indicates that flow rates at sub-lysimeter C are mostly affected by the coarse scanning resolution of 3 seconds. These error bounds are not as large as those presented based on the data interpolation scheme (± 20 – 35 %; Section 3.3.1), and will not be discussed further.  3.4  MOISTURE CONTENT  The change in moisture content of an unsaturated porous medium provides insight into the flow behaviour and storage capacity of that medium. Time Domain Reflectometry (TDR) probes were therefore installed to characterize the moisture regime of test pile 1.  3.4.1  Time Domain Reflectometry Data  Twenty-two TDR probes were installed inside the pile to provide continual moisture content data (Corazao Gallegos, 2007). Each probe is connected to a Moisture Point MP-917 TDR instrument (www.esica.com) located within the instrumentation hut. When instructed, the MP-917 box directs a pulse to a single TDR probe, calculates the travel time difference, and then outputs the result to the datalogger. An incorrect setting in the TDR instrument resulted in unrecoverable TDR data from the first day of the study period until 8-May-2008 (the beginning of the 2008 dry season), when the wrong setting was corrected (Appendix F).  Since the study period for this investigation extends until 30-Sept-2008, this means that the available TDR data during this 145-day period only occurs during the 2008 dry season. During this time, TDR data indicates that the pile is draining water but does not show any considerable responses to precipitation because no significant rain events occur during this period. Therefore, for the purpose of examining the moisture content regime of the pile, the study period is 81  extended to include TDR data from 1-Oct-2008 through 22-Jan-2009, as this is sufficient to observe the re-wetting of the pile (i.e., as it achieves the ‘wet season wet-up’ state) during the beginning of the 2008-9 wet season. Approximately 585 mm (760 m3 water) of precipitation and 225 mm (292 m3) of outflow from the basal lysimeter are calculated during this additional period (based on similar methodology used for the previous data). Raw TDR data (travel time offset in nanoseconds) is calibrated to m3/m3 (volume of water/bulk volume), but is generally presented as a volumetric water content percentage (% VWC).  3.4.2  TDR Probe Response  Two TDR probes are located in instrumentation lines 5 and 6 near the base, approximately 10 m below the surface (Appendix A). The remaining twenty are located in 4 near-vertical instrumentation lines (5 probes in each) at average depth intervals of 1.75 m below the surface (Appendix A). These probes are used to track a wetting front of moisture content as it penetrates to depth. The wetting front is estimated based on the time between the peak precipitation and the peak increase in moisture content (measured as the average between the upper and lower moisture content inflection points associated with that wetting front signal). TDR data is used to corroborate observations that were previously discussed in this chapter.  3.4.2.1 Protective Layer Two TDR probes were installed within the 2B-Protective Layer of the pile base in instrumentation lines 5 and 6 (Appendix A). It was decided to place these probes in the 2B material to determine the moisture variability of the 2B-Protective Layer to gain a better understanding of the effect of the protective layer on the flow in the pile and chemistry. On physical grounds, it would be expected that the 2B-Protective Layer largely remains saturated because the drainage system is 0.35 m above this 2B-Protective Layer and water has no place to exit from. The TDR probe in line 5 consistently exhibits a high moisture content ϴ = 45 % VWC that does not vary throughout the study period, possibly indicating that:  1. the probe is not functioning properly, 82  2. the calibration does not work well for the 2B crushed material, or 3. the 2B-Protective Layer is saturated.  Since this is the lowest probe in the pile, it may possibly indicate that the 2B-Protective Layer is saturated and the probe is below a water table. A VWC of 45 % is higher than expected, but falls within the calculated porosity range of n = 0.4 ± 0.1 for test pile 1 waste rock (Chapter 2). No porosity is known for the 2B-Protective Layer. While it is possible that this probe is located below the water table (if one is present), no conclusions can be confirmed until more data is obtained and a soil-specific TDR calibration equation is obtained for the test pile waste rock (Appendix F) and possibly the 2B-Protective Layer. The line 6 TDR probe shows more of a seasonal response, with ϴ = 25 % VWC in October and begins to increase shortly after in response to precipitation inputs until a maximum of ϴ = 31 % VWC is attained on 22-Jan-2009. That water content of this probe is variable likely indicates that it is above the water table (if one is present).  3.4.2.2 Compacted Surface The shallowest TDR probes in the pile (i.e., the highest TDR in lines 1 - 4) are located 1.6 – 2.4 m below the surface (Appendix A). The VWC for these shallow probes ranges between 11 - 30 % VWC, with the lowest values (11 – 18 % VWC) in September/October and the highest values (18 – 30 % VWC) in January/February (Figure 3.12).  Consecutive days of precipitation that are characteristic of the wet season begin in the second week of Sept-2008, yet these probes do not register a moisture content increase until 9-Oct, nearly 5 weeks after the rains begin. At this point, moisture content begins to increase relatively rapidly over the next few days in response to 30 mm of precipitation that falls over the course of 4 days (Figure 3.12). In order for these responses to occur, they need:  1. a sufficient volume of infiltrating water is required prior to these responses, and  83  2. a significant precipitation event (or consecutive events) to trigger a sharp response in moisture content. Without these event(s), infiltrating water from prior events will not likely result in a sharp increase in moisture content.  The highest probe in line 4 (L4TDR5) responds much earlier because it is under the batter (explained in the following section). That these TDR probes do not show any responses until 5 weeks after the heavy precipitation events began indicates that:  1. the permeability of this material is relatively low compared to the rest of the pile, particularly compared to the batters (Figure 3.8), 2. the rain water is initially taken up by matric forces to replenish the moisture content, which influences the rate at which the wetting front con propagate (Section 3.3.6), 3. matrix flow is dominant here, 4. water may be by-passing the TDR instruments, and/or 5. only a large rain event (> 10 mm/day) will lead to identifiable short-term changes in moisture content.  It has already been discussed that this material is relatively fine (due to preferential settling during the end-dumping process) and compacted (due to the heavy dump truck traffic during pile construction) compared to the rest of the pile. TDR data supports these observations and indicates that this material has a relatively low permeability compared to the pile batters (Chapter 2). In addition, that the shallow probes from instrumentation lines 1, 2 and 3 all indicate rather sharp rises in moisture content at around the same time indicates that the wetting front moves relatively uniformly through the upper 1.6 – 2.4 m of the pile, with an average wetting front velocity vwf of 0.06 m/day (7 x 10-7 m/s). Recall that results of the infiltrometer (Chapter 2) demonstrate that the average hydraulic conductivity of this material is 2.3 m/day (3 x 10-5 m/s). That these two rates differ by almost two orders of magnitude is due to the differences in methodology and the fact that they are both at different points of the ‘hydraulic conductivity – moisture content’ characteristic curve. For instance, the infiltrometer test was conducted in March, more than halfway into the wet season, and the material is likely the wettest it is throughout the entire year. Conversely, the TDR wetting front velocities are calculated based on 84  data from September/October, which is the beginning of the wet season when the material is relatively dry and some of the infiltrating water is taken up by matric forces.  It is also likely that no large moisture content changes are evident in the first 5 weeks of precipitation because no large precipitation events (> 10 mm) occur. Some moisture contents do respond a few days after a large precipitation event of 12 mm occurs on 7-Oct-2008.  3.4.2.3 Batter Instrumentation line 4 is located at the front of the pile, under the pile batter (Appendix A). Since each of these four TDR probes (the fifth does not function) are approximately 1 - 2 m below the surface, they are also ‘shallow’ probes. However, these probes differ from the probes at the top of the pile (above) because the batter consists of much coarser material than the top of the pile (Figure 3.8). Due to these differences in material, moisture contents are very different under the batter.  Line 4 experiences the largest variation in moisture content, which varies between 13 and 35 % VWC (Figure 3.13). TDR probes under the batter report clear and distinct responses to a 10 mmprecipitation event on 9-Sept, which is preceded by 3 days in which 10 mm fell. These probes responded almost one month prior to responses under the top layer. This corresponds to wetting front velocities as high as vwf = 0.5 m/day, which is one order of magnitude higher than calculated for the upper compacted layer. This wetting front velocity is similar to those calculated at two other waste rock test piles in which Marcoline (2008) and Neuner (2009) report the wetting front velocity to range between vwf = 0.1 and 1 m/day, and vwf = 0.1 – 5 m/day, respectively, using similar TDR probes.  TDR probes near the bottom of line 4 (L4TDR1 and L4TDR2 in Figure 3.13) show an increase from 13 % VWC near the end of the dry season (1-Sept-2008) to 36 % VWC during the wet season (19-Jan-2009). TDR probes near the top of line 4 (L4TDR4 and L4TDR5; L4TDR3 does not function) increase from 18 to 30 % VWC over the same period. All four probes show rapid responses to intermediate- and large-precipitation events, with the highest volumetric water change of nearly 3 % observed over the course of one day. Nearly six weeks later, these probes 85  reach maximum moisture contents, and remain near these levels for the remainder of the period (Figure 3.13). That L4TDR1 and L4TDR2 moisture contents differ from each other by almost 8 % throughout the period whereas L4TDR4 and L4TDR5 probes exhibit very similar responses likely indicates the heterogeneous nature of the waste rock pile.  3.4.2.4 TDR Variability TDR results are generally similar to Nichol (2002); Nichol et al. (2003), Marcoline (2008) and Neuner (2009), all of which report highly heterogeneous flow through waste rock and higher wetting front velocities than fluid velocities. Results show that, in general, the deepest probes encounter the highest sustained water contents compared to the shallowest probes under the top layer of the pile, which typically report the driest values and experience the largest range in moisture contents. Deeper probes also respond several weeks after the onset of the wet season, whereas probes under the unconsolidated batter respond only a few days after the onset of the wet season. These observations are generally consistent with Nichol (2002) who reports that water content changes from TDR probes near the surface are generally larger (of a 5 m-high waste rock pile at the Cluff Lake mine) in comparison to deeper probes where water content changes in response to the largest rainfall events are less. That some shallow probes report higher moisture contents than deeper probes at test pile 1 is likely a result of:  1. material heterogeneity surrounding each probe, 2. by-passing of water around TDR probes, and/or 3. preferential and/or non-vertical flow.  Based on a moisture content range between 11 and 36 % VWC, the residual saturation can be expected to be around 11 % VWC. The average moisture content measured during the particle size distribution was 3 %. Although the TDR probes measure the water content of the matrix and not the entire range of particle sizes, these moisture content differences indicate that the waste rock moisture content was below residual saturation during pile construction and explains why there was an initial delay in wetting-up during the 2006-7 wet season (Section 3.3.6).  86  3.4.3  Wetting Front Propagation and Evidence of Preferential or Non-Vertical Flow  An example of the wetting front propagation to depth is illustrated using line 1 TDR responses (Figure 3.14). The shallowest TDR probe (L1TDR5) is 2.4 m below the surface, and shows a rapid increase in moisture content on 9-Oct that is likely triggered by a 21 mm-precipitation event on 6- and 7-Oct. Over the course of a few days, the moisture content experiences a steady increase of 10 % VWC from 17 % to almost 27 % VWC. Nearly twenty days later, the closest functioning probe (L1TDR3) at a depth of 5.8 m below the surface, begins to respond. Shortly after, the next closest probe at a depth of 7.3 m (L1TDR2) begins to respond to increasing moisture content, followed by the deepest probe (L1TDR1) at a depth of 9.3 m during the second week of November. That the moisture contents increases in order with depth to surface effectively illustrates the wetting front as it propagates to depth during October and November of the 2008-09 wet season.  While TDR probes may not report macropore flow, it is still possible for them to report preferential and/or non-vertical flowpaths in the matrix. Another example of wetting front propagation to depth that provides evidence of preferential flow is illustrated using line 2 TDR responses (Figure 3.15). The shallowest TDR probe (L2TDR5) is located 1.6 m below the surface, and shows a rapid increase in moisture content on 10-Oct that is likely triggered by 21 mm of precipitation during 6- and 7-Oct. Nearly three weeks later on 1-Nov, both L2TDR4 (3.2 m depth) and L2TDR1 (7.9 m depth) moisture contents begin to increase. Since L2TDR1 moisture content increases before the two probes above it at depths of 4.9 (L2TDR3) and 6.5 m (L2TDR2), this provides evidence of preferential or non-vertical flow. It is likely that the heavy precipitation during the week prior to this increase in moisture content activates some macropores or other flow paths that channel water to depth while by-passing the middle TDR probes. This is supported by Smith and Beckie (2003) who report that because TDR probes are installed in a fine-grained matrix, they may not be able to indicate the presence of preferential flow through large macropores.  The presence of wetting fronts in response to high-precipitation events are more evident near the onset of the wet season when the pile is relatively dry compared to towards the end of the wet season when the pile is relatively wet. This indicates that: 87  1. antecedent soil moisture condition influences the wetting front propagation time, 2. TDR probes are of limited use in tracking fluid flow in wet waste rock, and/or 3. TDR probes preferentially sample from the matrix material in which they are installed and do not likely measure the macropore flow that by-passes the matrix regions.  3.4.4  Change in Storage  Based on observations of a 5 m-high waste rock test pile at Cluff Lake, Nichol (2002) and Marcoline (2008) observed that test pile moisture content returns to similar levels over the course of one water year, and that it can be assumed that there is no net change in storage (ΔS ~ 0) over that time period. No Antamina test pile TDR data yet is available over the course of one water year, so the actual change is storage ΔS could not be determined. However, due to very similar recession curves of the basal lysimeter D hydrograph and a return to minimum flow rates of nearly 0.1 m3/day at the end of both the 2007 and 2008 dry seasons (Figure 3.9d), it is assumed that the pile returns to similar moisture conditions after one water year (i.e., returns to a dynamic equilibrium steady state) and that ΔS ~ 0 over the course of one water year that fully encompasses a wet and a dry season (1-Oct to 30-Sept).  Without moisture data throughout the entire water year, the maximum and minimum water storage limits are estimated based on the 7.5 month-long data in which the TDR data is examined (May-2008 to Jan-2009). The following calculations are only included to provide a relative idea of the potential change in water content of the pile, and should not be considered absolute. Large uncertainty is expected, and is discussed after the calculations are presented.  Based on the available data, the pile is assumed to be at its driest point on 1-Sept-2008 and the wettest on 11-Jan-2009. For the purpose of water balance calculations, the pile volume is conceptually divided into individual discrete volumes, each one centered around a single TDR probe. Average wet and dry volumetric water contents are applied to each volume to obtain a representative water content of the entire pile during the two representative periods. TDR probes in line 1 and 2 are used to provide internal pile moisture contents whereas probes in line 4 are used to provide representative moisture contents along the front batter as well as the north- and 88  south-facing batters. Instrumentation lines 3, 5 and 6 are omitted for simplicity. When no TDR data is available (e.g., L1TDR4 and L4TDR3 do not function) linear approximations are made using nearby probes. Average moisture contents of ϴd = 18 % and ϴw = 25 % are estimated on the driest and wettest days, respectively, and result in an estimated change of storage of S = 850 m3. This corresponds to an estimated change in storage of 8.5 % of the entire pile volume (10,000 m3) over the entire 7.5 month period of study.  It should be reiterated that these calculations are only meant to present the relative differences in moisture content between wet and dry periods, and should not be used as any absolute values until further data is obtained. One of the principal flaws with these calculations is that the TDR data is assumed to represent moisture content of the entire pile when it is already discussed (above) that TDR data only measures the moisture content of the matrix. The presence of boulders and large voids leads to lower moisture contents for the entire pile than that of the matrix (see porosity calculations, Chapter 2). To account for the large boulders, Neuner (2009) multiplied the TDR moisture content by the proportion of fine-grained material to estimate the large-scale moisture content of the entire pile. While this is not relevant for calculating change in storage, it is useful in calculating flow velocities.  3.5  3.5.1  DISCUSSION  Water Balance  Water balance calculations are conducted over the course of one water balance year and include one complete wet and dry season (Table 3.2). The water year extends from 1-Oct-2007 to 30Sept-2008 as explained in Section 3.2.2. Evaporation estimates from waste rock piles are highly complex and can vary considerably depending on the methodology applied. Nevertheless, there is relative agreement when the error ranges are incorporated.  Test pile 1 discharge Q represents 41 % of estimated precipitation P over one water year, which is slightly higher than the discharge observed at a similar investigation (Q = 44 – 55 % P over three consecutive water years) conducted on a 5 m-high uncovered waste rock pile at the Cluff Lake mine in northern Saskatoon, Canada (Marcoline, 2008). At this waste rock test pile site, the 89  M.A.T is -0.3 °C, with the average daily temperature greater than 0 °C for 5 months of the year. Average annual precipitation (including snow) is 460 mm. Even though the climate of the Cluff Lake mine is very different from that of the Antamina mine, that the percentage of evaporation and outflow is relatively similar indicates that behaviour at test pile 1 is relatively typical of a waste rock test pile.  Based on the available data, the maximum estimated change in storage is calculated over a 4.5 month period (and not the preferred water year). Given the large error estimate attributed to the flow data, the storage value of S = 850 ± 298 m3 roughly corresponds to the total volume of water estimated to have discharged from the base of the pile Q = 670 ± 235 m3. This is conceptually easy to understand. According to the steady state constraint over the water year, any water that is added to storage must eventually be removed to comply with ΔS = 0. This means that over the course of the water year, ΔS should theoretically be equal to Q (longer-term trends may vary). That there is relatively good agreement between these two figures provides strong support that:  1. the instrumentation (tipping bucket, precipitation, moisture content) is functioning properly, 2. the data was accurately calibrated/interpolated/analyzed, etc., 3. the assumptions made throughout this chapter are reasonable, and 4. there is no leak in the lysimeter catchment system.  3.5.2  Pore Water Displacement and Macropore Flow Velocity  Smith and Beckie (2003) report that outflow responses in test piles are typically observed prior to TDR responses, indicating that the maximum pore water fluid velocity (i.e., through macropores) generally exceeds the wetting front velocity recorded by the TDR probes. TDR responses are typically slower than lysimeter outflow peak discharges and indicate that the spatially distinct preferential flow paths at high flow rates do not correspond to the majority of TDR probe locations (Nichol, 2002). Marcoline (2008) further states that the outflow flux is the best estimate of pore water velocity while TDR probes represent biased velocity ranges based on the fact that they preferentially sample from only the matrix and are not representative of the 90  overall pile behaviour. Similar results are observed in the Antamina test pile data, but only once the pile has sufficiently attained the ‘wet season wet-up’ state. Otherwise moisture contents near the top of the pile increase from days to weeks before outflow fluxes begin to increase after a considerably dry period (i.e., start of the wet season).  The pore water velocity vpw [m/s] in an unsaturated porous medium is related to the pore water flux q [m/s] and water content ϴ [m3/m3] in the following equation (presented in Chapter 2): v =q/ϴ  [3-6]  The highest flux of water calculated for basal lysimeter D is q = 12 mm/day (Q = 14.7 m3/day). Based on the average ‘wet season’ moisture content of the test pile (ϴ = 25 % VWC), the pore water velocity is calculated using equation 3-6 to be vpw = 0.05 m/day. Conversely, the lowest basal lysimeter flux is q = 0.088 mm/day (Q = 0.11 m3/day), and corresponds to vpw = 5 x 10-4 m/day at the ‘dry season’ water content of ϴ = 18 % VWC. Calculated using moisture contents from the bottom of the pile (e.g., L1TDR1 and L2TDR1 (L4TDR1, L5TDR1 and L6TDR1 are excluded (see above) and L3TDR1 is not functioning properly)), the ‘wet season’ porewater velocity is vpw = 0.04 m/day (ϴ = 32 % VWC) and the ‘dry season’ porewater velocity is vpw = 3 x 10-4 m/day (ϴ = 26 % VWC).  Smith and Beckie (2003) state that wetting-front velocities three to four orders of magnitude faster than the median water velocity have been observed in waste rock piles. Comparing the maximum pore water velocity (vpw = 0.05 m/day) with the maximum wetting front velocity (vwf = 0.5 m/day) for test pile 1, the maximum wetting front velocity is 1 order of magnitude higher than the pore water velocity. This supports the previous statement that rapid outflow fluxes in response to a heavy precipitation event largely consist of pre-event pore water that is displaced through the matrix of the pile (Section 3.3.7.2). This displacement mechanism may be described as a kinematic or pressure wave that moves through the unsaturated profile much faster than the maximum pore water velocity (Rasmussen et al., 2000; Neuner, 2009).  91  The pressure wave mechanism consists of distinct, rapid increases in moisture that propagates to depth in an unsaturated zone in response to precipitation inputs (Neuner, 2009). These waves are analogous to a garden hose - in which opening the tap causes an almost instantaneous increase in outflow well before the water entering at the higher flow rate reaches the end of the hose - and typically displaces pore water when activated. In test pile 1, infiltrating water into the unsaturated zone will typically propagate downwards in the form of a wetting front, which is defined as the interface between the region of higher water content and fluid velocity, and the underlying region with lower water content and fluid velocity (Smith and Beckie, 2003). Using pressure tensiometers to track the pressure wave in response to precipitation events, Neuner (2009) explains that pressure waves are quite common in unsaturated flow conditions, even under relatively minor rainfall events. The pressure wave is dependent on moisture content and will propagate more effectively if the porous medium is fully saturated.  These pressure waves should not be confused with the rapid velocities associated with macropore flow. In the case of macropore-flow, the wetting front may contain particles that travel at the speed of the wave, but wetting fronts are more often in the form of pore water displaced by a propagating pressure wave that results from the presence of infiltrating water. Since Rasmussen et al. (2000) explain that pressure wave diffusivity increases with depth, this likely explains why more displacement is observed in basal lysimeter D and sub-lysimeter C, and that the pressure wave signal may have been largely dissipated by the time it penetrates the full 10 m of the pile (sub-lysimeters A and B).  Results indicate that macropore flow is likely activated under high intensity rainfall events and that flow from pressure-induced matrix water displacement is also likely occurring. A lack of moisture content data throughout the study period does not allow the moisture content to be examined during the high outflow events that were previously examined. In addition, without any pressure data (i.e., from tensiometers) the direct confirmation of pressure waves can only be inferred and cannot be further investigated in this study. Additional work is required to fully characterize flow through the unsaturated zone, as this will affect the evolution of water quality over time.  92  3.5.3  Effects of the Batter on Pile Hydrology  Results demonstrate that evaporation from, and infiltration into, the fine-grained, trafficcompacted top layer of the pile is very different from the coarse-grained, non-compacted batter. Water can pool on the top surface of the pile and be exposed to higher evaporation and lower infiltration rates, whereas higher infiltration rates can occur on the porous slope of the pile and lead to more net infiltration and less evaporation (recall no significant runoff has been observed at the pile). The compacted top surface of the pile represents only 21 % of the total surface area of the pile. This indicates that a majority of the surface area of the pile is dominated by the batter. However, evaporation from the batters may be further complicated because:  1. the sloping nature of the batters leads to a larger exposed area in which rain may fall, and may be less concentrated so infiltration may not be as high (assuming vertical rainfall), 2. there is a higher proportion of internal evaporation at the course, well-aerated batters as wind-driven advection is likely occurring, bringing fresh (i.e., less humid) air into the batters (particularly the front), and/or 3. it is difficult to obtain accurate evaporation estimates for both the unconsolidated batters and compacted top of pile (e.g., Carey et al., 2005).  It is for these reasons that the most reliable evaporation estimate is that from the water balance method, as it is based on empirical calculations that integrate both climatic and meteorological properties.  3.6  SUMMARY  Using the suite of available data from meteorological data and in-situ pile instrumentation, this chapter characterized the hydrological behaviour of test pile 1. A reasonable agreement between the water balance components is attained, indicating that that the data analysis methodology was suitable for the purpose of this investigation. In addition, this chapter has shown that:  1. infiltration to test pile 1 is approximately 41 % of annual precipitation, and evaporation is 59 %, 93  2. flow is spatially and temporally variable, as evidenced by the basal and sub-lysimeter hydrographs, 3. the hydrological response of the pile is clearly dominated by the less consolidated batter compared to the less permeable top of the pile, 4. the average linear porewater velocity ranges from vpw = 0.2 m/year in the dry season to vpw = 17 m/year in the wet season, 5. the highest calculated wetting front velocity (vwf = 0.5 m/day), is one order of magnitude higher than the porewater velocity (vpw = 0.05 m/day), 6. evidence of preferential flow, macropore flow, non-vertical flow, and pressure-wave pistoning is prevalent in the data, and 7. average moisture contents at the end of the 2008 dry season are ϴ = 18 % VWC.  94  TABLES Table 3.1. Total estimated volume, weekly rate, and daily flux of precipitation and lysimeter outflow. Estimated Precipitation Lysimeter Outflow 3  3  [m ]  [m /week]  [mm/day]  Wet1  561  38.1  Dry  209  Wet  [m3]  [m3/week]  [mm/day]  D  C  B  A  D  C  B  A  D  C  B  A  4.2  437  8  0.8  1.5  29.7  0.54  0.06  0.1  3.4  4.86  0.51  0.92  9.6  1.1  95  1.1  0.1  0.4  4.3  0.05  0.006  0.02  0.49  0.44  0.05  0.18  1379  45.3  5.1  604  13.5  1.5  3.9  20  0.45  0.05  0.13  2.29  3.99  0.45  1.15  Dry  303  13.9  1.4  67  0.8  0.2  0.5  3  0.03  0.01  0.02  0.35  0.31  0.09  0.21  Average Wet3  749  41.7  4.6  38.1  10.7  1.2  2.7  24.8  0.5  0.05  0.12  2.84  4.42  0.48  1.04  Average Dry  197  11.7  1.3  9.6  0.9  0.2  0.5  3.7  0.04  0.008  0.02  0.42  0.37  0.07  0.19  Total  2452  27.6  21.2  1202 23.3  2.7  6.3  13.5  0.26  0.03  0.07  1.54  2.34  0.27  0.63  2006-7 2007-82  1  wet season data is not representative of complete wet period because the study period began on 18-Jan-2007 this is the 2007-8 water year 3 calculated over entire period 2  95  Table 3.2. Summary of evaporation estimates throughout the study period and water year. mm m3 % of Precipitation % Error Study Period (18-Jan-2007 to 30-Sept-2008) P  1891  2450  100  ±5  EPAN  1423  1760  75  ± 15  EPM  1617  1890  86  ± 20  Q  949  1230  50  ± 20 – 45  ΔS  N/A  N/A  N/A  N/A  Water Year (1-Oct-2007 to 30-Sept-2008)  1  P  1297  1681  100  ±5  EPAN  887  1150  68  ± 15  EPM  972  1260  75  ± 20  EWB  766  990  59  ± 25 – 501  Q  532  690  41  ± 20 – 45  ΔS  ~0  ~0  ~0  N/A  water balance estimate error combines precipitation and outflow errors.  96  Table 3.3. Cumulative interpreted flow for each lysimeter. Volume of Water [m3] Collection Area  Interpreted1 Lower Bound  Expected  Upper Bound  Manually  Estimated  Removed2  Unrecorded3  Total4 Lower Bound  Expected  Upper Bound  A  3.86  3.91  3.97  0.005  2.47  6.34  6.39  6.45  B  2.55  2.58  2.60  0.026  0.19  2.77  2.79  2.82  C  19.05  20.74  23.15  0.022  2.6  21.68  23.36  25.77  D  1114  1148  1186  2.5  51  1167  1202  1239  Entire Pile  1139  1175  1216  2.6  56.3  1198  1234  1274  1  resulting dataset from RANSAC interpretation equivalent volume of water of data points removed prior to correcting for missed tips due to maintenance, etc. (Appendix E) 3 estimated volume of water not recorded by the datalogger during periods in which site observations confirm lysimeter was flowing 4 sum of reported outflow, equivalent volume manually removed, and estimated unrecorded flow during gaps in lysimeter data  2  97  FIGURES  Figure 3.1. Antamina rain gauge locations with respect to the study site (elevation above sea level in brackets). Missing (Estimated)  Cumulative Precipitation [mm]  2000 1500 1000  Quebrada Antamina  Estimated (Study Period) Yanacancha  500 0  Measured + Estimated  Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Measured + Estimated  Missing (Estimated)  Estimated (Study Period)  Yanacancha  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition  Figure 3.2. Reported and estimated cumulative rainfall compared to the Antamina precipitation. 98  Cumulative Precipitation [mm]  800  Primary Site Rain Gauge 600 400 200  Secondary Site Rain Gauge 0 Sep-08  Nov-08  Dec-08  Feb-09  Figure 3.3. Rainfall comparison between the two site rain gauges.  2000  1500 20  Estimated  1000 10 500  0 Jan-07 Apr-07  Jul-07  Oct-07  Jan-08 Apr-08  Jul-08  Oct-08  Cumulative Precipitation [mm]  Daily Precipitation [mm]  30  0  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition; orange dotted line = gap in precipitation data.  Figure 3.4. Daily and cumulative estimated precipitation.  99  Precipitation [mm/day]  25 24-Hour 12-Hour 6-Hour  20 15 10 5 0 30-Dec  6-Jan  13-Jan  20-Jan  27-Jan  Figure 3.5. Variability of rain as measured by site rain gauge during Jan-2008  a)  b)  Figure 3.6. Class A Pan evaporation a) schematic and b) Yanacancha station, Antamina.  100  3-Feb  25 Precipitation: Measured and Estimated Evaporation: Penman-Monteith Evaporation: Pan Evaporation  Flux [mm/day]  20  Estimated  15  10  5  0  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition; orange dotted line = estimated precipitation data  Figure 3.7. Daily precipitation (measured and estimated) and evaporation (Penman-Monteith and Pan Evaporation).  a) top of pile  b) pile batter  Figure 3.8. Comparison between a) top and b) batter of pile.  101  0.04  a) Sub-Lysimeter A  Data  Volume[m 3/day]  Gap 0.03  0.02  0.01  0 Jan-07 Apr-07  Jul-07  Oct-07 Jan-08 Apr-08  Jul-08  Oct-08  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition.  0.04  Volume [m 3/day]  b) Sub-Lysimeter B 0.03  0.02  0.01  0 Jan-07 Apr-07  Jul-07  Oct-07  Jan-08 Apr-08  Jul-08  Oct-08  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition.  Figure 3.9. Outflow hydrograph for: a, b, c) the sub-lysimeters and d) basal lysimeter (continued on next page).  102  0.5  c) Sub-Lysimeter C  Data Gap  Volume[m 3/day]  0.4 0.3 0.2 0.1 0 Jan-07 Apr-07  Jul-07  Oct-07  Jan-08 Apr-08  Jul-08  Oct-08  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition.  15  10  3  Volume[m /day]  d) Basal Lysimeter D  5  0 Jan-07 Apr-07  Jul-07  Oct-07  Jan-08 Apr-08  Jul-08  Oct-08  *black solid line = wet-dry season transition; black dotted line = dry-wet season transition.  Figure 3.9. Outflow hydrograph for: a, b, c) the sub-lysimeters A, B and C, respectively; and d) basal lysimeter D (continued from previous page).  103  30 Precipitation Basal-Lysimeter D Sub-Lysimeter C Sub-Lysimeter B Sub-Lysimeter A  a) Study Period: 2007-08 Net Flux [mm/day]  25 20 15 10 5 0 Jan-07  Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 *black solid line = wet-dry season transition; black dotted line = dry-wet season transition  Oct-08  **note. missing precipitation from Feb – June, 2008  Net Flux [mm/day]  30 25  b) 2007  20 15 10 5 0 Feb-07  May-07  Apr-07  Mar-07  30 Net Flux [mm/day]  25  c) 2008  Precipitation Basal-Lysimeter D Sub-Lysimeter C Sub-Lysimeter B Sub-Lysimeter A  C  20 15  D  B  A  10 5  0 13-Jan-08 20-Jan-08 27-Jan-08 03-Feb-08 30-Dec-07 06-Jan-08 Figure 3.10. Precipitation and lysimeter fluxes for a) the study period b) February-May, 2007, and c) Jan-2008.  104  20 Volume [m3/day]  Volume [m 3/day]  16  18  12  a) Basal Lysimeter D  15 12 9 6 3  Upper Bound  0 5-Jan  19-Jan  2-Feb  16-Feb  Expected  1-Mar  Lower Bound 8  4  0 Jan-07  Apr-07  0.7  3  Volume [m /day]  0.5 0.4  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  0.8  Volume [m3/day]  0.6  Jul-07  b) Sub-Lysimeter C  0.6 0.4 0.2 0 5-Jan  19-Jan  2-Feb  16-Feb  1-Mar  Upper Bound Expected Lower Bound  0.3 0.2 0.1 0 Jan-07  Apr-07  Jul-07  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  Figure 3.11. Upper, lower and expected flow for basal lysimeter D and sub-lysimeter C.  105  L1TDR5  L2TDR5  L3TDR5  30  L4TDR5  2.1 m 0.25  Precipitation [mm/day]  Volumetric Water Content [m 3/m3]  0.3  2.4 m 0.2  20  1.6 m  0.15 1.8 m  0.1  10  0.05 0 Jul-08  Aug-08  Sep-08  Oct-08  Nov-08  Dec-08  Jan-09  0 Feb-09  0.4  L4TDR1  L4TDR2  L4TDR4  40  L4TDR5 2.1 m  0.3  30  9.3 m 3.7 m 20  0.2 7.3 m  10  0.1  0 Jul-08  Aug-08  Sep-08  Oct-08  Nov-08  Dec-08  *L4TDR3 (5.4 m depth) not functioning  Figure 3.13. Line 4 TDR moisture content response (under front batter).  106  Jan-09  Precipitation [mm/day]  3  3  Volumetric Water Content [m /m ]  Figure 3.12. Shallow TDR moisture content and precipitation in mm/day.  0 Feb-09  L1TDR1  L1TDR2  L1TDR3  30  L1TDR5 2.4 m  0.25  7.3 m 9.3 m 20  0.2 5.8 m 0.15  10  0.1 0.05 0 Jul-08  Aug-08  Sep-08  Oct-08  Nov-08  Dec-08  Jan-09  Precipitation [mm/day]  Volumetric Water Content [m 3/m3]  0.3  0 Feb-09  *L1TDR4 does not function  0.4  L2TDR1  L2TDR2  L2TDR3  L2TDR4  L2TDR5  7.9 m  0.3  40  30  6.5 m 0.2  20 4.9 m  0.1  3.2 m 1.6  10 1.6 m 3.2  0 Jul-08  Aug-08  Sep-08  Oct-08  Nov-08  Figure 3.15. Line 2 TDR moisture content and precipitation.  107  Dec-08  Jan-09  Precipitation [mm/day]  Volumetric Water Content [m 3/m3]  Figure 3.14. Line 1 TDR moisture content and precipitation.  0 Feb-09  4  RESULTS: TEST PILE GEOCHEMISTRY  It is often the case that sulphide minerals found in mining environments will lead to the creation of Acid Rock Drainage (ARD). Pyrite is a common sulphide mineral associated with ARD, the chemical oxidation of which generates acidic waters according to the following pathway (Blowes et al., 2007):  𝐹𝐹𝐹𝐹𝑆𝑆2(𝑠𝑠) +  15 2  7  𝑂𝑂2 + 2 𝐻𝐻2 𝑂𝑂 → 𝐹𝐹𝐹𝐹(𝑂𝑂𝑂𝑂)3 + 2𝑆𝑆𝑂𝑂42− + 4𝐻𝐻 +  [4-1]  In some cases, acid generated from sulphide oxidation reacts with the gangue material. If the gangue material includes a carbonate mineral, it can buffer the acidity and create Neutral Rock Drainage (NRD): 𝐶𝐶𝐶𝐶𝐶𝐶𝑂𝑂3(𝑠𝑠) + 𝐻𝐻 +(𝑎𝑎𝑎𝑎 ) → 𝐶𝐶𝐶𝐶2+(𝑎𝑎𝑎𝑎 ) + 𝐻𝐻𝐻𝐻𝐻𝐻3−(𝑎𝑎𝑎𝑎 )  [4-2]  𝐻𝐻2 𝐶𝐶𝐶𝐶3(𝑎𝑎𝑎𝑎 ) → 𝐶𝐶𝐶𝐶2(𝑔𝑔) ↑ + 𝐻𝐻2 𝑂𝑂  [4-4]  𝐻𝐻𝐻𝐻𝐻𝐻3−(𝑎𝑎𝑎𝑎 ) + 𝐻𝐻 +(𝑎𝑎𝑎𝑎 ) → 𝐻𝐻2 𝐶𝐶𝐶𝐶3(𝑎𝑎𝑎𝑎 )  [4-3]  Previous analyses have shown that there is a considerable buffering capacity in the Antamina host rock (Golder 2007a, Golder 2007b; Aranda et al., 2009) whereby acid generated from sulphide mineral oxidation is rapidly neutralized by carbonate mineral dissolution. Recall that nearly 2.2 % of the pile consists of sulphide minerals (mainly pyrite) and that 48 % of the entire pile consists of carbonate minerals (Table 2.3). As a result of these buffering reactions, effluent chemistry is pH-neutral, and most metals are immobile under these conditions. However, select metals such as zinc remain mobile under circum-neutral conditions, while the mobility of metals present as oxyanions, such as arsenic (as As(V)), increases. These conditions differ considerably from the majority of mine sites within Canada, which contain mostly acid-generating waste rock. Sulphide oxidation and acid neutralization involve several factors that contribute to the complexity of understanding and predicting acid and neutral rock drainage.  This chapter evaluates the chemical evolution of the neutral-pH drainage pore and outflow waters of test pile 1 between 18-Jan-2007 to 30-Sept-2008 by: 108  1. characterizing the geochemical composition of porewater and outflow waters, 2. examining the spatial and temporal variability of solute concentrations, 3. assessing the correlation between flow path length and concentration, 4. determining metal release mechanisms and possible attenuation mechanisms, particularly for mobile metals such as zinc, 5. providing supporting evidence for observations from hydrology (e.g., matrix vs. macropore flow), and 6. comparing effluent chemistry between test pile 1 and field cell experiments.  A comprehensive instrumentation network of real-time, continuously logged data, point-source chemistry data from within the pile, and effluent chemistry data are employed to better understand the internal processes and chemical evolution of test pile 1. Approximately 270 water samples were obtained from the sub-lysimeters, basal lysimeter, composite tank and sump and an additional 540 samples were obtained from the soil water samplers.  4.1 4.1.1  ELECTRICAL CONDUCTIVITY Lysimeters  Continually-logged electrical conductivity (EC) data from the four lysimeters is converted to specific conductance (SC) as described in Chapter 2 (Section 2.10.1.6). The EC probes used in the pile are functional, but experience degradation with time and result in several periods of missing data, particularly for the discharging water from basal lysimeter D. The highest SC values of 6000 µS/cm are reported at the beginning of the study period, which likely coincide with the initial flushing of weathering products of the most reactive material. Near the end of the study period, SC values typically range between 1000 and 3000 µS/cm. Basal lysimeter D effluent ranges between 1500 and 3500 µS/cm and is representative of the overall pile specific conductivity.  The high-resolution SC data (obtained at 30-minute intervals) indicates responses to large precipitation events (i.e. low SC values with high precipitation values), but does not register 109  smaller-scale, rapid fluctuations derived from specific wetting front arrivals via different flow paths (Figure 4.1). The best evidence of rapidly changing chemistry, in response to precipitation, is observed in lysimeter D and the front sub-lysimeter C, where sharp increases in flow result in sharp decreases in SC e.g., in March 2007 (Figure 4.1). This behaviour indicates the activation of macropore flow under heavy precipitation events (as described in the previous chapter) that results in higher discharge rates and a lower dissolved solute load. A similar response is anticipated in 2008 but is not as evident in the data due to incomplete EC-probe data.  Pile batters represent nearly 75 % of the horizontal surface area of the pile and will play an important role on pile response. Near the end of the first week in March-2007, the SC from basal lysimeter D drops from 3500 to 1700 µS/cm in less than 10 days, whereas sub-lysimeter C drops from 2600 to less than 2000 µS/cm over nearly the same period. Sub-lysimeters A and B do not show sharp responses, though slight dips observed on 22-Mar and 16-Mar (respectively) correspond to higher pile discharge rates.  Sub-lysimeter A (centre of the pile) reports the highest SC levels which are nearly twice as high as those reported at sub-lysimeter C (front batter of pile). Basal lysimeter D is representative of the entire pile, and reports intermediate levels with the largest seasonal variation (i.e., similar SC levels to A and B during the dry season and similar SC values to C during the wet season). Recall that the highest flow of water is reported from the batters and that a significant amount of the total pile flow occurs during the wet season (Chapter 3). This indicates that more macropore flow is mainly prevalent during the wet season, which is reflected in lower SC values compared to the dry season.  The weekly, manually-obtained field data is also plotted for comparison (Figure 4.1). Since SC values do not undergo rapid changes over the course of a few hours, this provides evidence that a significant portion of the effluent water originates from matrix flow. As matrix water is usually slow and diffuse (e.g., Chapter 1, 3), it is typically associated with higher dissolved loads compared to characteristic macropore flow. Comparing results from sub-lysimeters A and B (matrix dominated) with sub-lysimeter C and basal lysimeter D (stronger contributions from macropore flow) confirm this hypothesis. Recall that that the 4 x 4 m sub-lysimeters are 110  conceptualized as representing effluent water that spatially averages over matrix and macropore (if activated) flow pathways. The same applies to the 36 x 36 m basal lysimeter flow and chemistry, which averages over various flow and chemistry differences that result from the pile batter and the pile interior.  4.1.2  Composite Tank  The composite tank chemistry follows similar seasonal trends to the lysimeters, with higher reported values during the dry season (Figure 4.1). A regression comparison of SC between the composite tank E and the basal lysimeter D indicates that there is a reasonable correlation (Figure 4.2a). Average wet and dry season SC values for E are compared to those of D and indicate that average composite tank SC values are 5 % higher during the wet season but nearly 4 % lower during the dry season (Table 4.1). Further details about the composite tank are discussed in Appendix G.  4.1.3  Pile Interior  Fifteen (15) soil water samplers (SWS) were installed in instrumentation lines 1, 2 and 4 to provide chemistry data at various vertical points throughout the pile. Five soil water samplers (SWS) were placed at depth in near-vertical intervals of 1.5 m, and Figure 4.3 indicates that concentration changes with depth, similar to findings presented by Wagner (2001). Recall from Chapter 2 that test pile 1 was constructed by end-dumping waste rock in three stages, with Tipping Phase I at the back of the pile, Tipping Phase II in the middle and Tipping Phase III at the front of the pile (Figure 4.4).  The largest variability and highest SC values in response to the initial flushing of weathered products occurs in Tipping Phase II, in which instrumentation line 2 and sub-lysimeter A are located (Figure 4.4). SC values reported at sub-lysimeter A reach 6000 µS/cm at the beginning of the study period and drop below 3000 µS/cm towards the end (Figure 4.3). The SC values in line 2 range between 2000 and 3500 µS/cm at 1.9 m below the surface, and between 2000 and 4500 µS/cm 8.9 m below the surface. (Note, some data points report up to nearly 8000 µS/cm but are believed to be the result of improper data recording as they are one order of magnitude higher than the samples from the same location around that time and will be excluded from the general 111  discussion). SC values indicate slight seasonal variability, but the initial flushing of weathered products makes this pattern harder to resolve.  Conversely, the lowest SC values are reported in Tipping Phase III, in which sub-lysimeter C and instrumentation line 4 are located (Figure 4.4). SC values in sub-lysimeter C range between 1200 and 2600 µS/cm, with a strong seasonal signal that is rather consistent throughout the study period (i.e., reaches same general values during same time in 2007 and 2008) that is not observed towards the pile interior. There do not appear to be any strong differences in SC values at depth, with most values ranging between 500 and 2000 µS/cm. Since the instrumentation and sublysimeter in this region both have the shortest flow path and highest flow volumes compared to the other two sub-lysimeters (Chapter 3), it consequently has shorter porewater residence times as supported by evidence of a lower dissolved load.  4.1.3.1 Changes at Depth In general, SC values increase with depth in lines 1 and 2 as the infiltrating rainwater acquires a higher dissolved load along the flow path. Since average SC values at each SWS port in lines 1 and 2 generally increase with depth, this possibly indicates that no significant solubility controls are present at this time and that the pile is not tall enough to express these controls (i.e., the average specific conductivity may eventually flatten out if the pile was 100 m high, likely due to the presence of solubility controls). SC values in line 1 generally report higher values than the corresponding sub-lysimeter B during the first wet season, but report similar if slightly lower values a few months into the study period. Conversely, SC values in line 2 are almost consistently lower than SC values reported in the corresponding sub-lysimeter A. All SC values in line 4 are lower than those reported in the corresponding sub-lysimeter C. However, since SC values are relatively constant with depth and show similar values at the same time throughout the year, this possibly indicates that this region is subject to different chemical controls compared to lines 1 and 2 based on the fact that this material is closest to the surface and reasonably well aerated. It is also possible that these differences are due to preferential flow or heterogeneities in composition. These themes are discussed later in this discussion.  112  Each sub-lysimeter lies directly under one or two SWS’s, typically the middle ones, indicating that the highest and lowest SWS’s will not reflect similar concentrations to the corresponding sub-lysimeter. This may be further complicated by non-vertical flow effects discussed in the previous chapter. Further discrepancies between the SWS and lysimeter chemistries may result from the fact that the SWS’s extract a sample from the matrix (higher dissolved load) whereas the sub- and basal lysimeters collect the total flow of water that averages over matrix and macropore flow (lower dissolved load). The SWS’s extract point measurements and each sampler has a maximum capacity of only 500 ml that is extracted from an estimated region of influence that is up to 10 - 15 cm in diameter. In comparison, the 4 x 4 m sub-lysimeters have a much larger region of influence and integrates over a larger area.  4.2  TEMPERATURE  Thirty-six (36) thermistor probes were installed at various locations within the pile and instrumentation hut (Chapter 2; Appendix A). Temperature data provided by thermistors is principally used to assess whether heat is emitted during sulphide mineral oxidation and to estimate the extent of temperature-driven gas convection. However, several secondary reasons exist, namely for: geochemical speciation modeling and other calculations (i.e., mineral solubility), neutralization capacity (Sherlock et al., 1995), and to affect temperature corrections for TDR and EC. Temperature data also contributes in explaining flow anomalies observed in the sub-lysimeters during the cooler dry season months. The mean annual temperature (M.A.T.) ranges between 5.5 and 6 °C during the study period (18-Jan-2007 to 30-Sept-2008) and is 5.6 °C during the water year (1-Oct-2007 to 30-Sept-2008).  4.2.1  Lysimeters  In the instrumentation hut, a thermistor probe is installed in each of the four water conveyance systems from each lysimeter (Figure 2.9). Since the conveyance pipes are partially exposed outside the instrumentation hut, lysimeter outflow temperature ranges (-0.6 to +23 °C) are more pronounced than temperatures from inside the pile (see below) due to meteorological diurnal fluctuations. These observations do not provide any indication of the weathering behaviour but help to explain some anomalies in the outflow behaviour.  113  Water draining from the three sub-lysimeters exhibits very similar diurnal temperature responses, with temperatures generally ranging from approximately 3 - 14 °C during the wet season months and 2 - 18 °C during the dry season months (although temperatures extend beyond these ranges, resulting in short-lived periods in which temperatures drop to below zero and near 20 °C). Higher flow rates during the wet season correspond to a tighter temperature range. Throughout most of the year (although more pronounced in the wet season), the temperature range of sublysimeter C is slightly more constrained than that of sub-lysimeters B and A (by up to 2 - 3 °C). This pattern is even further evident in basal lysimeter D flow data, which experiences a significant difference in temperature ranges throughout the year. In the dry months, temperatures range from as much as -0.5 to 23.0 °C and as little as 6 - 8 °C in the 2007 wet months.  Temperature variations in the outflow water do not have any implications to weathering characteristics inside the pile, but do help to explain an anomaly in the flow behaviour. During the cooler low-flow periods of the dry season, the large diurnal temperature range is found to effect sub-lysimeter flow and create abnormal diurnal ‘pulsing’ of rapid water flow during the day and no flow at night. It is hypothesized that this is caused by the freezing of water in the exposed conveyance pipes. To prevent this from occurring, an insulating material was placed around the exposed pipes in November-2007. Consequently, the 2008 wet season temperature fluctuations for basal lysimeter D converge to a constricted range as tight as 6 to 6.5 °C from 2 to 15 °C at the end of the 2008-09 dry season. The range of temperatures decreases by up to 4 °C (in total) for the sub-lysimeters, and at no time does sub-lysimeter drop to freezing temperatures (min ~ 1 °C except for 11-Sept-2008 which registered 0.2 °C). To note, the range in sublysimeter C is more constrained by up to 4 °C compared to the other two sub-lysimeters. Despite these differences, the diurnal pulsing is still observed in the sub-lysimeter effluent during the next low-flow period (i.e., 2008 wet season). Based on the minimal volume of water associated with this behaviour, it is insignificant compared to the overall pile outflow and is not discussed further.  4.2.2  Pile Interior  Seasonal and spatial temperature variability is evident throughout the entire study period in test pile 1 (Figure 4.5). Temperature trends towards the pile interior indicate that while the 114  temperature regime is cyclical (i.e., seasonal variation), temperatures are all 0.5 – 1 °C cooler in 2008 compared to 2007. This may be explained by a small decrease in reactivity later in the pile lifetime as the highly reactive particles react first (releasing energy as a by-product of sulphide mineral oxidation). This may also indicate that the pile may not have reached dynamic thermal equilibrium soon after construction was completed.  4.2.2.1 Internal vs. External Temperatures Based on 30-minute data, internal pile temperatures range between 0.3 and 12 °C, which is much more constrained than the outflow temperature ranges, indicating the thermal insulating properties of the waste rock (Figure 4.5). Thermistors along the outer shell of test pile 1 respond to short-term meteoric changes whereas temperature probes located in the interior of the pile show more attenuated and subdued temperature variations and respond to long-term meteoric trends (Figure 4.5).  During the water year (1-Oct-2007 to 30-Sept-2008; Chapter 3), temperatures measured from the ‘deep probes’ (> 6 m from the top of the pile in lines 1 - 4, > 15 m from the outer perimeter for lines 5 and 6) in test pile 1 range between 4.5 and 7.5 °C, with a mean temperature of 5.8 °C (Table 4.2). The subdued temperature variations occur because of the insulating properties of waste rock which attenuate the short-term atmospheric signals. The ‘shallow probes’ (< 6 m from the top of the pile in lines 1 - 4, > 15 m from the outer perimeter in lines 5 and 6) exhibit a larger temperature range of 3 - 10 °C with a mean annual temperature of 7.6 °C, during the water year (Table 4.2). The mean annual air temperature (M.A.T.) during the water year is 5.6 °C.  A comparison of mean annual temperature calculated based on daily thermistor data in lines 2 and 3 during the water year (1-Oct-2007 to 30-Sept-2008) is summarized in Table 4.2. During this period, average temperatures near the top of the pile (i.e., shallowest probes) are 1.7 - 1.8 °C higher than the bottom. During the wet season (1-Oct-2007 to 30-Apr-2008) the temperature range is nearly the same (1.7 °C), but the difference slightly diverges during the dry season (2.0 °C). The M.A.T. is very stable during the wet and dry season (Table 4.2), and is very similar to the bottom pile temperatures (± 0.2 °C). This temperature difference (warm surface - cool interior) is the inverse of patterns observed or modeled in other more mature and reactive waste 115  rock piles, and may have implications on gas transport (Sracek et al., 2004; Andrina et al., 2006; Linklater et al., 2005; Marcoline 2008). It is hypothesized that the highest reported temperatures exist at the top of the pile because:  1. this material is the most reactive (i.e., finer grained due to compaction of heavy machinery and preferential settling of finer material near the top of the pile during end-dumping process) and generates the most heat during exothermic sulphide mineral (pyrite) oxidation, 2. heat-driven gas convection transports heat through the relatively well-aerated pile, and/or 3. solar radiation heats the top of the pile.  The latter item is most probable, but more comprehensive data and thermal modeling is required to form any solid conclusions.  4.2.2.2 Spatial Variability in Temperatures During the water year the average temperature in instrumentation lines 1, 4, 5 and 6 ranges between 5.5 and 5.9 °C, whereas instrumentation lines 2 and 3 exhibit a mean annual temperature of 6.7 °C. That the mean annual temperature in both lines 2 and 3 is more than one degree higher than the other instrumentation lines may be indicative of higher chemical oxidation rates in the material from Tipping Phase II in which lines 2 and 3 are situated (Figure 4.4). Figure 4.6 presents a cross-sectional profile of average temperatures in lines 1, 2 and 4 during the water year and indicates that temperatures are coolest near the bottom and increase with elevation in the pile.  4.2.2.3 Evidence of Advection and/or Evaporation from Front Batter Due to its proximity to the surface, each thermistor in line 4 displays fluctuations in response to atmospheric temperature changes and indicates that the waste rock material in the batter will likely undergo stronger temperature fluctuations compared to internal thermistors found in lines 1, 2 and 3. Line 4 thermistors located near the base register cooler and more variable temperatures than thermistors located closer to the top of the pile, which is the opposite trend 116  observed in the rest of the pile. This likely indicates that when a portion of the overburden above line 4 was removed to conform with initial design specifications, it created a situation where the deepest probes (7 and 8.9 m below surface) became situated closest to the surface at the pile toe (Figure 4.6). That average temperatures at 1.9 and 3.6 m below the surface in line 4 (4.5 and 4.8 °C, respectively) are below the mean annual temperature by one degree likely indicates that this region of the pile is well-aerated (e.g., wind-driven advection) and that energy is likely lost via evaporation (i.e., latent heat of vapourization) and leads to cooler average temperatures. While these differences may result from precision and calibration issues, based on performance of all of the thermistors, a one-degree difference is considered to be indicative of a significant temperature trend within the pile.  4.3  PH  In general, pH values range between 7 and 8.5, with the majority of the values between 7.5 and 8. Sampling station (lysimeters, composite tank) and interior (SWS) pH values are similar, and both exhibit a seasonal variation in response to precipitation. Based on equations 4-2, 4-3 and 44, higher pH values typically indicate that porewaters approach more open-system conditions (i.e., atmospheric CO2 near surface) whereas lower pH values approach more closed-system conditions (i.e., depleted CO2 towards pile interior). 4.3.1  Lysimeters  A seasonal variation in pH is apparent, with decreasing values during the dry season and increasing pH values during the wet season. There are several explanations for these trends:  1. the rate of sulphide mineral oxidation is slightly higher than calcite dissolution during the wet season, which results in lower pH values, 2. higher infiltration and discharge rates during the wet season result in lower porewater residence times which means that they have less time to equilibrate with calcite before exiting from the bottom of the pile, and/or 3. preferential and/or macropore flow are more prominent in the wet season and have less time to equilibrate with calcite before exiting from the bottom of the pile.  117  It is also apparent that pH values are nearly 0.5 pH points higher in 2008 than the same period in 2007, likely indicating that the pile is slightly less reactive during 2008. Due to the seasonality of the data, it is difficult to make any conclusions of pH trends based on this data at this time.  4.3.2  Composite Tank  Composite tank E pH values follow similar seasonal trends to the lysimeters. A regression analysis of the weekly and/or bi-weekly pH data between E and basal lysimeter D indicates that there is a relatively weak correlation (Figure 4.2b), which is largely due to the sensitivity of the parameter and the instruments required to measure it (Table 4.1). A relatively close range is observed between the two stations, as average pH values are 4 % higher in the composite tank during the wet season, and 7 % higher during the dry season. Basal lysimeter D pH values are lower during the dry season than during the wet season, and results in larger pH differences between D and E during the dry season (pH difference: ~ 0.5) compared to the wet season (pH difference: ~0.3).  Water collected in the composite tank generally has a higher pH than water collected directly from the lysimeter discharge. This is because water draining from the pile interior likely has a high dissolved CO2 concentration that exceeds atmospheric levels and releases CO2 gas as it attempts to equilibrate with atmospheric CO2 concentrations in the composite tank (and when passing through the flow splitter), which consequently raises the pH according to equation 4-4. A poor correlation between E and D may be the result of the fact that E includes a mixture of water from all four lysimeters.  4.3.3  Sump  Based on two samples (28-Nov-2007, 1-May-2008) an average pH of 7.5 was reported for the sump. These levels are within the lower range of pH values observed throughout the pile (Figure 4.7), possibly indicating that sump porewaters are approaching closed-system behaviour. This is expected based on the sump and 2B-Protective Layer design.  118  4.3.4  Pile Interior  Soil water sampler data indicates that pH remains relatively stable in lines 1 and 2 and increases with depth in line 4 (Figure 4.7). Line 1 and 2 pH values are generally the lowest reported pH in the pile, and likely indicates that these regions are representative of partially-closed systems within the pile. In general, pH reported in all 3 instrumentation lines (1, 2 and 4) is within a similar range to that reported in the sub-lysimeters, and basal lysimeter D is representative of the variability expressed in the three sub-lysimeters. At the lowest depth in line 4 (i.e., 8.9 m below the surface), pH is 0.5 units higher than the reported pH of the SWS above it (i.e., at 7 m depth). Recall that this SWS is closest to the pile toe and is likely covered by the least amount of material (Figure 4.7). While the higher pH may be due to a lesser amount of sulphide mineral oxidation occurring in this region compared to the rest of the pile, it is possible that the higher pH is due to the degassing of CO2 in this region in which wind-driven advection is responsible for keeping it well-aerated.  Preliminary gas analyses indicate that CO2 concentrations (in the gas phase) reach as high as 5 times atmospheric toward the pile interior, but are only 10 % higher than atmospheric in the well-aerated line 4. In the liquid phase, dissolved porewater CO2 content will equilibrate with the CO2 concentration in the pore gas. As water from the interior of the pile (elevated CO2 partial pressure) moves towards more aerated regions (lower CO2 partial pressure), it equilibrates with atmospheric CO2 and leads to degassing (equation 4-4). A PHREEQC analysis indicates that an approximate 0.4 unit increase in pH will result during the transport of porewaters from internal to external locations (i.e., moving from closed- to open-systems) while maintaining equilibrium with calcite. This indicates that pH changes may result from sulphide oxidation (which produces H+ ions, and decreases pH. A reduction in pH is limited by calcite dissolution, which in turn raises CO2 – under closed system conditions) and/or CO2 degassing (which removes CO2 and increases pH – under open system conditions).  4.4  SOLUTES  This section evaluates the porewater and effluent water solute concentrations. All solute concentrations exhibit seasonal variability, which increase during the dry season and decrease during the wet season (which is opposite to pH). Recall that 250 ml of a sample is required to 119  analyze the dissolved metals, 500 ml is required for total metals, and 1000 ml is required for the remaining solutes. Each SWS has a capacity of only 500 ml (Chapter 2).  For the interest of this investigation, Antamina’s discharging and receiving limits can be found in Table 4.3. The discharging limits refer to the solute concentrations that are allowed to leave the mine (and be mixed with re-directed river discharge) and the receiving limits are the maximum concentrations allowed at sampling stations downstream from the mine.  4.4.1  Metals  Dissolved arsenic (As), copper (Cu), molybdenum (Mo), iron (Fe), lead (Pb), and nickel (Ni) concentrations are at, or near, detection limits (i.e., < 0.05 mg/l). As is primarily released from arsenopyrite, enargite, realgar, tennantite and watanabeite; Cu is associated with chalcopyrite and iron oxyhdyroxide, iron sulphate and apatite; Mo is largely from molybdenite but also from some iron sulphates; Pb is mainly associated with galena, but also Fe-oxyhydroxide, chalcopyrite and Fe,Cu-silicates (Aranda, 2009). The only metal above detection limits in this pile is dissolved zinc (Zn), although total Cu and Fe are elevated in the sump (consistent with slightly lower pH levels in the sump).  4.4.1.1 Zinc Sphalerite is the principal source of Zn in this test pile, but other sources include Feoxyhydroxide, Fe-sulphate and apatite (Aranda, 2009).  4.4.1.1.1 Lysimeters, Composite Tank and Sump Dissolved Zn concentrations in the lysimeter drainage range between 0.5 and 2.5 mg/L (Figure 4.8). There is a clear seasonal variation, with concentrations increasing during the dry season and decreasing during the wet season. The lowest Zn concentrations occur at the end of the wet season and the highest concentrations occur at the end of the dry season. Figure 4.8 illustrates that Zn concentrations are highest at sub-lysimeter B (1-2.5 mg/L), and lowest in sub-lysimeter C (0.5 to 1 mg/L). Concentrations reported at D are typically between these two values except for the end of the 2008 dry season where values increase sharply to match B values.  120  Zinc concentrations in composite tank E exhibit similar seasonal fluctuations as observed in the lysimeters, with higher concentrations during the dry season. A regression comparison indicates that there is a strong correlation between E and basal lysimeter D, with slightly lower Zn concentrations in the composite tank (Figure 4.2c). Average Zn concentrations in the composite tank are roughly equal to basal lysimeter D concentrations during the wet season, but are nearly 15 % less during the dry season (Table 4.1).  Sump (F) Zn concentrations are similar to those observed in D and E. The sump produces the highest total Fe and copper Cu concentrations reported throughout the entire pile. Dissolved Fe and Cu levels are largely at or near detection limits (Figure 4.9). It is hypothesized that the area around the sump is saturated (Chapter 3) which would create relatively anoxic conditions compared to the rest of the pile thus explaining these high total Fe and Cu concentrations.  4.4.1.1.2 Pile Interior Zinc concentrations (< 2.5 mg/L) in lines 1 and 2 increase with depth (Figure 4.8) and remain below their respective sub-lysimeter (B and A, respectively) concentrations (see Section 4.1.3 for explanation). The opposite is true for line 4 and sub-lysimeter C. The trend in line 4 is more difficult to resolve, with values reaching as high as 5.5 mg/L at the shallowest soil water sampler at the beginning of the wet season. The highest reported values (> 3 mg/L) occur over a few weeks only and may be the result of:  1. analytical errors, 2. Zn attenuation mechanisms present throughout the rest of the pile may not be present here, 3. release of Zn from a high Zn-bearing region, 4. higher oxidation rates in this region, and/or 5. evaporation of porewaters, resulting in higher relative Zn concentrations during the dry season.  These differences highlight the highly heterogeneous nature of the pile, and that the highest Zn concentration in the effluent waters (2.5 mg/L) may likely increase in time. With the limited data 121  available for line 4, no definitive conclusions can be made with regards to Zn. Longer term monitoring is required before any conclusions can be drawn.  4.4.2  Cations  Dissolved calcium (Ca), sodium (Na), magnesium (Mg), and potassium (K) all have large concentration ranges, with maximum values of 1200 mg/L, 110 mg/L, 75 mg/L, and 60 mg/L, (respectively). The abundance of Ca indicates that calcite dissolution is significant; however, the presence of Na and K indicates that weathering of the waste rock is not limited to carbonate mineral dissolution as there is a considerable amount of silicate minerals in the pile (e.g., pyroxene, biotite, plagioclase, K-felspar; Aranda, 2009). Ca is also present in Ca-Mg-silicates silicates (diopside) and the calcium silicate minerals of hornfels (Aranda, 2009).  4.4.2.1 Lysimeters, Composite Tank and Sump The highest cation (Ca, Na, Mg, and K) concentrations are reported in sub-lysimeter A and the lowest concentrations are reported in sub-lysimeter C (indicating chemical differences between pile batter and the interior), and concentrations reported at sub-lysimeter B and basal lysimeter D are typically found in between (Figure 4.10, Figure 4.11, Figure 4.12, Figure 4.13). While seasonal variability is reported in each lysimeter, cation concentrations generally appear to be decreasing over time. This is particularly the case for Na, Mg and K, which indicates that the minerals containing these elements are not weathering as rapidly (silicate minerals generally weather more slowly than carbonate minerals).  Calcium concentrations exhibit a strong seasonal variation with increasing concentrations during the dry season and decreasing concentrations during the wet season, which is opposite to pH. These differences are likely due to the fact that, in equilibrium, an increase in H+ concentrations (decreasing pH) will result in an increase in Ca concentrations (equation 4-2). Although calcite solubility is affected by both temperature and equilibrium pH, temperature changes between the wet and dry season are not enough to affect these changes and that the likely driver of these seasonal differences is the concentration of Ca. A higher concentration of Ca that results from a higher porewater residence time during the dry season (which increases the dissolved load) leads  122  to decreasing pH values, and vice versa (equation 4-2). A slightly lower pH will thus enhance calcite solubility.  Cation concentrations in the composite tank E exhibit similar seasonal fluctuations as observed in the lysimeters, with higher concentrations during the dry season. A regression comparison indicates that there is a strong correlation between Ca measured at E and basal lysimeter D (Figure 4.2d). Average cation concentrations in E are slightly higher (< 4 %) than D concentrations during the wet season, but exhibit a larger difference during the dry season. For example, the average Ca concentration at E is 5 % less than D, whereas Na is nearly 9 % higher and Mg is 4 % lower during the dry season (Table 4.1).  Calcium concentrations in the sump (F) are slightly higher than concentrations observed in basal lysimeter D and composite tank E, which is supported by a lower pH in the sump (Figure 4.9).  4.4.2.2 Pile Interior Cation concentrations measured in the soil water samplers in line 2 report the highest concentrations found throughout the pile during the initial pile flushing in 2007. For example, Figure 4.11 indicates that Na concentrations along Line 2 and measured in sub-lysimeter A effluent are much higher in 2007 (55-110 mg/L) compared to 2008 (25 -55 mg/L). A depth trend for these metals cannot be discerned, as line 2 concentrations are much higher than any other concentrations reported at other pile locations. The soil water samplers also indicate a seasonal trend (Figure 4.10, Figure 4.11, Figure 4.12, Figure 4.13).  4.4.3  Trace Metals  Antimony (Sb) concentrations are less than 0.25 mg/L. Likely sources are stibnite and watanabeite (Aranda, 2009). However, the SWS’s in line 1 experience higher than average concentrations in the suction lysimeters located at 7 and 8.9 m depth during the wet season. Interestingly, comparable values are not reported in sub-lysimeter B effluent. Similarly, manganese (Mn) concentrations reported in the SWS’s of line 2 are nearly twice as high as its corresponding sub-lysimeter A. Overall, the mobility of trace metals other than Zn from Pile 1 is low. 123  4.4.4  Major Ions  Major ions such as alkalinity, sulphate, and nitrate are only analyzed if a 1000 ml sample is obtained. This was not usually a problem for the sampling stations (lysimeters, composite tank and sump), but was not possible for the internal pile data as the SWS’s are not capable of producing water samples of this size (maximum capacity of ~ 500 ml). Therefore, only major ions are only analyzed for the lysimeters, composite tank and sump.  4.4.4.1 Dissolved Sulphate Dissolved sulphate is an important solute to monitor because it indicates that the chemical oxidation of sulphide minerals such as pyrite (equation 4-1), chalcopyrite, pyrrhotite and sphalerite (to a lesser extent) is occurring. Concentrations range between 500 and 2000 mg/L and exhibit seasonal trends similar to other solutes with maximum values near the onset of the wet season and lowest values near the onset of the dry season (Figure 4.14). The highest concentrations are found in A and the lowest concentrations are found in C. Basal lysimeter D exhibits high concentrations matching A at the onset of the wet season, and matches the low C concentrations at the onset of the dry season.  Sulphate concentrations in the composite tank E exhibit similar seasonal fluctuations as observed in the lysimeters, with higher concentrations during the dry season (Figure 4.14). Sulphate concentrations measured in the basal lysimeter are strongly correlated to (and slightly higher than) those of the composite tank (Figure 4.2e). Average SO4 concentrations in the composite tank are roughly equal (2 % higher) to basal lysimeter (D) concentrations during the wet season, but are nearly 9 % lower during the dry season (Table 4.1).  Sulphate concentrations measured in the sump (F) are 500 mg/L higher than the composite tank on 28-Nov-2007 (1675 mg/L) and 100 mg/L higher on 1-May-2008 (1030 mg/L). This indicates that the porewater in the 2B-Protective Layer experiences longer residence times, which is expected based on the location of the sump.  124  4.4.4.2 Alkalinity Under the neutral-pH conditions of test pile 1, it can be assumed that almost all of the alkalinity is in the form of bicarbonate (HCO3-). Similar to Ca (and opposite to pH), alkalinity experiences seasonal variations in which alkalinity values increase during the dry season and decrease during the wet season (equation 4-2). These trends are also explained by a higher residence time during the dry season, which drives calcite dissolution (increasing Ca and HCO3-) and decreases pH (Section 4.4.2.1).  The highest alkalinity values are reported at the end of the observation period (100 mg/L), whereas the lowest values are observed at the start of the period (60 mg/L; Figure 4.15). Alkalinity values from peak 2008 dry season values are 20 % higher than those values from the peak 2007 dry season, although this is likely not indicative of any long-term trends because of the initial flushing of weathered products from the most reactive material. However, if alkalinity in the pile was decreasing over time, this would likely indicate that pH values will be increasing over this period. That HCO3- tends to be increasing while Ca concentrations tend to decrease over the study period likely indicates that there is a solubility control of Ca, most likely the precipitation of calcite (equation 4-2).  Composite tank E exhibits similar alkalinity values to those of the lysimeters, with higher concentrations during the dry season (Figure 4.15). Average alkalinity concentrations in the composite tank are nearly 3 % lower than basal lysimeter D concentrations during the wet season, and nearly 9 % lower during the dry season (Table 4.1).  Alkalinity concentrations measured in the sump are 9 mg/L higher than the composite tank on 28-Nov-2007 (83 mg/L) and 3 mg/L higher on 1-May-2008 (79 mg/L).  4.4.4.3 Nitrate Nitrate is not present in the host rock, but is found in high initial concentrations because it is used in the explosives that the mine uses in blasting. Early time pile concentrations nearly reach 300 mg/L, but drop to 100 mg/L at the end of the observation period. Basal lysimeter D values begin at 120 mg/L but quickly drop to nearly 20 mg/L, indicating that the nitrates are largely 125  being flushed out. That nitrates are still present in the pile indicates that some porewater is likely retained as residual saturation in the matrix (i.e., relatively immobile).  Nitrate concentrations measured in the sump are 22 mg/L higher than the composite tank on 28Nov-2007 (63 mg/L) and 1 mg/L higher on 1-May-2008 (7 mg/L).  4.5  PORE GASES  Pore gas analyses provide evidence that sulphide mineral oxidation is occurring based on the pathway described in equation 4-1 whereby oxygen is consumed, carbon dioxide is enriched, sulphate ions are produced and acidity is generated. Results indicate that the outer shell of the pile is reasonably well-aerated. Oxygen concentrations decrease with depth (towards pile interior), with the lowest oxygen concentrations at 85 % atmospheric levels. Conversely, carbon dioxide concentrations increase with depth (towards pile interior), with the most enriched samples containing up to 5 times atmospheric concentrations.  Gas data generally confirm observations described above (e.g., temperature, pH, Ca, alkalinity, etc). The interior of the pile behaves more like a partially-closed system in which higher CO2 gas concentrations result in higher dissolved CO2 gases in the porewaters and lower pH values (equations 4-3 and 4-4). The more-aerated pile exterior acts like an open system where CO2 partial pressures are lower in the pore fluids (gas and water) and results in a higher pH and possible CO2 degassing (equation 4-4). Wind-driven gas advection is likely the dominant process in these pile regions. The areas with the highest CO2 and lowest O2 gas concentrations coincide with instrumentation lines 2 and 3, which exhibit the highest average annual temperatures (Section 4.2.2). Since pyrite oxidation is exothermic, this provides reasonable evidence that significant sulphide mineral (pyrite) oxidation is taking place within test pile 1.  The highest temperatures recorded in lines 2 and 3 are near the top of the pile whereas the highest CO2/lowest O2 concentrations are at the bottom of lines 2 and 3 (Tipping Phase II; Figure 4.4). This provides evidence that gas advection (e.g., ‘chimney effect’) may not be able to penetrate to the pile interior and other gas-transport mechanisms (e.g., diffusion) may prevail. It is also possible that these differences could result in physical heterogeneities in: 126  1. composition (i.e., higher sulphide contents near the surface), and/or 2. particle size (e.g., finer particle size will likely have a higher water content and result in less oxygen available for pyrite oxidation),  Other indications that lines 2 and 3 are hot spots for sulphide mineral oxidation include the fact that SO4 concentrations in the effluent waters are highest for sub-lysimeter A (below line 2) and pH values are relatively low compared to the rest of the pile. While this provides good evidence that the middle of the pile (sub-lysimeter A, instrumentation lines 2 and 3) exhibits the largest sulphide mineral oxidation rates and that the top of the pile around lines 2 and 3 may be a ‘hot spot’, further analysis is still required to determine the principal mechanisms involved and to determine any long-term trends. This is further examined in Chapter 6.  4.6  GEOCHEMICAL MODELING  Outflow geochemistry from basal lysimeter D was analyzed with PHREEQC v. 2.15 using the following parameters: temperature, pH, Alkalinity (as CaCO3,) ,S(6) as SO4, Cl, N(5) as NO3, N(-3) as NH3, F, Al, Sb, As, Ba, B, Cd, Ca, Cr, Co, Cu, Fe, Pb, Mg, Mn, Hg, Mo, Ni, K, Se, Si, Ag, Na, Sr and Zn. Several species were below or near the detection limit (e.g., Al, Cd, Cr, Co, Cu, Fe, Pb, Hg, Mo, Ag) but were included to remain consistent with similar geochemical modeling of the field cells (Aranda, 2009). Modifications were made to PHREEQC’s Wateq4f database in order to account for Mo thermodynamics not included in the database (e.g., Carroll et al., 2006; Stollenwerk, 1998) as per Conlan (2009).  Chemical modeling of basal lysimeter D shows that, with the exception of 4 samples, charge balance errors are within an acceptable range of ± 10 % (Figure 4.16). Based on the fact that minerals will tend to precipitate when the saturation index (SI) > 1, dissolve at SI < -1, and are ‘in equilibrium’ when -1 < SI < 1 (Sherlock et al., 1995; Golder, 2007), speciation calculations indicate that the effluent is near saturation with respect to gypsum and calcite. These are anticipated due to the known abundance of carbonate minerals and presence of sulfide minerals within the pile. Gypsum SI values exhibit a seasonal variability, with increasing SI values during the dry season and decreasing values during the wet season. It would also be expected that 127  porewaters are undersaturated with respect to gypsum near the top of the pile but approach saturation further along the flowpath, near the bottom of the pile where gypsum precipitation may occur. This can only by hypothesized because no consistent SO4 data is available from the soil water samplers (Section 4.4.4.1).  Gypsum SI values are almost entirely < 0 except for brief periods in August/September during 2007 and 2008 when SI > 0 (Figure 4.17). During the dry season, gypsum may form coatings around minerals found in dry pockets, and may be first to dissolve when moisture returns to these areas, resulting in concentration spikes during the initial wet season flushing. Calculated saturation indices for calcite show a slight increasing trend with time (Figure 4.18). Gypsum and calcite SI in composite tank E indicate that while gypsum remains similar between basal lysimeter D and E, calcite is reportedly higher in E (i.e., SI > 0) providing supporting evidence that CO2 may be degassing, or solid phase calcite may be accumulating in the composite tank. In addition to calcite and gypsum, several other minerals are at or approaching saturation (Table 4.4). These minerals include sulphates, carbonates, and oxides, and do not contribute significantly to the assemblage of primary minerals found in the pile. The presence of these minerals may indicate that secondary mineral formation and attenuation mechanisms may be occurring. The sulphate minerals exhibit seasonal trends similar in fashion to gypsum whereas carbonate and oxide SI values are all increasing towards saturation much like calcite. The silica in the oxides is likely present from the silicate minerals in the pile. Phases that exhibit increasing SI values may eventually start to precipitate if they continue on this trend.  4.7 4.7.1  DISCUSSION Gypsum Precipitation and the SO42-/Ca2+ Ratio  Pyrite oxidation and calcite dissolution/acid buffering pathways will proceed as described in equations 4-1, 4-2, 4-3 and 4-4. Pyrite oxidation can drive the dissolution of carbonates through the production of hydronium ions. As such, for every mole of pyrite oxidized (in the presence of calcite), 2 moles of sulphate and 4 moles of calcium are produced.  128  An examination of basal lysimeter D effluent concentrations is useful in determining the release and attenuation mechanisms that are occurring within the pile. Figure 4.19 indicates that the effluent water is dominated by SO4 (anion) and Ca (cation) ions. Geochemical modeling indicates that the porewater at the base of the pile is saturated with respect to gypsum (Section 4.6), and will lead to the precipitation of gypsum under the following pathway: 𝐶𝐶𝐶𝐶2+ + 𝑆𝑆𝑂𝑂42− → 𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆4(𝑠𝑠)  [4-5]  A regression analysis comparing the molar concentration of calcium and sulphate illustrates that there is a near proportional positive molar relationship (r2 = 0.62) between Ca2+ and SO42 (Figure 4.20), which is consistent with the sulphate release by pyrite oxidation and calcite dissolution, and pH buffering (equations 4-1 and 4-3). These results indicate that CO2 is degassing (equation 4-4). On the other hand, the precipitation of gypsum includes a 1:1 removal of Ca and SO4 ions, and does not affect the SO4:Ca ratio. That the molar ratio is consistent indicates that there are no other primary sources or sinks of Ca and/or SO4. In an open, pH-neutral system where HCO3-, H2CO3, and CO2 are at equilibrium, the dissolution of calcite by acid produced from sulphide oxidation may result in an increase of dissolved CO2 which can lead to a release of CO2 gas (equation 4-4). This dissolution pathway is supported by the observed CO2 gas enrichment with depth (5 times atmospheric) which indicates that the pile is not entirely behaving like an open system and that the interior is likely behaving more like a partially-closed system. Although CO2 concentrations are not as high as reported in other waste rock piles (e.g., Pantelis and Ritchie, 1992; Wels et al, 2003; Sracek et al., 2004), this indicates that CO2 enrichment is generally occurring towards the pile interior. 4.7.2  Zinc: Source and Attenuation Mechanisms  The primary source of zinc in test pile 1 pore waters is from the chemical oxidation of sphalerite [(Zn, Fe)S] according to the following pathway: 𝑍𝑍𝑍𝑍𝑍𝑍(𝑎𝑎𝑎𝑎 ) + 2𝐻𝐻2 𝑂𝑂 + 𝑂𝑂2 → 𝑍𝑍𝑍𝑍2+(𝑎𝑎𝑎𝑎 ) + 𝑆𝑆𝑂𝑂42−(𝑎𝑎𝑎𝑎 ) + 4𝐻𝐻 +(𝑎𝑎𝑎𝑎 ) 129  [4-6]  Note that although the chemical oxidation of sphalerite generates sulphate ions and acidity, due to a small proportion of sphalerite present in the waste rock compared to pyrite, pyrite is still considered to be the principal source of acidity, heat and sulphate ions within test pile 1 porewaters.  Based on the spatial and temporal variability of Zn (Figure 4.8) and results of chemical modeling, there is no clear evidence that any attenuation mechanisms are active that would remove a considerable amount of Zn from the porewaters. Zinc concentrations generally increase with depth and experience seasonal variability, and geochemical modeling indicates that no Znbearing mineral phases are near saturation. Furthermore, Figure 4.21 illustrates that while there is a reasonable trend between sulphate and zinc in C, D and E, there is no strong trend in A or B at this time.  A possible attenuation mechanism for Zn may be the sorption of Zn on carbonate minerals (Conlan, 2009), but there is no strong evidence of this in the test pile data. In addition, it is also possible that Zn may be retained in secondary mineralization (if it is occurring), but this also cannot yet be confirmed.  4.7.3  Spatial Variability  Instrumentation lines 1, 2 and 4 each correspond with a sub-lysimeter based on location within the pile. It is observed that concentrations do not always increase with depth and do not always correspond well with the sub-lysimeter data, suggesting that internal chemistries may be higher than reported flowing out of the base of the pile. The following list postulates some possible explanations:  1. heterogeneity of mineral composition (similar material classification, but originates from different areas of the pit at different times and has different compositions), 2. mineral availability/reactivity (e.g., how much is occluded and how much is suitable for reaction sites), 3. heterogeneity of particle size,  130  4. spatial and temporal flow variability (variable porewater travel times and scales, effluent from sub-lysimeters is a combination of matrix and macropore flow), 5. porewater pathways (assuming vertical flow of water, may pass through 2 or possibly 3 different types of waste rock before reaching the base of the pile i.e., Tipping Phase I and II, or II and III +/- the protective layer), 6. biased sampling of soil water samplers (i.e., sampling from matrix and can overestimate surrounding porewater chemistry due to unknown macropore saturation), 7. sample volume (i.e., soil water samplers vs. sub-lysimeter), 8. soil water samplers do not always lie directly above a lysimeter, and 9. unequal rates of sulphide mineral oxidation and calcite dissolution.  4.7.4  Comparison to Field Cell Chemistry  A brief comparison of effluent chemistry between field cells constructed by Aranda (2009) using the same material as used in the construction of test pile 1 is examined over the entire study period. Seasonal field cell solute trends are generally consistent with the test pile in which concentrations increase during the dry season and decrease during the wet season, indicating that the dissolved load is also a function of the porewater residence time in the field cells. Field cells also indicate an initial flushing of the weathering products from the most reactive materials early in the study period.  Based on the pH, Ca and alkalinity evidence, the field cells are well aerated. For example, a majority of the field cell pH values range between 7.5 and 9 (compared to 7 and 8.5 for test pile 1), and higher pH values are more indicative of open system conditions. High pH values (for field cells and pile at the end of the wet season) correspond to higher Ca concentrations (equation 4-2). Ca concentrations range between 300 and 450 mg/L (end of dry season) to 150 mg/L (end of wet season). In comparison, Ca levels in test pile 1 reach up to 1200 mg/L earlier on, but decline to 700 mg/L at the end of the study period. Alkalinity (as CaCO3) in the field cells ranges between 30 and 60 mg/L, which are similar to pile levels (consistent with relatively well-aerated system where CO2 is degassing).  131  Further evidence of sulphide mineral weathering is found in the electrical conductivity (i.e., specific conductance) and SO4 levels (which are relatively well correlated, as described in Chapter 5). Field cell SC varies between 200 and 1000 µS/cm, but reaches as high as 1000 2000 µS/cm (at the end of the dry season and in the black marble and marble diopside samples in particular). Test pile SC values reach up to 6000 µS/cm at the beginning of the study period, but decline to maximum values of 3000 µS/cm at the end of the study period. This signifies that the accumulation of weathering products (dissolved) along the pathway is decreasing over time, likely indicating an overall decrease in waste rock weathering rates (solid phase). Higher EC values correspond to the black marble and marble diopside material. Test pile SO4 concentrations range between 500 mg/L (end of wet season) up to 2000 mg/L (end of dry season) whereas the field cells range between 200 - 300 (end of wet season) to 700 - 900 mg/L (end of dry season).  The largest discrepancy in effluent chemistry between the field cells and test pile 1 occurs with zinc concentrations. Test pile Zn concentrations range from 0.5 – 2.5 mg/L in the effluent and up to 5.5 mg/L in the internal porewaters as measured by the soil water samplers. On the other hand, field cell Zn concentrations range between 0.4 to 17 mg/L, with the highest concentrations occurring in the marble diopside field cells. These high concentrations may be the result of the chance selection of high Zn-bearing materials, or perhaps due to the reasons previously described in Section 4.4.1.1.2.  That the test pile and field cell pH and alkalinity values are generally similar is rather expected based on open-system conditions. That specific conductance values for test pile 1 are 3 – 6 times higher than those of the field cell, despite the fact that the test pile is almost 10 times taller than the field cells, indicates that there is an accumulation of weathering products along the pathway, but that there is not a linear proportion between test pile and field cell SC values. That these results are non-linear may be a function of:  1. particle size differences, 2. preferential flow, 3. mixing, 4. continuous release, 132  5. oxygen availability, and/or 6. solubility controls (e.g., sorption, secondary precipitation, etc.).  Previous Class B field cells constructed by Golder (2006) indicate that there is a large variety in mineralogical composition of Class B waste rock at Antamina. Zinc concentrations in marble, hornfels and limestone field cells range between 0.002 and 0.2 mg/L (Golder, 2007b), which is 2 orders of magnitude below the test pile 1 field cells (Aranda, 2009). This large discrepancy between solute concentrations in various Class B effluent indicates that further mineralogical classification is required to distinguish between the more and less reactive sub-classes of this waste rock material. Further work is required to determine whether other Class B material will leach such high Zn concentrations.  4.8  SUMMARY  The following are the most salient conclusions from test pile 1 data:  1. test pile 1 chemistry results support evidence of sulphide mineral oxidation and a subsequent buffering of the generated acid, 2. chemistry shows distinct spatial and temporal variability in the pile with the outer region of the pile behaving like an open system whereas the internal pile behaves more like a partially-closed system, 3. elevated porewater solute concentrations and increased temperatures during the first wet season indicate an initial flushing of the weathering products of the most reactive particles. Increasing concentrations typically correlate with longer flowpath lengths, but the observed variability suggests the presence of pockets of elevated oxidation and dissolution rates, and reactive mineral contents, 4. concentrations do not change dramatically, which suggests that chemistry will evolve slowly over time (i.e., days and weeks and not hours or minutes), 5. pile structure and instrumentation location lead to variations in pile chemistry that are further complicated by transient flow mechanisms (matrix vs. macropore), 6. concentrations generally increase with depth in lines 1 and 2, and remain the same or actually decrease at depth in line 4, the latter is likely due to location differences (batter). 133  The effects of the pile batters and initial waste rock composition have direct results on effluent chemistry, with line 4/sub-lysimeter C showing the lowest concentrations, line 2/sub-lysimeter A showing the highest and line 1/sub-lysimeter B showing intermediate concentrations. Basal lysimeter D typically shows a range that is representative of all three sub-lysimeters, 7. despite the coarse nature of the waste rock, matrix-flow is the dominating flow mechanism in this pile (e.g., concentrations do not undergo rapid changes over a few hours), but SC data provides evidence that suggests macropore flow is present e.g., sharp increases in basal lysimeter D and sub-lysimeter C outflow result in sharp decreases in specific conductance. Further evidence of macropore flow is evident in the pH and Ca data, 8. temperature data indicates that sulphide mineral oxidation is occurring, and slight increases in some pile regions provides additional support that the heterogeneous pile will experience varying rates of pyrite oxidation that may result in creating local ‘hot’ spots of higher sulphide mineral oxidation, 9. despite Cu, Fe, Pb and other metals present in the waste rock (Chapter 2), these are all rapidly immobilized upon dissolution, indicating that the only metal in which attenuation mechanisms are not significantly present is zinc (i.e., exceeds receiving and discharging limits, 10. test pile 1 is relatively young and it is likely that conditions are not providing an optimal environment (pH, sulphide oxidation, temperature, etc.) for the establishment of significant microbial consortia that catalyze sulphide mineral oxidation (Lin and Herbert, 1997), 11. there is little evidence of thermal-induced convection, which may indicate that most of the gas exchange between the pile and the atmosphere is supported through wind-driven advection near the outer shell and diffusion-dominated transport towards the interior, and 12. initial pore gas analyses indicate that oxygen remains at near-atmospheric levels throughout the pile, signifying that either sulphide mineral oxidation is occurring at low levels or there is rapid replenishment of oxygen succeeding these reactions. Based on the presence of sulphate ions in solution, it is evident that sulphide mineral oxidation is occurring and that oxygen is being replenished to some extent. 134  TABLES  Table 4.1. Parameter comparison between basal-lysimeter and composite tank. Average Specific Conductance  pH  Zn  Ca  Na  Mg  SO4  Alkalinity  (Field)  (Field)  Dissolved  Dissolved  Dissolved  Dissolved  Dissolved  Total  mg/L  mg/L  mg/L  mg/L  mg/L  mg/L  uS/cm Wet Dry  D  2251  7.716  1.19  504.3  10.9  20.5  1081.2  72.1  E  2369  8.055  1.20  509.2  11.3  20.4  1102.9  69.7  D  2606  7.595  1.50  592.6  12.4  23.2  1390.6  81.6  E  2505  8.120  1.27  561.5  13.5  22.3  1263.4  74.3  Table 4.2. Temperature comparison between top and bottom of pile.  Top of Pile Bottom of Pile Difference Air Temperature  Mean Temperature [°C] Wet Season (1-Oct-2007 to 30-Apr-2008) 7.6 5.8 1.7 5.7  Water Year (1-Oct-2007 to 30-Sept-2008) 7.6 5.8 1.8 5.6  *based on temperature data from instrumentation lines 2 and 3  135  Dry Season (1-May-2007 to 30-Sept-2008) 7.7 5.7 2.0 5.5  Table 4.3. Solute discharging and receiving limits Parameter As - dissolved Cu - dissolved Fe - dissolved Pb - dissolved pH - field Mo - dissolved Zn - dissolved S  Units mg/L mg/L mg/L mg/L pH Unit mg/L mg/L mg/L  Discharge Limit1 0.5 – 1 0.3 – 1 1–2 0.2 – 0.4 6-9 – 0.5 – 1  Notes MEM2 MEM2 MEM2 MEM2 MEM2 – CMA3/WB4  Receiving Limit1 0.2 0.022 – 0.025  Notes GWL-III5 CMA-RW6  0.1 6–9 0.2 0.13 – 0.21 0.002  GWL-III5 GWL-III5 ECT7 GWL-III5, ECT7 GWL-III5  1  as of 2006 (Golder, 2006) MEM: Peruvian Ministry of Environment and Mines 3 CMA: Compañia Mineral Antamina standards 4 WB: World Bank. 5 GWL-III: General Water Law – Class III 6 CMA-RW: Dissolved Cu and Zn to replace CMA Total Cu and Zn. 7 ECT: eco-toxicity test performed by Stantec. 2  Table 4.4. Minerals modeled to be near saturation in basal lysimeter effluent. Mineral Type Sulphate  Carbonate  Oxide  Other  Mineral Name Anhydrite Barite Celestite Aragonite Dolomite (ordered and disordered) Smithsonite Chalcendony Cristobalite Quartz Fluorite  136  Mineral Composition CaSO4 BaSO4 SrSO4 CaCO3 CaMg (CO3)2 ZnCO3 SiO2 SiO2 SiO2 CaF2 Zn(OH)2(γ), ZnCO3:H20 ZnO(active).  FIGURES  16  8000  6000  12 10  4000  8 6  2000  Effluent [m3/day]  Specific Conductance [uS/cm]  14  Pile Discharge  4 2  0 Jan-07  Apr-07  Jul-07  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  0  6000  600  5000  500  4000  400  3000  300  2000  200  1000  100  0  A High-Resolution B High-Resolution C High-Resolution D High-Resolution A Low-Resolution B Low-Resolution C Low-Resolution D Low-Resolution  Effluent [L/hr]  Specific Conductance [uS/cm]  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Pile Discharge 1-Mar-07 15-Mar-07  0 1-Apr-07 15-Apr-07  Figure 4.1. Continual and weekly-obtained specific conductivity vs. and pile discharge.  137  4000 y = 0.81x + 401  3000 2000 1000 1000  2000  3000  4000  b) pH  Basal Lysimeter D  Basal Lysimeter D  a) Specific Conductance [µS/cm]  9 y = 0.71x + 1.9  8.5 8 7.5 7 7  8  Composite Tank E  Composite Tank E  d) Zinc [mg/L]  1000 800 600 400 y = 0.95x + 23  200 200 400 600 800 1000  Basal Lysimeter D  Basal Lysimeter D  c) Calcium [mg/L]  2 1.5 1 y = 0.81x + 0.26  0.5 0.5  Composite Tank E  1  Basal Lysimeter D  1.5  Composite Tank E  e) Sulphate [mg/L]  2000 1500 1000 y = 0.95x +57  500 500  9  1000 1500 2000  Composite Tank E Figure 4.2. Parameter comparison between basal lysimeter and composite tank.  138  2  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.3. Temporal and spatial distribution of specific conductance.  139  Figure 4.4. Cross-sectional diagram of four construction stages and four lysimeters  140  Temperature [oC]  9  a) Pile Temperature – Deep Probes Air Temp L1T1 L1T3 L2T1 L2T2 L2T3 L3T1 L3T2 L5T2 L5T3 L6T3  8 7 6 5  4 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08  Temperature [oC]  10  b) Pile Temperature – Shallow Probes Air Temp L2T5 L2T6 L3T5 L3T6 L4T1 L4T2 L4T3 L4T4 L4T5 L4T6 L5T1 L6T1 L6T2  8  6  4  2 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 *all temperature probes are located within test pile 1  Figure 4.5. Internal and external pile temperature variation.  141  Figure 4.6. Average temperatures in lines 1, 2 and 4 over the water year.  142  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.7. Temporal and spatial distribution of pH.  143  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.8. Temporal and spatial distribution of Zn.  144  0.5 0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 4 2 0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 0.02 0.01  0.05  400  Zn-D [mg/L]  Cu-T [mg/L]  0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 0.1  Ca-D [mg/L]  Cu-D [mg/L] Fe-T [mg/L] Fe-D [mg/L]  1  0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 800  0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08  2 1 0 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08  Basal Lysimeter D Composite Tank E Sump F Figure 4.9. Sump comparison to basal lysimeter and composite tank.  145  Concentration [mg/L] *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.10. Temporal and spatial distribution of Ca.  146  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.11. Temporal and spatial distribution of Na.  147  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.12. Temporal and spatial distribution of Mg.  148  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.13. Temporal and spatial distribution of K.  149  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.14. Temporal and spatial distribution of SO4.  *solid lines indicate wet-dry season transition; dotted lines indicate dry-wet transition.  Figure 4.15. Temporal and spatial distribution of alkalinity.  150  Number of Samples  20 15 10 5 0 -15  -10  -5  0 Percent Error  5  10  15  *the largest absolute error is 41 %.  Figure 4.16. Percent charge balance error of PHREEQC model of basal lysimeter data.  G ypsum Saturation Index  0.5  Basal Lysimeter D Composite Tank E  0.25 0 -0.25  -0.5 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Figure 4.17. Gypsum SI for the basal lysimeter D and composite tank E.  Saturation Index  1  Calcite Basal Lysimeter D Composite Tank E  0.5 0 -0.5 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08  Figure 4.18. Calcite SI for the basal lysimeter D and composite tank E.  151  Figure 4.19. A trilinear Piper Plot of basal lysimeter D effluent  Basal Lysimeter Ca [mmol/L]  25 20 15 10 5 0 0  y = 0.78x + 3.48 r2 = 0.62  5  10 15 Basal Lysimeter SO 4 [mmol/L]  Figure 4.20. Regression analysis of Ca and SO4 at basal lysimeter D. 152  20  25  0.04 B Zn [m]  A Zn [m]  0.04  0.02  0 0  0.02  0 0  5 10 15 20 25  B SO4 [mmol/L]  A SO4 [mmol/L]  0.04 D Zn [m]  C Zn [m]  0.04  0.02  0 0  5 10 15 20 25  0.02  0 0 5 10 15 20 25  5 10 15 20 25  D SO4 [mmol/L]  C SO4 [mmol/L]  E Zn [m]  0.04  0.02  0 0  5 10 15 20 25 E SO4 [mmol/L]  Figure 4.21. Correlation between SO4 and Zn at sub-lysimeters A, B & C, basal lysimeter D, and composite tank E.  153  5  RESULTS: TEST PILE MASS LOADINGS  Solute mass loadings represent a critical component in understanding and predicting the metal release capability of a waste rock environment. The solute mass loading behaviour of test pile 1 is examined based on flow and chemistry results presented in Chapters 3 and 4, respectively. This chapter will:  1. examine the relationships between flow, chemistry and solute mass loading of test pile 1, 2. determine solute-specific correlations to specific conductance at various lysimeters, 3. compare low- and high-resolution mass loading estimates for SO4, Ca and Zn, 4. determine mechanisms that control solute mass loadings, 5. develop first-pass estimations of metal depletion rates that will aid in characterizing the relative strength of metal release mechanisms dominating the first 2 years of pile operation, and 6. compare test pile 1 loading rates with those from field cells of the same material.  5.1 INITIAL PILE COMPOSITION The solid-phase elemental composition of waste rock placed in the pile from each of the four material placement stages (protective layer and Tipping Phases I-III) is summarized in Table 2.4. The average elemental mass is calculated using the solid phase elemental concentration s [%] and mass M [kg] of waste rock associated with each placement stage. When normalized by the total pile mass MT [kg], the average proportion of each element in the pile, sP [%], is calculated using the following weighted average calculation:  sP = [(s1 * M1) + (s2 * M2) + (s3 * M3) + (s4 * M4)]/MT  [5-1]  where materials used in the protective layer, Tipping Phase I, Tipping Phase II and Tipping Phase III are denoted by subscripts 1, 2, 3 and 4, respectively. The most notable results indicate that calcium, sulphur and zinc make up nearly 27.1 and 0.61 and 0.09 % of the pile by weight, respectively (Table 2.4).  154  5.2 SOLUTE MASS LOADINGS Solute mass loading is a transient term based on transient flow and variable effluent chemistry. Mass loading ML [mg/week] is calculated from the volume of water Q [L/week] and solute concentration c [mg/L] at a given time based on the following relationship:  ML = Q * c  [5-2]  Using equation 5-2, the solute mass loading of this test pile is calculated using two different approaches:  1. a ‘low-resolution’ approach (Section 5.2.1, below), and 2. a ‘high-resolution’ approach (Section 5.2.2, below)  The former approach is based on near-weekly geochemical sampling and flow, whereas the latter is based on continually-logged electrical conductivity and flow measurements. The resulting mass loading rates are presented in mg/week. Loadings are then normalized by the waste rock mass of the pile (MT ~ 20,000 metric tonnes; Chapter 2) to obtain a mass loading ML [mg/kg/week] described as ‘mg of solute released per kg of waste rock per week’. To obtain mass loading rates for each of the four lysimeters, the loadings rates are normalized by the estimated mass of waste rock lying directly above the 36 x 36 m basal lysimeter or 4 x 4 m sublysimeters.  Solute mass loadings are based on the flow estimates presented in Chapter 3, where methods used to account for data gaps in flow are explained, and chemistry presented in Chapter 4. Sulphate is the dominant anion with the highest concentration in the effluent waters, calcium is the dominant cation, and Zn is the only trace heavy metal found at any level of interest (e.g., above the discharge limit). Calcium is an indicator of acid neutralization by carbonates. These three solutes serve as the basis for the following loading discussion. Based on the sampling schedule described in Chapter 2, weekly chemistry data is not available for all lysimeters (e.g., only when they are flowing). For the following low-resolution mass loading calculations, chemistry values are interpolated between the nearest points before and after the given date. 155  5.2.1  Low-Resolution Mass Loadings  Low resolution mass loadings ML are calculated using weekly QL totals that correspond to the chemistry sampling days in which the cL values are obtained (refer to Chapter 4). Table 5.1 presents the cumulative and weekly-averaged flow and loadings separated into wet and dry periods. It is observed that pronounced wet and dry periods result in significantly different effluent and loading trends. Average weekly discharge and loadings rates are calculated to account for differences in sampling frequency and period length (i.e., wet and dry seasons are 7 and 5 months long, respectively). Average cumulative wet season loadings and time-averaged loading rates are larger than the average dry season loadings by a factor of at least 6. For example, the average wet season SO4 loading rate is 1.45 mg/kg/week and the dry season loading rate is 0.23 mg/kg/week (Table 5.1). Discharge volume differences vary by nearly the same factor. This indicates that flow is of greater influence to mass loadings compared to solute concentration, and is consistent with trends reported by Wagner (2004).  5.2.1.1 Cumulative Loadings The cumulative SO4, Ca, and Zn loadings for the entire period are shown in Figure 5.1 and indicate that 1343 kg, 613 kg and 1.4 kg (respectively) are released from the pile throughout the entire study period. These figures include combined loadings from the three sub-lysimeters and the basal lysimeter. All three curves exhibit the same shape as the total estimated precipitation (Chapter 3), which indicates that the seasonal change between wet and dry seasons are the dominant influence governing these mass loadings at this site and that the greatest proportion of the loadings is released during the wet season.  5.2.1.2 Time-Averaged Loading Rates Average weekly mass loadings are calculated based on the outflow from each lysimeter. Results are then normalized by the estimated mass of waste rock associated with each respective lysimeter (assuming only vertical flow). For example, pile mass is estimated at 20,000 tonnes of waste rock (Chapter 2), of which an estimated 326 tonnes lie directly above sub-lysimeter A, 330 tonnes lie above B, and 240 tonnes lie above C. These masses are estimated based on a 4 x 4 vertical prism of waste rock that lies directly above each sub-lysimeter (assumes vertical flow). A significantly larger amount of material (~ 19,104 tonnes) contributes to the flow from basal 156  lysimeter D. The time-averaged, weekly-normalized loading rate [mg/kg/week] is referred to throughout the remainder of this investigation as the weekly loading rate.  The weekly SO4, Ca and Zn loading rates calculated at each lysimeter are presented in Figure 5.2. All lysimeters exhibit strong seasonal variations in loadings that vary by 2 - 3 orders of magnitude. Sub-Lysimeters A and C exhibit similar mass loading behaviour to basal lysimeter D, with maximum wet season SO4 loading rates between 4 and 6 mg/kg/week, Ca loadings between 2 and 3 mg/kg/week and between 0.003 and 0.005 mg/kg/week for Zn. Sub-lysimeter B wet season solute loading rates do not reach similar levels. This is likely due to both the flow behavior and composition of the material in this region of the pile. All lysimeters exhibit relatively low loading rates in the dry season. Differences between sub-lysimeter B and sub- and basal lysimeters A, C and D, respectively, as well as seasonal differences are further explained in the following chapter.  5.2.1.3 Flow vs. Chemistry Weekly SO4, Ca and Zn loading rates exhibit opposite seasonal trends from the respective solute concentrations (Figure 5.3). Despite decreasing solute concentrations during the wet season, mass loadings continue to increase, indicating that flow rate has a larger influence on mass loading behaviour than solute concentrations. This is explained by flow rates (Q) that vary by up to three orders of magnitude whereas solute concentrations (c) decrease by a factor of 2 to 5 (equation 5-2). In a similar investigation on a 5 m-high waste rock pile at the Cluff Lake mine, Wagner et al. (2006) report similar trends except for the period directly after the onset of large infiltration events where a short but rapid increase in concentration is reported. This is explained by a flushing effect where weathering products that have accumulated over a period of relative dryness are easily dissolved under large precipitation events, and lead to higher dissolved loads in the porewater. Over a larger time frame (e.g., monthly) results of the 5 m test pile indicate that flow rates and dissolved metal concentrations are positively correlated, which is the opposite of data presented for Antamina’s test pile 1.  Chapter 4 explains that the solute concentrations are in part dependent on the residence time of porewater in the pile, and Chapters 3 and 4 explain that the different transport pathways (i.e., 157  matrix vs. preferential flow) also affect solute concentration and loading behaviour. In similar investigations, long-term mass loading behaviour is found to be generally dominated by matrixflow water, with peak mass loadings resulting from preferential flow during periods of highintensity infiltration (Wagner et al., 2006; Neuner, 2009). Similar to these investigations, in response to large precipitation events, the chemistry of test pile 1 continues to increase until a (relatively) fresh slug of water penetrates to depth and decreases the dissolved load of the effluent water. Test pile 1 thus responds similarly to large precipitation events, however the dataset contains some uncharacteristically high solute concentrations during periods of high flow (e.g., SO4 concentrations associated with basal lysimeter D on 1-May-2007; Figure 5.3). 5.2.1.4 Specific Conductance-Solute Correlations Weekly solute concentrations are correlated to the weekly-measured specific conductance (SC) reported in Figure 4.1. Figure 5.4 provides an example of the correlations calculated based on basal lysimeter D effluent, and Table 5.2 summarizes the SC-SO4, -Zn and -Ca correlations (and r2 values) calculated at each lysimeter. Solute concentrations are estimated based on the correlation equations provided in Table 5.2. For example, the SC-SO4 correlation is: [ SO4 ] = 0.532 * SC - 37.3  [5-3]  where, sulphate concentrations ([SO4]) are given in mg/L and the specific conductance (SC) is given in µS/cm. Table 5.2 indicates that correlations between SC and SO4, Zn and Ca range from strong (r2 = 0.86) to non-existent (r2 = 0.01). Basal lysimeter D exhibits a strong SC-Ca relationship (r2 = 0.82), with a moderate correlation for SC-SO4 (r2 = 0.56) and a weak correlation for SC-Zn (r2 = 0.19). Sub-lysimeter C has the best SC-solute correlation and all lysimeters have relatively weak SC-Zn correlations. Poor SC-Zn correlations (Figure 4.21) are likely a result of the fact that Zn is a trace metal and has little effect on the SC, which is largely dominated by SO4 and Ca ions. These correlations are used in the following section to estimate high-resolution solute concentrations and mass loadings.  158  5.2.2  High-Resolution Mass Loadings  Under certain conditions (e.g., in response to a large precipitation event), solute concentrations may change over a short period of time (Wagner et al., 2006). High-resolution mass loadings are therefore required to monitor the short-term fluctuations in mass loadings that the previous (lowresolution) method does not account for.  High-resolution mass loadings MH are calculated using solute correlations to specific conductivity (Section 5.2.1.4) to estimate cH, and flow measurements QH are presented for the same time interval. One of the principal benefits of installing custom-made EC probes is that they provide high-resolution SC data (i.e., 30-minute intervals) that can show rapid responses to precipitation inputs. The continual SC data is similar to the weekly, manually collected SC data (Figure 5.1).  5.2.2.1 Basal Lysimeter SO4 Loading Using the regression correlations provided for basal lysimeter D in Table 5.2 and the continuous SC data, SO4 loadings are estimated over three (3) separate weeks during the 2006-07 wet season (7- to 15-Mar-2007, 7- to 15-Apr-2007 and 8- to 15-May-2007). Results are compared to flow, specific conductance and estimated sulphate concentrations (Figure 5.5; Table 5.3). These time periods are chosen because all of the chemistry (lab, field, and continual SC data) and flow data are present. For data processing purposes, SC and flow data are reduced to hourly intervals. Estimated hourly solute concentrations [mg/L] are multiplied by hourly lysimeter flow [L/hr] to obtain a loading rate that is converted to an equivalent kg SO4/day. The inverse relationship between flow rate and solute concentration (Section 5.2.1.3) also applies to mass loadings and specific conductance, indicating that flow rate and mass loadings are linearly proportional to each other much like SC and solute concentrations are to one another (Section 5.2.1.4). However, SC and solute concentration both respond slower and less proportionally than flow rate and mass loadings (due to magnitude in changes, Section 5.2.1.3). For example, under extreme events, flow and loading rates can increase by more than 50 % over one day whereas SO4 and SC experience less proportional, more attenuated responses (Figure 5.5). 159  5.2.2.2 Lysimeter SO4, Ca and Zn Loadings Whereas Section 5.2.2.1 examines the SO4 loadings at D, results presented in this section compare the SO4, Ca and Zn loading rates at each lysimeter during an 8-day period during the week of 11- to 19-Apr-2007.  Weekly sub-lysimeter C and basal lysimeter D loading rates are higher than A and B for SO4, Zn and Ca (Figure 5.6). The total pile flow from all four lysimeters over the course of the period is also included, and illustrates that sub-lysimeter C and basal lysimeter D loading rates are more responsive to the changing discharge compared to sub-lysimeters A and B, which experience much more gradual, attenuated responses. Basal lysimeter D responds more rapidly than sublysimeter C. The highest SO4 and Ca loading rates are found at the front of the pile (sublysimeter C), whereas the lowest loading rates are found towards the pile interior (sub-lysimeter B). There is a poor SC-Zn correlation for all lysimeters (Table 5.2), which possibly explains why basal lysimeter D has a higher Zn loading rate than sub-lysimeter C. Otherwise, loadings are generally consistent among the three solutes as measured at the four lysimeters, and the relative difference in loading rates between lysimeters is consistent with the flow patterns reported in Chapter 4.  5.2.3  Comparison Between Low- and High-Resolution Loadings  Low- and high-resolution loading rates are compared. High-resolution data indicates that significant changes in effluent chemistry are occurring that the low-resolution (weekly) data is not effectively capturing.  Section 5.2.2.1 examines SO4 loadings at basal lysimeter D over three one-week periods. The high-resolution specific conductance and SO4 concentrations are averaged over the entire period to enable a comparison with low-resolution parameters. In general, both approaches produce similar results with SO4 concentrations and SC values having inverse behaviours to increasing flow and mass loading rates.  160  Section 5.2.2.2 compares the low- and high-resolution solute mass loading estimates for each lysimeter during 11- and 19-Apr-2007. Both methods (low and high) produce similar results, with minor differences (Table 5.4). With the exception of SO4 at sub-lysimeter A, this trend (low-resolution result < high-resolution result) is observed for all three solutes at each lysimeter. High- and low-resolution loading rates are similar for SO4 (i.e., 10.6 (low-resolution) and 12.7 (high-resolution) mg/kg/week at basal lysimeter D) and Ca (5.3 (low) and 5.5 (high) mg/kg/week at D). Low-resolution SO4 rates are nearly 85 % of the high-resolution rates, whereas lowresolution Ca rates are 97 % of the high-resolution rates. Zn experiences the largest discrepancy between low and high resolution solute loading values (i.e., 0.010 and 0.015 mg/kg/week), and low-resolution rates are roughly 70 % of the high-resolution rates. The larger discrepancy from Zn likely arises from a poor SC-Zn correlation (Table 5.2).  Based on Table 5.3 and Table 5.4, average high-resolution SO4 concentrations are up to 50 % higher than the low-resolution values, whereas high-resolution Ca and Zn loadings are 2 - 8 % and 30 - 50 % higher than the low-resolution loadings, respectively. It is difficult to determine whether the low-resolution approach underestimates loadings or whether the high-resolution approach overestimates loadings. Comparing solute correlations between the basal lysimeter and composite tank (Figure 4.2), this suggests that the low-resolution approach likely produces more accurate results than the high-resolution approach. This is because each composite tank sample represents an integration of flow from all four lysimeters over the course of the week, whereas the basal lysimeter sample represents an instantaneous (weekly) snapshot of concentration that is not indicative of how concentrations varied since the last sample (Appendix G). Since weekly fluctuations in the basal lysimeter concentrations are not significant on the overall pile loadings (Figure 4.1), this suggests that the low-resolution approach likely produces more accurate results than the high-resolution approach, which is based on an empirical relationship and introduces even more error.  5.2.4  Comparison to Field Cells  To determine the weathering characteristics of waste rock in a more controlled domain than the test pile and to isolate distinct rock types which are mixed within the pile, eight 55-gallon kinetic field cell experiments were constructed using random samples of the same waste rock material 161  that was placed in the experimental pile (Aranda, 2009). Due to a smaller size fraction of waste rock employed in the field cells (i.e., < 10 cm), these cells typically have a higher surface area and a higher general reactivity compared to the test pile. Nearly two years of data is available from these field cells, the chemistry of which was briefly introduced in Chapter 4.  The average, maximum and minimum SO4, Ca and Zn loading rates [mg/kg/week] calculated for the field cells are compared at each lysimeter and the composite tank (Table 5.5). Results are based on data extending from Jan-2007 to Sept-2008. Lysimeter loading estimates are based on interpolated datasets when flow and chemistry data are absent (Section 5.2.1).  All solute mass loading rates (lysimeter and field cells) vary by nearly 2 - 3 orders of magnitude, demonstrating the large variability in loading rates throughout the study period. Average field cell loading rates are larger than the lysimeters by between a factor of 2 to two orders of magnitude. For example, while average SO4 loadings rates are nearly 3 - 4 times higher for the field cells than for the total pile, the maximum rate is 4 times higher and the minimum rate is 2 orders of magnitude higher. Similar trends are found in Ca, and Zn reactivity in the field cells is nearly 2 - 3 orders of magnitude higher (Table 5.5).  5.3 ELEMENT DEPLETION ESTIMATES The time of element depletion is crudely calculated based on mass loading rates. This simplified calculation is only included to provide a rough estimate of the possible leaching potential of the pile into the future.  Based on pile design and assuming vertical flow, infiltrating water will always come into contact with material originating from two or more material placement phases (i.e., protective layer and Tipping Phases I, II & III; Chapter 3). Water reporting to sub-lysimeter B will likely pass through Tipping Phases I and II and the protective layer, whereas water reporting to sublysimeter A will likely come into contact with material from Tipping Phases II and III, and the protective layer. Water reporting to sub-lysimeter C will come into contact with Tipping Phase III and the protective layer. The basal lysimeter will contain water in contact with all four  162  materials. It is for this reason that whole-pile depletions will only be estimated, and will not distinguish between various lysimeters or tipping phases.  Using average pile S, Zn and Ca concentrations and pile mass (Table 2.4), and the total mass of those three elements already released from the pile during the first 2 years (Table 5.1), the depletion estimates are calculated. Nearly 0.4 % of total S has been depleted from the pile, while 0.01 % of Ca and Zn have so far been depleted. To note, Golder (2007b) reported that less than 1 % of available metals in field cell kinetic tests have been depleted from a majority of Class B materials. Based on the assumption that the current rates stay constant in time, then it will take S nearly 500 years to be depleted from test pile 1, Zn nearly 23,000 years and Ca nearly 18,000 years. 5.3.1  Mineral Depletion Uncertainties  The purpose of the above calculation is to illustrate that the pile has the potential to release poorquality drainage water many years into the future. These metal depletion calculations have several inherent limitations. One major weakness is the assumption that the time-averaged solute mass loading rate remains constant. Calculations do not address the possibility of future decreases to solute release rates in response to a loss of reactivity, mineral coatings, etc. To improve the accuracy of the depletion estimation, longer term observations are required. With a more profound understanding of the geochemical and flow dynamics, higher-order mechanistic models can be used to better predict future loadings and depletion estimates.  To increase the accuracy of these calculations and minimize the limitations, several physical and chemical constraints should be addressed in the future. Some considerations are, but are not limited to, the following:  1. sulphide mineral availability (and surface area for reaction sites), 2. particle size heterogeneity (relative amount of more reactive fine-grained material), 3. composition heterogeneity (i.e., proportion of sulphide minerals), 4. rate and development of mineral coatings (e.g., secondary mineralization), 5. flow pathways and mechanisms (matrix vs. macropore flow), 6. rate of oxidation (availability/replenishment of oxygen, water, sulphide, etc.), 163  7. precipitation/dissolution tendencies (saturation indices), and 8. acid neutralization capability (and carbonate mineral availability).  5.4 SUMMARY A total 1343 kg of SO4, 613 kg of Ca, and 1.4 kg of Zn are estimated to have been removed from test pile 1 during the 2-year long study period. Weekly loadings vary by 2 – 3 orders of magnitude between the wet and dry season, with the highest rates of 4 – 6 mg/kg/week SO4, 2 – 3 mg/kg/week Ca, and 0.003 – 0.005 mg/kg/week Zn calculated for basal lysimeter D and sublysimeters A and C.  Flow rates at test pile 1 have a larger influence on the mass loading rate compared to solute concentration. In general, an increase in flow and mass loading rates at test pile 1 will typically cause a decrease in solute concentration and specific conductance. The high-resolution mass loading calculations indicate that mass loading rates can change by over 50 % over the course of a few days. First-pass elemental depletion estimates indicate that the pile has the potential to release solutes many years into the future. Field cell loading rate estimates are 2 – 200 times larger than those for test pile 1.  164  TABLES Table 5.1. Cumulative and average wet and dry season loading rates. Discharge  Time  Cumulative 3  Loadings  Average  Cumulative [kg]  3  Average [mg/kg/week]  Wet 2006-07 Dry 2007 Wet 2007-08 Dry 2008 Average Wet Average Dry  [m ] 447 96 623 65 535 81  [m /week] 30.4 4.4 20.5 3.0 25 4  SO4 513 115 633 82 573 98  Zn 0.42 0.11 0.76 0.10 0.59 0.10  Ca 242 55 282 33 262 44  SO4 1.59 0.29 1.31 0.18 1.45 0.23  Zn 1.1E-03 2.3E-04 1.1E-03 1.7E-04 1.1E-03 2.0E-04  Ca 0.79 0.14 0.59 0.08 0.69 0.11  Total / Average  1231  14  1343  1.4  613  0.76  7.9E-04  0.35  Table 5.2. Specific conductance-solute lysimeters and composite tank regression correlations. Station A B C D E  [SO4] = 0.0246(SC)+1354.5 0.375(SC)+275.2 0.465(SC)-0.98 0.532(SC)-37.3 0.393(SC)+226.7  r2 0.01 0.33 0.63 0.56 0.58  [Zn]= 0.00007(SC)+0.822 -0.0003(SC)+2.67 0.0004(SC)-0.010 0.0004(SC)+0.49 0.0002(SC)+0.794  r2  [Ca]=  r2  0.20 0.05 0.48 0.19 0.10  0.180(SC)+74.6 0.223(SC)-18.7 0.222(SC)-12.37 0.250(SC)-56.81 0.225(SC)-11.45  0.86 0.65 0.75 0.82 0.80  Table 5.3. High- and low-resolution basal lysimeter loading comparison; 2007. 2007 March 7-15 Cumulative Volume Specific Conductance SO4 Concentration Cumulative Mass Loading April 19-24 Cumulative Volume Specific Conductance SO4 Concentration Cumulative Mass Loading May 8-15 Cumulative Volume Specific Conductance SO4 Concentration Cumulative Mass Loading 1 2  Unit  Low Resolution1  High Resolution2  m3 µS/cm mg/L kg SO4  40.6 1988 760 30.9  40.6 1963 1013 39.8  m3 µS/cm mg/L kg SO4  15.7 2440 1055 16.6  15.7 2375 1233 19.3  m3 µS/cm mg/L kg SO4  12.9 2460 862 11.1  12.9 2368 1229 15.9  low-resolution specific conductance and SO4 concentrations are from the end of the sampling period high-resolution specific conductance, and SO4 are averaged over each period.  165  Table 5.4. High- and low-resolution lysimeter loading comparison; 11- to 19-April-2007. Solute  Parameter Cumulative Volume  SO4  Ca  Zn  Unit 3  [m ]  Specific Conductance  [µS/cm]  Concentration  [mg/L]  Mass Loading  [mg/kg/week]  Concentration  [mg/L]  Mass Loading  [mg/kg/week]  Concentration  [mg/L]  Mass Loading  [mg/kg/week]  Resolution  A  B  C  D  Low  0.166  0.08  0.7  33.09  31.9  m3  High Low High* Low High* Low High* Low High* Low High* Low High* Low High*  0.166 4300 4377 1488 1462 4.7 4.6 800 860 2.51 2.69 0.88 1.13 0.0027 0.0035  0.08 3050 3012 1355 1405 2.0 2.1 634 652 0.95 0.97 1.25 1.76 0.0019 0.0027  0.7 1800 1666 689 774 11.5 13.7 351 357 5.85 6.31 0.54 0.66 0.0090 0.0116  33.09 2640 2359 1055 1218 10.6 12.7 525 533 5.27 5.52 1 1.44 0.0100 0.0151  31.9 33.5 39.4 16.7 17.2 0.031 0.047  m3  * High-resolution averages over time period for comparison to low-resolution data  166  Cumulative  kg kg  kg kg  kg kg  Table 5.5. Average field cell and lysimeter loading rates. Loadings [mg/kg/week] SO4  Zn  Ca  Station  Average  Maximum  Minimum  Ratio: Field Cell : Station  Pile1 D C B A Field Cell Pile1 D C B A Field Cell Pile1 D C B A Field Cell  0.794 0.792 1.164 0.149 1.244 2.83 8.18E-04 8.28E-04 8.44E-04 1.86E-04 8.90E-04 3.58E-02 0.357 0.355 0.54 0.066 0.596 1.44  4.734 4.796 5.817 0.807 4.28 16.596 4.12E-03 4.18E-03 4.84E-03 8.99E-04 3.05E-03 3.11E-01 1.546 1.542 2.806 0.367 2.005 6.512  0.029 0.03 0.003 0.002 0.003 0.894 2.39E-05 2.49E-05 4.41E-06 3.28E-06 3.73E-06 6.94E-03 1.37E-02 1.43E-02 7.49E-04 9.50E-04 2.14E-03 0.375  4 4 2 19 2 1 44 43 42 192 40 1 4 4 3 22 2 1  1  Total pile rates are based on weighted average based on estimated mass of material.  167  FIGURES  SO4 [kg]  1500 1000 500  Ca [kg]  0 800  400  0  Zn [kg]  1.5 1 0.5 0  *Solid line = wet-dry season transition; Dotted line = dry-wet season transition  Figure 5.1. Best estimated cumulative pile loadings.  168  Loading [mg/kg/week]  6 5 4 3 2 1 0 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.005  SO4  C  B A D  Ca  Zn  0.004 0.003 0.002 0.001  *Solid line = wet-dry season transition; Dotted line = dry-wet season transition  Figure 5.2. Best estimated weekly SO4, Ca and Zn mass loadings for all four lysimeters.  169  Oct-08  Jul-08  Apr-08  Jan-08  Oct-07  Jul-07  Apr-07  Jan-07  0  2000  4  1500  2  1000  0 Jan-07  Apr-07  Jul-07  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  Basal Lysimeter Sulphate Concentration [mg/L]  Basal Lysimeter Sulphate Loadings [mg/kg/week]  6  500  -3  x 10  4  2.5  3  2  2  1.5  1  1  0 Jan-07  Apr-07  Jul-07  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  Basal Lysimeter Calcium Loadings [mg/kg/week]  2  Basal Lysimeter Zinc Concentration [mg/L]  3  0.5  1000  1.5  800  1  600  0.5  400  0 Jan-07  Apr-07  Jul-07  Oct-07  Jan-08  Apr-08  Jul-08  Oct-08  *Solid line = wet-dry season transition; Dotted line = dry-wet season transition  Figure 5.3. Weekly SO4, Ca and Zn solute concentration and mass loading.  170  200  Basal Lysimeter Calcium Concentration [mg/L]  Basal Lysimeter Zinc Loadings [mg/kg/week]  5  2500 2000  SO4  1500 1000 y = 0.5324x - 37.257 r² = 0.5425  Basal Lysimeter Concentration [mg/L]  500 0 1000 800  Ca  600 400 y = 0.2502x - 56.807 r² = 0.8172  200 0 3 Zn 2 1  y = 0.0004x + 0.4921 r² = 0.191  0 1000  2000 3000 Specific Conductance [µS/cm]  4000  Figure 5.4. Specific conductance correlation to SO4, Ca and Zn as measured at D.  171  Lysimeter Loadings vs. Pile Flow 2500  10 Flow  SC SO4 Loading  6 SO4  Specific Conductance [µS/cm] 1 Sulphate [mg/L] 1  500 Mar 06  08  10  12  14  2 16  3000  6  2000  4  1000 April 19  2 21  23  25  3000  4  2000  2  1000  0  May 08  10  12  14  Flow Rate [m3/day] 1 Load Rate [ kg SO4 / day] 1  1500  16  SC 1  based on hourly data  Figure 5.5. Basal lysimeter loading comparison to concentration and flow rate; 2007.  172  Lysimeter Loadings vs. Pile Flow 4  SO4  C D  2 A  4  B  0  0  2  8  Ca  1  4  0  0  0.004  8  Zn  0.002  4  0.000  0  11-Apr 13-Apr 15-Apr 17-Apr 19-Apr  Figure 5.6. Lysimeter loading variation, 11- to 19-Apr-2007. 173  Cumulative Pile Flow [m3/day]  Lysimeter Loadings [mg/kg/week]  Flow  8  6  DATA INTEGRATION  This chapter combines selected results previously presented in Chapters 2 to 5. Physical and chemical waste rock properties are integrated with the flow, chemistry and loading results in order to obtain a better understanding of the mechanisms that mobilize metals in test pile 1. Particular emphasis is placed on examining evidence of:  1. different flow mechanisms (matrix vs. macropore flow), 2. pyrite oxidation, and 3. spatial variability of weathering rates in the pile.  6.1  EVIDENCE OF DIFFERENT FLOW MECHANISMS  The 4 x 4 m sub-lysimeters are conceptualized as representing effluent water that spatially averages over matrix and macropore (if activated) flow pathways. The same applies to the 36 x 36 m basal lysimeter flow and chemistry, which averages over various flow and chemistry differences that result from the pile batter and the pile interior. Evidence of matrix flow is identified in Chapter 3, and specific conductivity is used to identify evidence of slow and diffuse matrix flow in Chapter 4. The chemical signature of the matrix flow typically contains a higher dissolved load compared to the relatively fresh slugs associated with macropore flow. In response to higher moisture contents and larger infiltration rates due to large precipitation events, specific conductivity measurements decrease by up 50 % over nearly one week after the precipitation event.  Evidence of preferential macropore flow, pressure-wave responses and non-vertical flow in response to large precipitation events near the middle to end of the wet season was discussed in Chapter 3 and is further confirmed using the effluent chemistry in Chapter 4. Sharp fluctuations correspond with large outflow events characteristic of fresh, low conductivity macropore slugs. Furthermore, higher porewater concentrations during the dry season reflect steadier, gradually changing rates representative of matrix flow, which also confirms flow findings from Chapter 4.  174  6.2 PYRITE OXIDATION: ESTIMATING SO4 The total estimated SO4 generated from pyrite weathering is calculated over the course of the water year (1-Oct-2007 to 30-Sept-2008). Assuming that the temperatures at the top and bottom of the pile reflect the temperatures as measured at the shallowest and deepest porewaters, then ΔT = 1.8 °C (Chapter 4) can be used to describe the average temperature difference between the top and bottom of the pile over the course of the water year. In further assuming that the higher temperatures result from a conservative release of energy from the chemical oxidation of pyrite (and this energy did not leave the system through conduction, convection, etc.), then the total amount of energy released due to pyrite oxidation could be calculated based on the following equation: ΔQ = mcΔT  [6-1]  where ΔQ is the heat energy associated or taken out of the system [kJ], m is the mass of the substance [kg], c is the specific heat capacity of water (4.185 kJ/(kg K)), and ΔT is the change in temperature [°K]. Using a total of 6.9 x 105 kg of water (690 m3) estimated to have discharged from the base of the pile (Chapter 4), a total energy of ΔQ = 5.2 x 106 kJ of energy was released during pyrite oxidation over the course of the water year. This value corresponds to an estimated 0.2 W of energy that is continuously released by the entire pile, which is a modest amount given the total volume of the pile (approximately 10,000 m3; Chapter 2). Based on the Gibbs Free Energy released per mol of pyrite mineral oxidation (of ΔGr° = -1440 kJ/mol; Ritchie, 1994), an estimated 12 kJ of energy is released for every gram of pyrite oxidized. Based on the calculated ΔQ, an estimated 430 kg of pyrite was oxidized in order to release that much heat. Using a 1:2 molar ratio between S in pyrite (FeS2; 120 g/mol) and sulphate (SO4; 96 g/mol), an estimated 693 kg of sulphate was associated with the amount of generated heat during the pyrite oxidation described in equation 4-1. Recall that 715 kg of sulphate is reported to have been released from the pile during the same period (Chapter 5).  Even though these two methods produce similar results, they both represent minimum values due to several limitations. The SO4 estimated based on pyrite oxidation is strongly dependent on a perfect mass balance of S between pyrite oxidation and generated SO4. This method also assumes that pyrite is the only sulphide mineral oxidizing and there is no loss of energy via 175  convection, conduction, evaporation, etc. In addition, both methods do not account for the precipitation of gypsum within test pile 1. While it is likely that these limitations may be balanced in both approaches, more quantitative thermal modeling would better constrain these estimates.  6.3 OXYGEN DEPLETION ESTIMATE To estimate the ability of air to travel through the waste rock pile (and confirm whether it is wellaerated), a first-pass approximation of the total amount of oxygen consumed compared to the total estimated amount of oxygen in all of the void spaces. The number of ‘pore volumes’ of oxygen consumed during pyrite oxidation in test pile 1 throughout the study period (18-Jan-07 through 30-Sept-2008) is calculated.  The volume V of oxygen in 1 mol of pore space in test pile 1 is calculated using the Universal Gas Law: 𝑝𝑝𝑝𝑝 = 𝑛𝑛𝑛𝑛𝑛𝑛  [6-2]  where p is the absolute pressure corrected for altitude (57.5 kPa at 4374 m a.s.l.; Appendix D), V is the gas volume (m3), n is the amount of substance of the gas (1 mol), R is the gas constant (8.314472 J K−1mol−1) and T is the absolute temperature, which is the mean annual temperature of 279.15 K (Chapter 2). The volume of gas V in 1 mol of air is calculated at 0.04 m3, and the total volume of oxygen in this volume (at PO2 = 0.21) becomes 5.2 mol O2/m3 of pore gas in the pile. Based on an estimated porosity of n = 0.4, a pile volume of 10,000 m3 (Chapter 2), and an average moisture content of ϴ = 0.22 (Chapter 3), the void volume is calculated to be Vv = 0.18 x 10,000 m3 = 1,800 m3.  The total amount of oxygen in one pore volume of test pile 1 is: 1,800 m3 x 5.2 mol O2/m3 = 9,360 mol O2  176  Based on a total mass of 1343 kg of SO4 released from the pile during the entire study period (Table 5.1), this corresponds to an equivalent molar concentration of O2 (based on equation 4-1) to be 52,460 mol of O2. Comparing the estimated pore volume of oxygen (9,360 mol O2) to the total amount of O2 consumed during the study period (52,460 mol of O2), then roughly 5.5 pore volumes of O2 have moved through the pile. This provides further support indicating that the pile is reasonably wellaerated.  6.4 SPATIAL VARIABILITY IN PILE BEHAVIOUR It has already been established that the loading behaviour of the pile is not spatially uniform (Chapter 5). That is, loading rates have been shown to vary (based on seasonal input) and these rates demonstrate a strong spatial variation throughout the pile.  Based on previous observations that solute loadings are dominated by flow (Chapter 5) and that sub-lysimeter C and basal lysimeter D experience the highest discharge fluxes (Chapter 4), it is no surprise that these two lysimeters experience the highest flow rates. However, it is interesting to note that although sub-lysimeter A experiences lower flow rates that are most similar to sublysimeter B, loading rates at A are almost just as high as C and D (Chapter 5). Chapters 3 (hydrology) and 4 (chemistry) assume that the particle size and material composition is similar across the entire pile (i.e., material properties from each placement stage was relatively homogeneous) and that flow and chemistry differences between lysimeters likely result from the structural differences between the pile interior (and thus the upper, less-permeable, trafficcompacted layer) and pile batter (less-consolidated, more-permeable, coarser-sized material). However, these assumptions are not entirely valid because they do not take into account the mineralogical heterogeneity of the waste rock which must also be considered when comparing lysimeter loading rates.  Although the particle size distribution of material from each placement stage is relatively consistent (Figure 2.10), material from each stage originates from different areas of the open pit and will thus have physical and chemical differences (Table 2.3; Table 2.4). To better understand the mechanisms that lead to the spatial loading variability within the pile, an integration of 177  selected data is required. This involves a comparison between material from Tipping Phases I to III and their respective sub-lysimeters (Figure 4.4). Table 6.1 summarizes the material properties (mass, solid phase concentration, neutralization potential, and relative uniformity), sub-lysimeter attributes (estimated volume of material directly above, total volume of water to have passed through, effluent chemistry, average loading rates) and other data (pile temperature, gas concentrations, etc). Material from the protective layer is discussed first, followed by Tipping Phases I, II and III, and their corresponding sub-lysimeters. In addition, Figure 4.4 is presented again in Figure 6.1 summarizing the ‘big picture’ differences between each material type.  6.4.1  Protective Layer  The protective layer (waste rock) consists of black marble Class B waste rock, which has the lowest average density compared to the rest of the pile (2770 kg/m3; Table 2.6). The weathering behaviour of the protective layer material cannot be deduced because it is mixed with material from the other three tipping phases. Effluent water that drains to any of the four lysimeters will have come into contact with the protective layer as well as one or two other materials. Based on results from the Mineral Liberation Analyzer (MLA; Chapter 2), this material consists of approximately 56 % carbonate, 41 % silicate and 0.5 % sulphide minerals (Table 2.3). This is supported by ICP analyses which indicate that the protective layer material contains the highest reported solid phase Zn concentration (830 ppm) amongst the materials used in the pile (Table 2.4). Compared to the rest of the pile, material from the protective layer contains intermediate masses of Ca, Zn and S (based on solid phase concentration and total mass of the protective layer from Table 2.4), an intermediate net neutralizing potential (NNP; Chapter 2) and the highest neutralization potential ration (NPR) levels (Table 6.1). Figure 6.1 presents an integration of data pertaining to the protective layer.  In general, the particle size distribution of this material is very similar in shape to Tipping Phases I and II and shows that material from all placement stages consists predominantly of coarse particles (Figure 2.10). Approximately 9 % of the material placed in the protective layer has a diameter of 4.75 mm or less and is assumed to represent the proportion of fine material. The median particle size is 250 mm.  178  6.4.2  Tipping Phase I - Sub-lysimeter B  Tipping Phase I consists of marble diopside (Table 2.3). Approximately 7.5 % of this material has a diameter of 4.75 mm or less and is assumed to represent the proportion of fine material. The median particle size is 200 mm. Water passing through Tipping Phase I may also pass through Tipping Phase II near the top of the pile before passing through the protective layer and reporting to the back sub-lysimeter B (Figure 6.1). The contribution of material from this tipping phase to the overall pile mass is low because a significant proportion of it was dumped along the back slope until material was finally placed within the pile footprint (Chapter 2; Table 2.6).  Based on MLA results (Table 2.3), this material consists of approximately 48 % carbonate, 50 % silicate and 2.0 % sulphide minerals. This is supported by ICP analyses which indicate an intermediate solid phase Zn concentration (average) compared to the rest of the pile (Table 2.4), yet this material contributes the least amount of zinc to the pile based on the total mass to reside within the pile (Table 6.1). Interestingly, effluent water at B reports the highest Zn concentrations (Figure 4.8) and likely results due to the low velocity of water passing through this region and a higher porewater residence time (Chapter 3).  A 4 x 4 m vertical column of waste rock, amounting to approximately 326 tonnes, lies directly above sub-lysimeter B. A majority of this material originates from Tipping Phase I. The contribution to entire pile flow from this region is proportionally low (~ 0.02 % of total pile), which may be explained by 3 possible reasons:  1. infiltrating water is attracted to the French drain along the back of the pile – slope contact, 2. non-vertical flow focuses water away from the sub-lysimeter, and/or 3. sub-lysimeter B is located under the most compacted region of the pile (i.e., most trafficcompaction).  The most likely explanation includes the latter option, as a less permeable top layer leads to lower fluid infiltration rates which lead to higher porewater residence times and lower pH values as O2 cannot ingress and CO2 cannot escape as rapidly. 179  Accordingly, water chemistry indicates relatively high average SO4, Zn and Ca concentrations compared to the rest of the pile, with the highest Zn concentrations reported compared to the rest of the pile. However, due to a low flow rate reported at this lysimeter, this region still exhibits the lowest average solute loading rates (Table 6.1). Based on the distance away from the surface, there is evidence that this region behaves more like a partially-closed system (as evidenced by lowest pH values throughout the pile) and thus explains the chemical differences from this region compared to the rest of the pile, particularly those exposed to the surface.  6.4.3  Tipping Phase II - Sub-lysimeter A  Tipping Phase II consists of the most dense (3140 kg/m3) marble diopside found within the pile (Table 2.3). Approximately 4 % of this material has a diameter of 4.75 mm or less and is assumed to represent the proportion of fine material. The median particle size is almost 300 mm (Figure 2.10) and the largest uniformity coefficient (Section 2.13.4; Table 6.1). Based on Table 2.3, material from Tipping Phase II consists of approximately 11 % carbonate, 82 % silicates and 4.5 % sulphide minerals (Aranda, 2009). This is supported by ICP analyses which indicate that that this material contains the lowest solid phase Ca and Zn concentrations, yet the highest solid phase S concentrations (Table 2.4).  Water passing through Tipping Phase II likely passes through Tipping Phase III near the top of the pile before passing through the protective layer and into sub-lysimeter A in the middle of the pile (Figure 6.1). Average SO4 and Ca concentrations in sub-lysimeter A effluent are higher than at any other lysimeter, and Zn levels are intermediate (Chapter 4). This material also has the lowest NPR ratio (Section 2.13.3) indicating that it is the ‘dirtiest’ and has the largest potential for acid generation (i.e., lowest acid buffering potential) and metal release. In addition, the average solid phase Fe concentration of this material is more than twice the amount of Fe reported from the rest of the pile (Table 2.4), which confirms MLA results that nearly 4.5 % of this material consists of sulphide minerals. The primary sulphide minerals are pyrite, chalcopyrite, and pyrrhotite.  Based on depleted O2 concentrations, enriched CO2 concentrations, pH and an average temperature in lines 2 and 3 that are 1.2 °C higher than lines 1 and 4 (Chapter 4), there is 180  reasonable evidence indicating that this is a relative ‘hot spot’ in which elevated rates of chemical oxidation are occurring. Furthermore, that the highest dissolved Ca concentrations are reported in sub-lysimeter A despite the lowest solid phase Ca concentration likely indicates that elevated sulphide oxidation is driving the dissolution of calcite to buffer the generated acid, and that this region may be the first to be depleted of its buffering capacity.  A 4 x 4 m vertical column of waste rock, amounting to approximately 330 tonnes, lies directly above sub-lysimeter A. A majority of this material originates from Tipping Phase II. Despite reporting only 0.5 % of the total pile outflow, the average loading rates at A are nearly half of those calculated at C. Average solute loadings rates are larger than those at sub-lysimeter B by a factor of 1.5 - 2, and indicate that both water flow and solid phase composition (and susceptibility to weathering) are dominant factors affecting loadings at this site.  That this relatively coarse material contains the highest solute loading rates within test pile 1 is contradictory to previous statements that water flowing through a finer material will have a higher dissolved load (based on the surface area). This is largely because of the material properties and initial sulphide mineral content, which causes the mineral weathering rates to be the highest in test pile 1. This is further corroborated by mineralogical work found in Aranda (2009). When these high weathering rates are combined with the intermediate flow rates, this leads to relatively high loading rates compared to the rest of the pile (i.e., similar to sublysimeter C and basal lysimeter D). In addition, since a majority of this material consists of silicate minerals (which are more resistant to weathering than carbonates), this means that the buffering capacity of this material may be depleted much sooner than the buffering capacity in other regions of the pile.  6.4.4  Tipping Phase III - Sub-lysimeter C  Material from Tipping Phase III consists of approximately 68 % marble diopside and 32 % grey hornfels. No particle size distribution test was performed on this material. Water passing through Tipping Phase III passes through the protective layer before reporting to the front sub-lysimeter C which is located under the front batter (Figure 6.1).  181  Based on Table 2.3, material from Tipping Phase III consists of, on average, approximately 74 % carbonate (highest in pile), 24 % silicate (lowest in pile) and 1.9 % sulphide (intermediate) minerals. This is supported by ICP analyses which demonstrate that this material has the highest initial solid phase Ca concentration, with intermediate Zn and S concentrations (Table 2.4). The NPR is the second highest compared to the rest of the pile (Table 6.1).  A 4 x 4 m vertical column of waste rock, amounting to approximately 240 tonnes, lies directly above sub-lysimeter C. Because C is under the batter, not all of the column is filled with waste rock and thus explains why the amount of material above C is approximately 70 % of the amount above sub-lysimeters A and B. All of this material originates from Tipping Phase III.  The total volume of water reported at sub-lysimeter C over the study period is an order of magnitude higher than that reported at sub-lysimeter B and nearly 4 times higher than sublysimeter A. Evidence of non-vertical flow is reported in Chapter 4 and possibly results from inter-bedded fine- and coarse-grained layers that help to channel water to certain locations (likely resulting from the end-dumping method; Chapters 1 and 2). This is the material that makes up the front batter, and is generally coarse and unconsolidated. Water reporting to sub-lysimeter C only has to infiltrate 6 – 9 m (as opposed to the 10 m for the other sub-lysimeters), so C has the fastest response to precipitation events (Chapters 3 and 4).  Solute concentrations in sub-lysimeter C effluent are generally the lowest (Chapter 4) whereas loadings rates are relatively the highest (Chapter 5). The lower concentrations are likely the result of less material, faster infiltration rates and a lower porewater residence time. That carbonate levels are highest in this material despite the lowest dissolved Ca concentrations may be because this material is very well-aerated (as oxygen concentrations are near atmospheric). Since calcite has a higher solubility in open-systems, then calcite dissolution rates may be lower compared to the pile interior. Lower concentrations may be the result of solute dilution based on the high flow rates under the batter. It is possible that the high loading rates result from flow through tipping phase II that reports to sub-lysimeter C, but would imply a net horizontal flow of 5 or more metres, which is largely unlikely.  182  6.4.5  Entire Pile / Basal Lysimeter D  The average elemental composition of the pile is found in Table 2.4. No NNP or NPR values are available for the entire pile, but based on the (linear) relationship between the proportion of total solid phase Ca and NNP [kg CaCO3/tonne], the pile has an estimated NNP value of greater than 50, which means that the pile is most likely ‘non-acid generating’ and has a moderate buffering capacity (Chapter 3). The average pile silicates content is 48.4 %, with 47.5 % carbonate minerals, and 2.2 % sulphide minerals (Table 2.3). The remaining 1.9 % consists of phosphates, oxides, sulphates, and others.  Sub-lysimeter D is conceptualized as representative of the entire pile outflow. Average solute concentrations for basal lysimeter D are generally at intermediate levels within the pile compared to the sub-lysimeters, but generally exhibits high loading rates (Table 6.1). Overall, basal lysimeter D is representative of the entire pile response because it incorporates both highflow/low-concentration and high-concentration/low-flow behaviour of the pile.  6.5 SUMMARY This chapter examined the different flow mechanisms, pyrite oxidation, O2 consumption and spatial variability within the pile. The most salient conclusions are that:  1. slow and diffuse matrix flow in test pile 1 has a higher dissolved load than the fresher, more rapid slugs associated with macropore flow, 2. energy calculations indicate that a significant proportion of heat generated from sulphide mineral oxidation is likely from pyrite oxidation, and the mass of SO4 generated based on the heat released from pyrite oxidation is very similar to the calculated mass of SO4 removed from test pile 1 based on the cumulative loadings, 3. the pile is relatively well-aerated and has consumed an estimated 5.5 pore volumes of oxygen over the 2-year study period, and 4. there is a strong spatial variability of weathering rates in the pile, with evidence of a ‘hot spot’ of sulphide mineral oxidation present in Tipping Phase II (instrumentation lines 2 and 3) and sub-lysimeter A. Since dissolved Ca concentrations are highest in this sub-lysimeter, despite the lowest proportion of carbonate material in this tipping 183  phase, this likely indicates that the buffering capacity of this material may be depleted much earlier than material from other parts of the pile.  In addition, it is evident that the quantification and prediction of mineral weathering rates is complicated. A high solid phase mineral or metal content does not necessarily mean that the effluent water will have a correspondingly high concentration of that mineral or metal. Low solute concentrations will not always correspond with low loading rates. Material susceptibility to physical and chemical weathering, in conjunction with certain attenuation tendencies must also be considered. The characterization and prediction of mass loadings is site-specific, and is highly heterogeneous with loading rates that vary by at least an order of magnitude over a distance of just a few meters within the test pile.  184  TABLES Table 6.1. Integration of waste rock material and lysimeter effluent properties. Material Properties  Uniformit y Coefficien t D60/ D10  Protective Layer Tipping Phase I/B Tipping Phase II/A Tipping Phase III/C  All / 1- D  < 4.7 5 mm  Mass  %  Metric Tonnes  Lysimeters  Estimat ed  29  9  19  7.5  48 N/ A  4.1 N/A  Tot al 5,5 60 2,8 35 5,3 85 5,7 80  33  7.7  19, 560  Ca 1,7 57 75 8 1,0 18 1,9 31 5,4 63  Zn 7. 2 1. 8 2. 4 4. 8 1 6. 2  S 13. 0 8.3 59. 2 27. 8 108 .4  Behaviour NN N P P [kg R CaC O3/t ] 77 9 192 59 7 75 14 7 6 83 35 7 3170  -  -  185  Mass  Flow  Average Concentratio n  tonn es  m3  mg/L SO  -  Average Normalized Loadings Rate  mg/kg/week  Z n  C a  SO  4  4  Zn  Ca  1. 86 1. 09 0. 65  62 8 73 7 38 8  0.1 5 1.2 4 1.1 6  1.9E04 8.9E04 8.4E04  -  1. 35  55 0  0.7 9  8.3E04  330  2.8  326  6.4  240  23.4  13 74 14 50 84 2  19,1 04  1234  12 45  0.07 0.60 0.54  0.36  FIGURE  Figure 6.1. Integration of physical properties and observations throughout the pile.  186  7  CONCLUSIONS  7.1 PROJECT OVERVIEW AND SIGNIFICANCE There is relatively little information about neutral rock drainage at mine sites (Chapter 1). In 2005, a collaboration among the University of British Columbia, Teck’s Applied Research & Technology (ART) group, the Natural Science and Engineering Research Council of Canada (NSERC) and Compañía Minera Antamina S.A. undertook a comprehensive study into the mineralogical, biological, chemical, and hydrological waste rock leaching behaviour at the Antamina mine, Peru (Figure 1.1). Due to the presence of limestone in the region, careful planning of mine waste management must be carried out to avoid far-reaching effects of lowquality drainage conditions.  One component of the Antamina project is the construction, instrumentation and monitoring of a 36 x 36 x 10 m high waste rock test-pile in 2006. The waste rock is classified as non-acid generating (NAG), however mineral weathering leads to the mobilization of several metals and other solutes. No other experiment of this size has ever been constructed to examine neutral waste rock drainage, partly due to the heterogeneous properties of the waste rock material that make it difficult to accurately characterize. In this thesis, the initial two years of data from this experimental waste rock pile were analyzed.  7.2 SUMMARY OF RESULTS The following summarizes the principal conclusions based on the first two years of data collected from this waste rock test pile:  7.2.1  Material Properties  1. Pile design and instrumentation was sufficient to characterize the hydrogeochemical response of the waste rock test-pile under investigation. 2. Test pile 1 is very heterogeneous in particle composition, size fraction and internal structure, and is not currently generating acidic drainage. 3. Based on the solid phase composition of the waste rock (NPR 6 – 3170), neutral drainage conditions can be expected in the future. 187  7.2.2  Hydrology  4. Pile hydrology is spatially and temporally variable: pile flow rates reach as high as 15 m3/day (12 mm/day) and as low as 0.1 m3/day (0.09 mm/day), respectively. 5. The maximum wetting front velocity of vwf = 0.5 m/day is one order of magnitude higher than the maximum porewater velocity of vpw = 0.05 m/day. 6. Some parts of the pile generate drainage throughout the entire year, whereas some parts of the pile stop discharging water during the dry season. The pile batter experiences a disproportionately high flux of outflow water compared to the pile interior. 7. Moisture content rises most rapidly under the batters, and less so near the less permeable, compacted top of the pile. 8. Rapid response to rain events in two of four lysimeters suggests that flow is highly dependent on waste rock heterogeneity, preferential flow paths and the influence of the batter. 9. A reasonable agreement between the water balance components is attained, indicating that the data analysis methodology that was carried out is suitable for the purpose of this investigation. 10. Over the course of the water year, 41 % of precipitation reports as outflow while 59 % evaporates in this humid climate. 11. A majority of outflow originates from matrix flow. 12. The initial wet-up period of test pile 1 took 3 to 4 months. 13. A certain time after a considerable period of rain follows a period of dry conditions, the pile has ‘re-wetted up’ and several flow mechanisms are activated. These mechanisms include: preferential and macropore flow, ‘pistoning’ or pressure-wave displacement and non-vertical flow.  7.2.3  Chemistry  14. Pile chemistry is spatially and temporally variable: effluent water is pH-neutral (pH = 7 8.5), with seasonally varying SO4 (500 – 2000 mg/L) and Zn (0.5 – 2.5 mg/L) concentrations. Tipping Phase II is generally associated with the highest effluent concentrations and Tipping Phase III generally has the lowest. 188  15. Pyrite is the principal source mineral of SO4, sphalerite is the principal source of Zn, and calcite is the principal source of Ca. The presence of SO4 and Ca indicates that sulphide mineral oxidation is occurring and that calcite dissolution is buffering the generated acid. 16. Despite a relative increase in CO2 and a decrease in O2 concentrations at depth, and with 5.5 pore volumes of O2 consumed over the study period, this demonstrates that test pile 1 is reasonably well-aerated. 17. The pile interior acts more like a partially-closed system whereas the pile exterior acts more like an open system. 18. Solute concentrations are generally inversely proportional to flow rate, with highest concentrations at the end of the dry season. 19. The range of porewater concentrations generally increases with depth in lines 1 and 2, and remains relatively stable in line 4. 20. Shallow temperature probes in the pile exhibit seasonal responses to atmospheric temperature variations, whereas deeper probes do not fluctuate as much. Instrumentation lines 2 and 3 are on average 1.2 °C higher than lines 1 and 4. 21. Test pile 1 likely underwent an initial pile flushing of weathering products from the most reactive material during the first year of pile operation. Long-term trends cannot be deduced because 2008 values were different from 2007 (i.e., flow, chemistry, temperatures, etc.), but it is clear that effluent water will contain a high dissolved load many years into the future. 22. There is a nearly proportional positive molar relationship between Ca2+ and SO42, which is consistent with acid neutralization. These results indicate that CO2 is degassing and SO42- concentrations can be used as a relative (not absolute) indicator for pyrite oxidation within the pile. 23. Geochemical modeling indicates that there are no attenuation mechanisms for Zn, and that calcite and gypsum are both at, or near, saturation.  7.2.4  Solute Loadings  24. Solute loadings are spatially and temporally variable. The maximum calculated SO4, Ca and Zn loading rates are 4 – 6, 2 – 3 and 0.003 – 0.005 mg/kg/week, respectively. Loading rates are lowest near the back (interior) of the pile. 189  25. Solute loadings are generally proportional to flow rate and inversely proportion to solute concentrations because of differences in magnitude of flow and concentration ranges. 26. Under extreme events, flow and loading rates increase by more than 50 % over one day whereas SO4 and SC experience less proportional, more attenuated responses. 27. Mineral composition and net infiltration rate are more influential in governing solute mass loadings than solute concentration. 28. Nearly 0.4 % of the total initial solid phase S has been depleted, whereas less than 0.01 % of the total initial solid phase Zn and Ca has been depleted. 29. Loading rates from the test pile are smaller than field cell loading rates by a factor of 2 up to 2 orders of magnitude.  7.2.5  Data Integration  30. Each region of the pile is dominated certain key attributes: a. back (Tipping Phase I – sub-lysimeter B): the back of the pile encounters the lowest permeability based on compaction of heavy machinery during pile construction. This results in the lowest infiltration rates reported by the pile, but relatively high concentrations based on a low porewater residence time. b. middle (Tipping Phase II – sub-lysimeter A): although it has a relatively low permeability due to heavy machinery, this region experiences higher flux rates than towards the back. Compared to the rest of this pile, this material also contains a low proportion of carbonate minerals but high sulphide content. Loading rates are almost just as high as under the batter. c. front (Tipping Phase III – sub-lysimeter C): this material is under the batter, which typically consists of the coarsest, unconsolidated material. Infiltration rates are highest along the batter (because evaporation is the lowest). Although the mineralogy of this material is intermediate (compared to the rest of the pile) and the solute concentrations are the lowest because they are diluted, the rate of solute release and transport is the highest in the pile. High loading rates under the batter likely indicate that mineral weathering rates are higher under a faster infiltration flux (despite the shorter porewater residence time).  190  31. Particle size and/or composition does not always dictate flow behaviour or solute concentrations Higher solute concentrations are typically due to a higher solid phase composition, but may also be due to a lower infiltration rate and high porewater residence time (e.g., higher susceptibility to weathering). 32. Spatial patterns are likely controlled by rock-type heterogeneity and variability in flow path length and residence time that result from the end-dump construction process. 33. Some regions of the pile (particularly Tipping Phase II, Lines 2 and 3, and sub-lysimeter A) may lose their buffering capacity faster than other parts of the pile.  7.3  RELEVANCE TO THE ANTAMINA MINE  The principal physical, hydrological, chemical, and solute mass loading observations summarized above can be used to better understand the mechanisms that lead to metal and solute leaching from the Antamina waste rock. Results are valuable in helping to determine the most cost-effective ways in which to classify and dispose of the waste rock with minimal environmental impact in short- and long-term timeframes and to limit current and future operational and post-closure costs.  The three principal points that the Antamina mine should take from these observations are that:  1.  results indicate that not all Class B material is benign and must be selectively dumped within the Tucush and East Dumps,  2.  there is a need for a re-classification of Class B waste rock into sub-classes, and  3.  the batters have a strong influence on loadings and have implications for the management of current operational piles.  7.3.1  Tucush Dump  This test-pile represents the closest existing simulation of the Tucush Dump and sheds light on uncertainties surrounding Class B waste rock effluent drainage quality. Results from this test pile cannot be directly extrapolated to the Tucush-scale, but can be used to provide a better idea of the expected leaching behaviour of this material. Since some Class B material is capable of producing poor quality drainage, it cannot be assumed that the wetlands at the base of the Tucush 191  Dump will be able to remove a sufficient amount of solutes from the discharging water, requiring additional water treatment considerations. A better approach would be to divert the more reactive Class B waste rock away from the Tucush Dump and send it to the East Dump. This requires a re-classification of the Class B waste rock scheme.  7.3.2  Re-Classification of Class B Waste Rock  Some Class B material has the ability to generate very poor-quality drainage that behaves more like Class A material than Class C. Different grades of Class B material should be identified using more variables than just rock description, rock classification, zinc, arsenic, sulphides and visible oxides (Table 3.2). These parameters should be used in combination with additional aspects such as more comprehensive mineralogical descriptions, lithology and friability. Results should also be combined with other investigative components of this neutral drainage waste rock study (e.g., Aranda (2009), Haupt (in progress)).  7.3.3  Possibility to Reduce Solute Concentrations Based on Dump Modifications  Based on the conclusions summarized above, it is clear that there are strong differences between the batter and the pile interior. As material is added to the existent waste rock dumps, the most reactive material could be placed near the back slope of the pile where non-vertical flow and compaction of the upper layer will help to prevent and re-direct water from infiltrating to depth. Although this region will also have the longest porewater pathways and residence times, the resulting effluent will likely be diluted with water from the rest of the pile when it reaches the base. The least-reactive material should then be placed under the batters where the infiltration rate is the highest, but solute concentrations are the lowest. Based on evidence of non-vertical flow due to the end-dumping procedure, it is likely that some of the pockets of high-reactivity material may be by-passed by infiltrating water which may be routed to less reactive parts of the pile.  Results should also be compared with the cover-study that is underway as part of this project and the test plot vegetation study that is currently conducted along waste rock slopes as an internal investigation by Antamina’s environmental department.  192  7.4  RECOMMENDATIONS FOR CURRENT PILE MAINTENANCE AND FUTURE PILE CONSTRUCTION  7.4.1  Instrumentation  a) always obtain soil-specific calibration (e.g., TDR, ECHO probe) and test all instrumentation before installing, b) replace custom-made EC probes with a more robust, improved design or commercial probes, c) always use two rain gauges (same manufacturer) in case one fails, d) tipping buckets and rain gauge should be connected to their own dedicated datalogger, e) quantify and minimize tipping bucket and rain gauge systematic errors, f) tipping bucket under the batter (sub-lysimeter C) should have a larger capacity that tips at 10 + seconds at the highest flow rates,  7.4.2  Modeling  g) examine other methods to estimate missing precipitation and interpolate flow data to determine how results differ and validity of the RANSAC algorithm and performance, h) determine reliable scale-up procedures for various physical and chemical properties, i) thermal modeling – determine where heat generation is originating from, and whether gas convection is likely occurring within the pile, j) further quantification of sulphide mineral oxidation rates, k) gas modeling – determine transport mechanisms, l) geochemical and reactive-transport modeling (between sub-lysimeters, subsequent piles, etc.), m) examine sensitivities of temperature on solubility, solute mobility, etc.,  7.4.3  Data Management  n) ensure that the water sampling protocol is clear and understood before sampling begins. Maintain consistency (i.e., sampling interval, procedures, priorities, units, QA/QC, etc.) throughout all components of the study: field cells, test piles, cover study, etc., o) better data system management, organization and sharing, 193  p) when a new program is sent to the datalogger, the ‘date implemented’ should be commented at the top, q) detailed maintenance notes should be encouraged,  7.4.4  Further Tests  r) characterize evaporation differences between top of pile and batter, s) characterize infiltration of batters, t) install tensiometers to provide pressure measurements within the pile to better characterize unsaturated flow behaviour, u) isotopic analysis (residence time, etc), v) conduct a tracer test (to determine porewater travel time, extend of preferential pathways, etc.), w) deconstruct pile. There is no current plan for deconstruction, but this should occur to verify findings and provide a more complete dataset that will help in understanding flow and chemistry mechanisms within this pile, x) more lab and field tests to determine: matrix porosity, hydraulic conductivity (bulk and fine-grained), soil water characteristic curve, secondary mineralization, heterogeneity, mineral accessibility, surface area, susceptibility to weathering, texture, etc. of waste rock, y) determine physical properties of ‘lastre’ utilized in the 2B-Protective Layer (e.g., water content) and possible TDR calibration in this material, z) determine mechanisms that lead to secondary mineralization (and identify them if possible) and ways in which to promote these surface coatings in waste rock environments as viable strategies for remediation or prevention of poor drainage quality conditions, aa) mixing tests for material at this pile – different types of Class B material with itself, with Class A, Class C, etc., bb) better and more reliable measurement of (automated) gas samples,  194  7.4.5  Personnel  cc) always have someone on site (UBC or UBC-representative) to deal with instrumentation on site (including datalogger program), dd) always have more than one person supervising construction on site, and ee) better communication between all parties: UBC, Environmental Department, Contractors, Mining Operations, etc. 7.5  CONCLUDING REMARKS  The fate and transport of water and solutes in unsaturated waste rock environments is difficult to quantify and accurately predict. This investigation has shed light on the various mechanisms that contribute to the neutral drainage waste rock environment at the Antamina mine, Peru. Although long-term trends cannot be determined based on the intial two-years of data collected from this test pile, they are useful in characterizing the initial pile flushing and understanding which flow processes are dominant under certain conditions. The pile experiment is continuing and a suite of data is being collected that will be combined and compared to data from four subsequent test piles and additional mineralogical, microbiological, field cell and cover study experiments to obtain a comprehensive characterization of neutral pH drainage from waste rock. The extensive dataset will continue to improve the characterization of the leaching potential of Class B waste rock and waste rock in general, and will contribute in minimizing the short- and long-term environmental impact of poor quality waste rock drainage at the Antamina mine.  195  8  REFERENCES 1.  Allen, R, Pereira, L., Raes, D., and M. Smith. 1998. Crop evapotranspiration: Guidelines for computer crop water requirements. FAO Irrigation and Drainage Paper No. 56, Food and Agricultural Organization of the United Nations, Rome.  2.  Andrina, J., Miller, S. and A. Neale. 2003. The design, construction, instrumentation and performance of a full-scale overburden stockpile for mitigation of acid rock drainage, Grasberg Mine, Papua Province, Indonesia. In: Proceedings of the 6th International Conference on Acid Rock Drainage, Cairns, Queensland, Australia.  3.  Andrina, J., Wilson, W., Miller, S. and A. Neale. 2006. Performance of the acid rock drainage mitigation waste rock trial dump at the Grasberg Mine. In: Proceedings of the 7th International Conference on Acid Rock Drainage, St. Louis, MO, USA.  4.  Antamina, 2003. “The geology and metal zonation of the Antamina deposit”, PowerPoint presentation presented by Lipten, E. and M. Pacheco. Geology Department, Compañía Minera Antamina S.A., Yanacancha, Peru.  5.  Aranda, C., Klein, B., Beckie, R. and U. Mayer. 2009. Assessment of waste rock weathering characteristics at the Antamina mine based on field cells experiment. In: proceedings of the 8th International Conference on Acid Rock Drainage, Skellefteå, Sweden.  6.  Aranda, C. 2009. Assessment of waste rock weathering characteristics at the Antamina mine based on field cells experiment. M.A.Sc. thesis, University of British Columbia, Vancouver, Canada.  7.  AMEC, Resource Modeling – Huaraz, Peru. Project No. 157041. Ch 18.  8.  Azam, S., Wilson, W., Herasymuik, G., Nichol, C. and L. Barbour. 2007. Hydrogeological behaviour of an unsaturated waste rock pile: a case study at the Golden Sunlight Mine, Montana, USA. Bulletin of Engineering Geology and the Environment. 66: 259-268  9.  Aubertin, M., Mbonimpa, M., Bussière, B. and R. Chapuis. 2003. A model to predict the water retention curve from basic geotechnical properties. Canadian Geotechnical Journal. 40(6): 1104-1122  10.  Baver, L., Gardner, W. and W. Gardner. 1972. Soil Physics. 4th Ed. John Wiley & Sons, Inc., Toronto, Canada.  11.  Bay, D. Corazao Gallegos, J.C., Peterson, H., Dockrey, J., Singurindy, O., Beckie, R., Mayer, U., Smith, L., Klein, B. and W. Wilson. 2008. Preliminary geochemical and flow responses of neutral drainage from a field-scale experimental waste rock pile at 196  the Antamina mine, Peru. In: Proceedings of the Water Air Land Sustainability in Mining Symposium, 47th conference of metallurgists, Winnipeg, Manitoba, Canada. 12.  Bay, D., Peterson, H., Singurindy, O., Aranda, C., Dockrey, J., Sifuentes, F., Mayer, U., Smith, L., Klein, B. and R. Beckie. 2009. Assessment of neutral pH drainage from three experimental waste-rock piles at the Antamina mine, Peru. In: Proceedings of the 8th International Conference on Acid Rock Drainage, Skellefteå, Sweden.  13.  Bellehumeur, T. 2001. Mechanisms and spatial variability of rainfall infiltration on the Claude waste rock pile. M.A.Sc. thesis. University of British Columbia, Vancouver, Canada.  14.  Beven, K. and P. Germann. Water flow in soil macropores II. A combined flow model. Journal of Soil Sciences. 32: 15-29.  15.  Blowes, D., Jambor, J. and C. Hanton-Fong, C., Lortie, L. and D. Gould. 1998. Geochemical, mineralogical and microbiological characterization of a sulphide-bearing carbonate-rich gold-mine tailings impoundment, Joutel, Quebec. Applied Geochemistry. 13(6): 687-705.  16.  Blowes, D. and C. Ptacek. 1994. Acid-neutralization mechanisms in inactive mine tailings. In: Jambor, J., Blowes, D. (Eds), Environmental Geochemistry of Sulfide Mine-Wastes, Mineral Association of Canada. Short Course, 22: 271 – 292.  17.  Blowes, D., Ptacek, C., Jambor, J. and C. Weisner. 2003. The geochemistry of acid mine drainage. Treatise on Geochemistry. 9: 149-204.  18.  Bouwer, H. 1986. Intake Rate: Cylinder Infiltrometer. In: Methods of Soil Analysis, Part 1: Physical and Mineralogical Methods – Agronomy Monograph no. 9 (2nd Ed). Madison, WI. American Society of Agronomy – Soil Science Society of America. Chapter 32.  19.  Brett, J. and J. Banfield. 2003. Microbial communities in acid mine drainage. FEMS Microbiology Ecology. 44: 139-152  20.  Brierley, J. and C. Brierley. 2000. Present and future commercial applications of biohydrometallurgy. Hydrometallurgy. 59(2-3): 223-239.  21.  Bussière, B., Aubertin, M. and R. Chapuis. 2003. The behaviour of inclined covers used as oxygen barriers. Canadian Geotechnical Journal. 40(3): 512-535  22.  Buytaert, W., Celleri, R., Willems, P., De Bievre, B. and G. Wyseure. 2006. Spatial and temporal rainfall variability in mountainous areas: A case study from the south Ecuadorian Andes. Journal of Hydrology. 329: 413– 421  197  23.  Carey, S., Barbour, S. and M. Hendry. 2005. Evaporation from a waste-rock surface, Key Lake, Saskatchewan. Canadian Geotechnical Journal. 42: 1189 – 1199.  24.  Carroll, K, Artiola, J. and M. Brusseau. 2006. Transport of Molybdenum in a BiosolidAmended Alkaline Soil. Chemosphere. 65: 778-785.  25.  Ciach, G. 2003. Local random errors in tipping-bucket rain gauge measurements. Journal of Atmospheric and Oceanic Technology. 20: 752-759.  26.  Conlan, M. 2009. Attenuation mechanisms for molybdenum in neutral rock drainage. M.A.Sc. thesis. University of British Columbia, Vancouver, Canada.  27.  Corazao Gallegos, J.C., 2007. The design, construction, instrumentation, and initial response of a field-scale waste rock test pile, M.A.Sc. thesis. University of British Columbia, Vancouver, Canada.  28.  Corazao Gallegos, J.C., D. Bay, R. Beckie, B. Klein, U. Mayer, L. Smith, W. Wilson, S. Brienne and H. Letient, 2007. Design and Construction of Field-Scale Waste Rock Test Piles at The Antamina Mine, Peru. Geotechnical News. 25(1): 49-5.  29.  Dawson, R. and N. Morgenstern. 1995. Liquifaction flowslides in Rocky Mountain coal mine waste dumps, Phase 3, Final Report. Natural Resources Canada, Ottawa, Ontario. Contract report 23440-3-9135/01.  30.  Diodato, D. and R. Parizek. 1994. Unsaturated hydraulic properties of reclaimed coal strip mines. Ground Water 32(1): 108–119.  31.  Evans, D., Letient, H. and T. Aley. 2005. Aquifer Vulnerability mapping in karsitc Terrain. In: The Proceedings of 2005 Society for Mining, Metallurgy, and Exploration (SME) Annual Meeting, Salt Lake City, Utah.  32.  Eriksson, N., Gupta A. and G. Destouni. 1997. Comparative analysis of laboratory and field tracer tests for investigation preferential flow and transport in mining waste rock. Journal of Hydrology. 194: 143-163.  33.  Fala, O., Aubertin, M., Molson, J., Bussiere, B., Wilson, G. and Chapuis, R. 2003.  34.  Numerical modeling of unsaturated flow in uniform and heteregenous waste rock piles. In: Proceedings of the 6th International Conference on Acid Rock Drainage, Cairns, Queensland, Australia. 895-902.  35.  Fala, O., Molson, J., Aubertin, M. and B. Bussière. 2005. Numerical modeling of flow and capillary barrier effects in unsaturated waste rock piles. Mine water and the environment. 24: 172-185.  198  36.  Feasby, G. and G. Tremblay G. 1995. New technologies to reduce environmental liability from acid generating minewastes. In: Proceedings of Sudbury 1995 Mining and the Environment.643-7.  37.  Fines, P., Wilson, W., Williams, D., A. Tran and S. Miller. 2003. Field Characterization of two Full-Scale Waste Rock Piles. In: Proceedings of the 6th International Conference on Acid Rock Drainage, Cairns, Queensland, Australia.  38.  Fischler, M. and R. Bolles. 1981. Random Sample Consensus: A paradigm for model fitting with applications to image analysis and automated cartography. Graphics and Image Processing. 24: 381-395.  39.  Flury, M., Flühler, H., Jury, W. and J. Leuenberger. 1994. Susceptibility of soils to preferential flow of water: A field study. Water Resources Research. 30(7): 1945-1954.  40.  Frostad, S., Klein, B. and Lawrence, R., 2005. Determining the weathering characteristics of a waste rock dump with field tests. International Journal of Mining, Reclamation and Environment. 19(2): 132-143.  41.  Gelinas, P., Lefebvre, R., and Choquette, M., 1992. Monitoring of acid mine drainage in a waste rock dump. In: Environmental issues and Waste Management in Energy and Minerals Production, 747-756.  42.  Gerke, H. and M. van Genuchten. 1993. A Dual-Porosity Model for Simulating the Preferential Movement of Water and Solutes in Structured Porous Media, Water Resource. Research. 29(2): 305–319  43.  Germann, P. and K. Beven. 1981. Water flow in soil macropores I: An experimental approach. Journal of Soil Sciences. 32: 1-13.  44.  Golder Associates, 2006. Waste dump and stockpile water quality estimates – Phase 3 – Antamina Mine, Peru. Final Version 4.0. February, 2006. Golder Associates Perú S.A.  45.  Golder Associates, 2007a. Estudio de Impacto Ambiental - Proyecto de Expansión del Tajo Abierto y Optimización del Procesamiento [Environmental Impact Assessment: Expansion of the open pit and processing optimization], Compañia Minera Antamina S.A. July, 2007. Golder Associates Perú S.A.  46.  Golder Associates, 2007b. Waste Rock and Tailings Geochemistry, Field Cell Monitoring 2002 to 2006, Antamina Mine Peru, Version 4.0, February 2007, Golder Associates Perú S.A.  47.  Herasymuik, G., Wilson, G., Barbour, S. and T. Smith, 1995. The characterization of hydrologic properties and moisture migration pathways of a waste rock pile. Mine reclamation, reclamation in extreme environments. In: 19th British Columbia Ministry of Energy and Mines, Victoria, BC. 199  48.  Hayashi, M. 2004. Temperature-electrical conductivity relation of water for environmental monitoring and geophysical data inversion. Environmental Monitoring and Assessment. 96: 119-128.  49.  ICOLD, 1996. A guide to tailings dams and impoundments: Design, construction, use and rehabilitation. International Commission on Large Dams, Bulletin (United Nations Environment Programme). No. 106: 239  50.  Jambor, J. and D. Blowes. 1998. Theory and application of mineralogy in environmental studies of sulphide bearing mine wastes. In: Modern Approaches to Ore and Environmental Mineralogy. Cabri, L and D. Vaughan. (Eds). Mineralogical Association of Canada, Canada, p. 267-401 (Chapter 12).  51.  Kleinmann, R., Edenborn, H. and R. Hedin. 1991. Biological treatment of mine water—an overview. In: 2nd International Conference on Abatement of Acidic Drainage. MEND Secretariat, Rome. 1: 27–42.  52.  Lawrence, W. and Y. Wang. 1996. Determination of neutralization potential for acid rock drainage prediction. MEND project 1.16.3. Vancouver, Canada.  53.  Lefebvre, R. 1994. Characterization and numerical modeling of acid mine drainage in waste rock dump. Ph.D dissertation, Université Laval, Quebec, Canada.  54.  Lefebvre, R. and P. Gelinas. 1995. Numerical modeling of AMD production in waste rock dumps. In: Sudbury '95, Conference on Mining and the Environment, Sudbury, Ontario. 1: 869 - 878.  55.  Lefebvre, R., Hockley, D., Smolensky, J. and P. Gélinas. 2001. Multiphase transfer process in waste rock piles producing acid mine drainage 1: Conceptual model and system characterization. Journal of Contaminant Hydrology. 52: 137-164.  56.  Linklater, C., Sincalire, D. and P. Brown. 2005. Coupled chemistry and transport modeling of sulphidic waste rock dumps at the Aitik mine site, Sweden. Applied Geochemistry. 20: 275-293  57.  Lipten, E. and S. Smith. 2004. The Geology of the Antamina Copper-Zinc Deposit, Peru, South America; in Porter, T.M. (Ed.), Super Porphyry Copper & Gold Deposits: A Global Perspective; PGC Publishing, Adelaide.  58.  Love, D., Clark, A. and K. Glover. 2004. The lithologic, stratigraphic and structural setting of the giant Antamina copper-zinc skarn deposit, Ancash, Peru. Economic Geology. 99: 887-916  200  59.  Machibroda, R. 1994. M.Sc. thesis: Evaluation of evaporative fluxes from mint tailings using the modified Penman formulation. Placer Dome Cover and waste Rock Research program. University of Saskatchewan, Saskatoon, Canada.  60.  Malmström, M., Destouni, G, S. Banwart and B. Strömberg. 2000. Resolving the sacledependence of mineral weathering rates. Enviornmental Science and Technology 34(7): 1375-1378.  61.  Marcoline, J. 2008. Investigations of water and tracer movement in covered and uncovered unsaturated waste rock. Ph.D. dissertation. University of British Columbia, Vancouver, Canada.  62.  Mayer, K., Benner, S., Blowes, D., and E. Frind, 1999. The reactive transport model MIN3P: Application to acid mine drainage generation and treatment. Nickel Rim Mine Site, Sudbury, Ontario. In: Sudbury '99, Conference on Mining and the Environment, Sudbury, Ontario. 1: 145-154.  63.  Mbonimpa, M., Aubertin, M., Aachib, M. and B. Bussière. 2003. Diffusion and consumption of oxygen in unsaturated cover materials. Canadian Geotechnical Journal. 40(5): 916-932.  64.  Milczarek, M., Hammermeister, D., Buchanan, M., Vorwaller, B. and T. Conner. 2009. In-situ monitoring of a closed waste rock facility. In: Proceedings of the 8th International Conference on Acid Rock Drainage, Skellefteå, Sweden.  65.  Moncur, M., Ptacek, C. Blowes, D. and J. Jambor. 2005. Release, transport and attenuation of metals from an old tailings impoundment. Applied Geochemistry. 20: 639-659  66.  Morin, K., Gerencher, E., Jones, C., and D. Konasewich. 1991. Critical Literature Review of Acid Drainage from Waste Rock. Canadian MEND Report 1.11.1, CANMET, Government of Canada, Ottawa  67.  Natural Resources Canada, 2004. Design, construction and performance monitoring of cover systems for waste rock and tailings. Mine Environment Neutral Drainage (MEND) 2.21.4. (ed. O’Kane Consultants Inc.). OKC Report No. 702-01.  68.  Nichol, C. 2002. Transient flow and transport in unsaturated heterogeneous media: Field experiments in mine waste rock. Ph.D. dissertation. University of British Columbia, Vancouver, Canada.  69.  Nichol, C., Beckie, R. and L. Smith. 2002. Evaluation of uncoated and coated time domain reflectometry probes for high electrical conductivity systems. Soil Science Society of America Journal. 66: 1454-1465.  201  70.  Nichol, C., Smith, L and R. Beckie. 2003. Time domain reflectometry measurements of water content in coarse waste rock. Canadian Geotechnical Journal. 40: 137-148.  71.  Nichol, C., Smith, L and R. Beckie. 2005. Field-scale experiments of unsaturated flow and solute transport in a heterogeneous porous medium. Water Resources Research. 41:W05018, doi:10.1029/2004WR003035.  72.  Neuner, M. 2009. Water flow through unsaturated mine waste rock in a region of permafrost. M.Sc. thesis, University of British Columbia, Vancouver, Canada.  73.  Neuner, M., Gupton, M., Smith, L., Pham, N., Smith, L., Blowes, D. and D. Sego. 2009. Diavik waste rock project: Unsaturated water flow. In: proceedings of the 8th International Conference on Acid Rock Drainage, Skellefteå, Sweden.  74.  Newman, L. 1999. Preferential flow in vertically oriented unsaturated soil layers. M.Sc. thesis. University of Saskatchewan, Saskatoon, Canada.  75.  Noborio, K. 2001. Measurement of soil water content and electrical conductivity by time domain reflectometry: a review. Computers and Electronics in Agriculture. 31: 213-237.  76.  Olson, G. and T. Clark. 2008. Bioleaching of molybdenite. Hydrometallurgy. 93(1-2): 10-15.  77.  Otwinowski, M. 1995. Scaling analysis of acid rock drainage. Canadian MEND report 1.19.7, CANMET, Government of Canada, Ottawa.  78.  Parkhurst, D.L., Appelo, C.A.J. and Geological Survey (U.S.), 1999. User's guide to PHREEQC, Version 2: a computer program for speciation, batch-reaction, onedimensional transport, and inverse geochemical calculations. Water-resources investigations report. U.S. Geological Survey: Earth Science Information Center, OpenFile Reports Section, Denver, Colorado, xiv: 312.  79.  Pantelis, G. and Ritchie, A. 1992. Rate-limiting factors in dump leaching of pyritic cores. Applied mathematical modeling. 16: 553-560.  80.  Penman, H. 1948. Natural evapotranspiration from open water, bare soil and grass. In: Proceedings of the Royal Society of London Series A. 193:120-145.  81.  Petrunic, B., Al, T. and L. Weaver. 2005. A transmission electron microscopy analysis of secondary minerals formed in tungsten-mine tailings with an emphasis on arsenopyrite oxidation. Applied Geochemistry. 21: 1259-1273.  82.  Price, W.A., 1997. Guidelines and Recommended Methods for the Prediction of Metal Leaching and Acid Rock Drainage at Mine sites in British Columbia - Draft, Smithers, BC, Canada 202  83.  Price, W. 2003. The mitigation of acid rock drainage: Four case studies from British Columbia. Report prepared for CANMET, Natural Resources Canada, Smithers, B.C.  84.  Rasmussen, T., Baldwin, Jr., R., Dowd, J. and A. Williams. 2000. Tracer vs. pressure wave velocities through unsaturated saprolite. Soil Science Society of America Journal. 64: 75-85.  85.  Redwood, D. 1999. The Geology of the Antamina Copper-Zinc Skarn Deposit, Peru. The Gangue, Geological Association of Canada Mineral Deposits Division Newsletter, January. 1999. 60: 1-7.  86.  Rhode, T. and D. Williams. 2009. Early hydrological monitoring of Cadia’s instrumented trial waste rock dump. In: proceedings of the 8th International Conference on Acid Rock Drainage, Skellefteå, Sweden.  87.  Richards, L. 1931. Capillary conduction through porous mediums. Physics. 1: 313-318.  88.  Ritchie, A. 1994. Sulphide oxidation mechanisms. Controls and rates of oxygen transport. In: The Environmental Geochemistry of Sulphide Mine-wastes (eds. J. Jambor and D. Blowes). Mineralogical Association of Canada, Nepean, Ontario. 22: 201-246.  89.  Saretzky, G. 1998. M.A.Sc. thesis: Hydrological characterization of a sulphide waste rock dump. Department of Civil Engineering. University of Saskatchewan, Saskatoon, Canada.  90.  Sherlock, E., R. Lawrence and R. Poulin, 1995. On the neutralization of acid rock drainage by carbonate and silicate minerals. Environmental Geology. 25: 43-54.  91.  Sobek, A., Schuller, W., Freeman, J. and R. Smith. 1978. Field and laboratory methods applicable to overburdens and minesoils, EPA-600/2-78-054. West Virginia Geological and economic Survey, West.  92.  Soil Cover, 2000. Soil Cover User’s Manual. Unsaturated Soils Group, Department of Civil Engineering, University of Saskatchewan, Saskatoon, Canada.  93.  Sracek, O., Choquette, M., Gélinas, P., Lefebvre, R. and R. Nicholson, 2004. Geochemical characterization of acid mine drainage from a waste rock pile, Mine Doyon, Québec, Canada. Journal of Contaminant Hydrology. 69: 45-71.  94.  Smith, L. and R. Beckie. 2003. Hydrologic and geochemical transport processes in mine waste rock, Environmental Aspects of Mine Wastes, Ed. J. Jambor, D. Blowes, and A.I.M. Ritchie, Short Course Series v. 31, Ed. R. Raeside, Mineralogical Association of Canada.  203  95.  Stockwell, J., Smith, L., Jambor, J.L., and R. Beckie. 2006. The relationship between fluid flow and mineral weathering in heterogeneous unsaturated porous media: a physical and geochemical characterization of a waste rock pile, Applied Geochemistry. 21: 1347-1361.  96.  Stollenwerk, K. 1998. Molybdate transport in a chemically complex aquifer: field measurements compared with solute-transport model predictions. Water Resources Research 34: 2727–2740.  97.  Strömerg, B. and S. Banwart. 1999. Experimental study of acidity-consuming processes in mining waste rock: some influences of mineralogy and particle size. Applied Geochemistry. 14: 1–16.  98.  Swanson, D., Barbour, S., Wilson, G. and M. O’Kane. 2003. Soil-atmosphere modeling of an engineered soil cover for acid generating mine waste in a humid, alpine climate. Canadian Geotechnical Journal. 40: 276-292.  99.  Tuller, M. and D. Or. 2002. Unsaturated hydraulic conductivity of structured porous media: A review of liquid configuration-based models. Vadose Zone Journal. 1:14-37.  100. USDA, 1993. Acid drainage from mines on the National Forest: a management challenge. Forest Service Publication. 1505: 1–12. 101. Wagner, K. 2004. Characterization of the geochemistry of discharge waters, pore waters, primary and secondary minerals of an experimental waste rock pile, Cluff Lake mine, Saskatchewan, Canada. M.Sc. thesis. University of British Columbia, Vancouver, Canada. 102. Wagner, K., Smith, L. and R. Beckie. 2006. Hydrogeochemical characterization of effluent from mine waste rock, Cluff Lake, Saskatchewan. In: Proceedings of the 7th International Conference on Acid Rock Drainage, St. Louis, MO, USA. 103. Webb, G., Tyler, S., Collord, J., Van Zyl, D., Halihan, T., Turrentine, J. and T. Fenstemaker. 2008. Field-scale analysis of flow mechanisms in highly heterogeneous mining media. Vadose Zone Journal. 7: 899-908. 104. Wels, C., Lefebre, R. and A. Robertson. 2003. An overview of prediction and control of air flow in acid-generating waste rock dumps. In: Proceeding of the 6th International Conference on Acid Rock Drainage, Cairns, Queensland, Australia. 105. Wilson, W., Fredlund, D. and S. Barbour. 1994. Coupled soil-atmosphere modeling for soil evaporation. Canadian Geotechnical Journal. 31(2): 151-161. 106. Wilson, W., Fredlund, D. and S. Barbour. 1997. The effect of soil suction on evaporative fluxes from soil surfaces. Canadian Geotechnical Journal. 34:145-155.  204  107. Wunderly, M., Blowes, D., Friend, E. and C. Ptacek. 1996. Sulphide mineral oxidation and subsequent reactive transport of oxidation products in mine tailings impoundments: A numerical model. Water Resources Research. 32(10): 3173-3187. 108. Yazdani, J. 1995. Soil water characteristic curve for mine waste rock containing coarse material. M.Eng. Thesis, University of Saskatchewan, Saskatoon, Canada. 109. Yazdani, J., Barbour, L, and W. Wilson. 2000. Soil water characteristic curve for mine waste rock containing coarse material, from Proceedings of the 6th Environmental Engineering Specialty Conference of the Canadian Society of Civil Engineers, London, Ontario, Canada.  205  APPENDIX A:  PILE DESIGN AND CONSTRUCTION  Figure A.1. Pile base and instrumentation lines 5 and 6. .......................................................................... 207 Figure A.2. Construction of pile base. ...................................................................................................... 208 Figure A.3. Tipping Phase I. ..................................................................................................................... 209 Figure A.4. Instrumentation Line 1........................................................................................................... 210 Figure A.5. Tipping Phase II. .................................................................................................................... 211 Figure A.6. Instrumentation line 2. ........................................................................................................... 212 Figure A.7. Instrumentation line 3. ........................................................................................................... 213 Figure A.8. Instrumentation line 4. ........................................................................................................... 214 Figure A.9. Datalogging system. .............................................................................................................. 215 Figure A.10. Top of test pile 1. ................................................................................................................. 216  206  APPENDIX A:  PILE DESIGN AND CONSTRUCTION  N  Plan View of Test Pile 1  Figure A.1. Pile base and instrumentation lines 5 and 6.  207  Sub-Lysimeter Lower Corner  2B-Protective Layer Upper Corner  Drainage Pipe  Protected Drainage Pipe  * view from north side  Figure A.2. Construction of pile base.  208  Lysimeter Berm  Tipping Phase I  Protective Layer (waste rock)  * view from north side  Figure A.3. Tipping Phase I. 209  Cross Section  Protective Layer (waste rock) 2B-Protective Layer  South Side  Front View  Protective Layer (waste rock) 2B-Protective Layer  East Side  * view from south and east sides, respectively  Figure A.4. Instrumentation Line 1 210  Tipping Phase II  Protective Layer (waste rock)  * view from north side  Figure A.5. Tipping Phase II. 211  Cross Section  Protective Layer (waste rock) 2B-Protective Layer  South Side  Front View  Protective Layer (waste rock) 2B-Protective Layer  East Side * view from south and east sides, respectively  Figure A.6. Instrumentation line 2. 212  Cross Section  Protective Layer (waste rock) 2B-Protective Layer  South Side  Front View  Protective Layer (waste rock) 2B-Protective Layer  East Side  * view from south and east sides, respectively  Figure A.7. Instrumentation line 3. 213  Cross Section  Protective Layer (waste rock) 2B-Protective Layer  South Side  Front View Protective Layer (waste rock) 2B-Protective Layer  East Side * view from south and east sides, respectively  Figure A.8. Instrumentation line 4. 214  Figure A.9. Datalogging system. 215  N  Tipping Phase III  Tipping Phase II  Tipping Phase I Figure A.10. Top of test pile 1. 216  APPENDIX B: B.1 B.2 B.3 B.4  PILE 2 & 3 CONSTRUCTION MODIFICATIONS FROM PILE 1  BASE ........................................................................................................................... 218 END-DUMPING AND INSTRUMENTATION ........................................................ 220 INSTRUMENTATION HUT ...................................................................................... 222 REFERENCES……………………………………………………………………….225  217  APPENDIX B:  PILE 2 & 3 CONSTRUCTION MODIFICATIONS FROM PILE 1  This appendix summarizes modifications in the original test pile 1 (2006) design (Corazao Gallegos, 2007) that were implemented during the construction of test piles 2 and 3 (2007). Whenever possible, consistency in design was maintained between the piles. This design was also closely followed during the construction of Piles 4 and 5 (2008). Departures from the original pile design are highlighted (bold) in this section. Recommendations for test pile 4 and 5 modifications are also presented (italics). Test piles 2 and 3 were built almost exactly the same, and will be referred to throughout this section as a single test pile unless otherwise specified.  June 2008 (revisions May 2009)  B.1 BASE This section describes the construction of the pile foundation and instrumentation. Principal components include the liner and instrumentation lines 5 and 6. a) A waste rock foundation pad was compacted with a slope of 3 % towards the south and east. A 0.20 m layer of lastre was placed on top of the foundation pad and subsequently compacted. b) The outer berm was placed in 3 separate layers, each separately compacted to a final height of 0.80 m. Upon verification of the distance between berm apex to apex (N-S and E-W), it was clear that a mistake in the 36 x 36 m capture basin perimeter was not obtained. Modifications resulted in slight departures from the original berm design, but the resulting dimensions of the basal lysimeter remained 36 x 36 m. c) A ditch was dug for the drainage pipes from the sub-lysimeters to the front of the pile. Protective tubes for instrumentation lines 1 and 2 were also placed in this trench. A possible modification would be to place all protective tubes in their respective places on the slope after the tipping. This will avoid spending time sealing the geomembrane around the protruding pipes and the trench, will require less time protecting the pipes with material before dumping and subsequently excavating the pipe for instrumentation. It would also be easier to pass the instrumentation lines through the protective pipes. Further discussion weighing the advantages and disadvantages is recommended. 218  d) The drainage pipe for the large lysimeter was placed at the lowest (southeast) corner of the pile, but was modified to leave from the front of the pile (eastward) so as to reduce the distance that water travels to the instrumentation hut and to maintain an appropriate gradient. This pipe was placed within the slope (e.g., between the 4373and 4371-m platforms; Figure 2.5) u