Proceedings of Tailings and Mine Waste 2023

Citation preview

Proceedings of

Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, BC, Canada

Editor Joe Goodwill, Vancouver, BC, Canada

ORGANIZED BY:

Disclaimer Any views and opinions expressed in the articles published in these proceedings are solely those of the authors and do not necessarily represent those of the University of British Columbia (UBC), C3 Alliance Corp. (C3) or of the panel of editors. The authors take full and exclusive responsibility for technical content and accuracy of the information published herein. This information is not intended nor implied to be a substitute for professional advice. UBC, C3, and the panel of editors are not responsible for any damage to property or persons that may occur as a result of the use of the information contained in this volume.

ii

CONTENTS Organizers .................................................................................................................................... xiv Committees ................................................................................................................................... xv Sponsors ..................................................................................................................................... xvii Exhibitors .................................................................................................................................... xix Foreword........................................................................................................................................ xx Chapter 1: Analysis ..................................................................................................................... 1 Seismic Evaluation of a Tailings Dam Using Uncoupled and Fully Coupled Soil Constitutive Models │ Zana Karimi, Pooya A. Sheykhloo, Lisa Yenne, Ron Hickman, Sam Saunders, Masood H. Kafash and Amin Gheibi ...................................................................................... 3 Seismic Deformation Assessment of Tailings Dams in Subduction Zones │ Jorge Macedo, Jonathan Bray and Chenying Liu .......................................................................................... 15 Numerical Analysis for the Stress-Strain Evaluation of the Conversion of a Conventional Slurry Tailings Storage Facility to Filtered Tailings Storage Facility, Considering Different Scenarios │ Carlos Omar Vargas-Moreno, Jorge Bricio Guillén-Guillén, Rigoberto Ramírez-Chávez, Humberto F. Preciado, Eduardo Ang and Daniel Servigna ...................... 29 Adapting Hoek-Brown Failure Criterion to Rockfill Material and Its Use in 3D Analysis of a Waste Dump Facility │ Ömer Ardıç and Aykut Ayderman .............................................. 43 Deformation Modelling of an Upstream Raised Tailings Dam │ Ignacio A. Cueto, Osvaldo N. Ledesma, Raul Norambuena Mardones and Noe Guerra Acosta ..................................... 57 Calibration of Modified Cam-Clay Parameters for Red Mud Tailings – A Case Study │ Jessé Joabe Vieira Carneiro, André de Oliveira Faria and Mauro Pio dos Santos Júnior ..... 69 An Energy-Based Analysis of the Feijão Tailings Dam Failure │ Kaitlyn O’Sullivan and Abouzar Sadrekarimi ............................................................................................................. 81 Evolution of the Geotechnical and Phreatic Conditions in Tailings of an Upstream-Raised Dam │ Anton Novikov, Raul Norambuena and Ignacio Cueto ............................................ 93 Coupling Tailings Deposition Modelling with Hydrus-1D and Python for Pseudo-3D Seepage Predictions │ M. Noël, I.J. Vega, F. Lopez Rivarola, I. Ezama, M. McGregor and A.G. Terlisky ....................................................................................................................... 105 Effects of State-Dilatancy Measurement Uncertainty on Some Analyses of Tailings Storage Facilities │ Marcus Gunnteg, David Reid, Andy Fourie, Kyle Smith, Mason Ghafghazi, Riccardo Fanni, Simon Dickinson and Juan Garfias ........................................................... 117 Preliminary Analysis of the Failure Process of a Tailings Dam │ Jiarui Chen, Alfonso Cerna-Diaz and Scott M. Olson .......................................................................................... 131

iii

Three-Dimensional Seepage Simulation of a Tailings Storage Facility: A Case Study │ Zana Karimi, John Redmond, Masood H. Kafash, Kanyembo Katapa, Richard Davidson and Hadleigh Tyler .............................................................................................................. 143 Influence of the Suspension Chemical Property on the Differential Settling Behaviour of Copper Tailings Suspension │ Yuan Li, Manthan Varia and Dirk van Zyl ....................... 155 Chapter 2: Application of GISTM ......................................................................................... 167 Utilized Approaches in the Application of GISTM to Legacy Tailing Facilities │ Matthew S. Gore and Ali Nasseri-Moghaddam .................................................................. 169 Deviance Accountability – A Register-Based Approach │ Martyn Bryan Willan, Jennifer Brash and Andy Haynes ....................................................................................................... 181 GISTM – Defining “Substantial” Conformance: Gold Fields’ Substantial Implementation of the GISTM at the Cerro Corona Mine in Peru and Tarkwa Mine in Ghana │ Louise McNab, Johan Boshoff, Javier Gutierrez, Edwin Zegarra, Diana Honores, Nathaniel Asifu Mensa, Fernando Rodriguez and Josh Rogers .................................................................................. 189 Engineers of Record, Professional Registration, and the Mining Industry in Mexico │ Jesus E. Romero .................................................................................................................. 201 Effective EoR Succession Planning Recommendations for Implementation of GISTM │ Madeline R. Sova, Hülya Salihoğlu Ertürk and Christopher N. Hatton ............................. 213 Chapter 3: Breach and Inundation Estimates ...................................................................... 225 Waste Dump Failure Runout Analyses: Applying Improved Empirical Correlation Methods to Waste Dump Datasets │ Trevor White, Andrew Mitchell, John Whittall and Scott McDougall ................................................................................................................. 227 Physical and Numerical Modelling of Tailings Dam Breach Processes (CanBreach Project) │ Megan McKellar, Andrea Walsh, Kayleigh Barlow, Ryan P. Mulligan, Scott McDougall, Stephen G. Evans and W. Andy Take ................................................................................. 239 Methods for Empirical and Numerical Analysis of Tailings Flow Runout (CanBreach Project) │ Negar Ghahramani, Daniel Adria, Sally Innis, Nahyan M. Rana, Evelyn Dina, Nadja Kunz, Stephen G. Evans, W. Andy Take and Scott McDougall ......................................... 251 Limitations and Possible Improvement of Dam Breach Studies │ Holly Williams, Daryl Hockley, Chad LePoudre and Michal Kozikowski ............................................................. 263 Improvements in Case History Knowledge for Tailings Dam Failures by Statistical and Remote Sensing Methods (CanBreach Project) │ Nahyan M. Rana, Negar Ghahramani, Andy Small, Scott McDougall, W. Andy Take and Stephen G. Evans .............................. 275 Comparison of Methods in Estimating the Released Volume in a Tailings Dam Break Analysis │ Jeymy Huamanyauri, Jhoel Huanchi, Jose Hinostroza and Miguel Huamán ... 287

iv

Multi-model Analyses of Physical Simulations of Tailings Impoundment Failure and Runout: Part II │ Armin Saeedi, Zahra Esfandiari, Xuexin Xia, Elena Zabolotnii, Mario Martinelli, Nicholas A. Beier, Gonzalo Zambrano and Paul Simms .................................................... 297 Tailings Dam Break Analysis Performed for the PFS Study of a Downstream Raised TSF in South America │ Holman Tellez, Miguel Medina and Abdolvahid Mohammadian ......... 309 Chapter 4: Case Histories ....................................................................................................... 325 Jet Grouting Field Trial Program in Soft Tailings │ Mauricio Pinheiro, Richard Dawson, Paolo Gazzarrini, Rogerio Torres, Jessica Miranda, Anelisa Vasconcelos and Akira Koshima .. 327 Tailings Storage Facility Risk Reduction by Use of Integrated Mine Waste Management │ Rob Longey, Myra Bacalzo and Glenn Duncan .................................................................. 339 TMF Extension at the Björkdal Mine: Dam Foundation Challenges │ Annika Bjelkevik, Mikael Stenberg, Marcus Vestman, Dennis Olofsson, Lena Printzell, Tobias Holmström and Viktor Wiklund ............................................................................................................. 349 Tailings Stockpile Runout: The Case of Xingu, Brazil │ Marcos Túlio Fernandes, Pedro de Caralho Thá, Willyan Giorgio Debastiani, Daniel de Oliveira Dourado, Felipe Jorge Teixeira, Leonardo Corradi Coelho, Zandra Almeira da Cunha and Gabriel Henrique Calais ...................................................................................................... 361 Case Study: Decharacterization of Tailings Dike in the Iron Quadrangle, Minas Gerais, Brazil │ Thatyane Martins Gonçalves, Ricardo Cabette Ramos, Frank M.S. Pereira, Gino Calderon Vizcarra and Alessandro Lage de Castro ............................................................................ 375 From Design to Construction. The Role of the EoR in the Construction of a Filtered TSF Located in a Remote Area │ Camilo Morales, Esteban Barría, Francisco Vera and Johan Boshoff ...................................................................................................................... 387 Rockfill Stockpile Construction Over Loose Tailings │ Michael Etezad, Ali El Takch, Ken Bocking, Monica Ansah-Sam and Renee White ......................................................... 399 Transition to Filtered Tailings at LaRonde Gold Mine │ Edouard Masengo, Edouardine-Pascale Ingabire, Sylvain Boily, Yanick Létourneau, Patrick Laporte, Francis Guay, Marielle Limoges Shaigetz, Jessica Huza and Michel R. Julien ....................................................... 413 The Development of LKAB’s Tailings Facilities, from the Beginning and into the Future │ Sara Töyrä, Dan Lundell, Roger Knutsson and Thomas Bohlin ......................................... 427 Evolution of Tailings Storage at the Campbell Mine over the Past 40 Years │ Desiree Wilkins and Tara Rothrock ............................................................................................................... 439 Modelling of Static Liquefaction of Cadia Failure with Material Point Method │ Pepe Aynaya, Fabricio Fernández and Raquel Quadros Velloso ................................................ 451 Case Study: Geotechnical Studies to Improve Understanding of Material Parameters and Address Changes in Stability Requirements, Granny Smith Gold Mine, Western Australia │ Louise McNab, Johan Boshoff and Calvin Wang ........................................... 461

v

Back Analysis of Runout Mechanism of a Static Liquefaction Induced Failure Using the Material Point Method │ Masood H. Kafash, Alfonso Cerna-Diaz, Lisa Yenne and Richard Davidson ................................................................................................................ 467 Evaluation of Flow Liquefaction Susceptibility of Uranium Tailings from CPTu │ Stefhany Melendez and Brahian Roman ............................................................................................ 481 Downstream Improvements on Upstream Dams: An Owner’s Perspective of a Paradigm Shift │ Christopher R. Winter ............................................................................................. 489 Chapter 5: Closure Issues ....................................................................................................... 501 Boodarie Tailings Storage Facility – Our Journey to Closure │ Nathalia Revelo Mendez, Daniel Meunier, Darren Springer, Bria Whiter and Simon Harris ...................................... 503 Lessons Learned from Closure Design and Post-Closure Performance of a Uranium Tailings Storage Facility │ Clint Strachan and Kevin Raabe ........................................................... 515 Long-term Evolution of the Phreatic Surface in a Tailings Dam following Closure │ Nicholas A. Beier, Haley L. Schafer and Renato Macciotta .............................................................. 527 Beta and Skeeter Lake Tailings Dams: Progressive Dam Safety Review Approach for the Abandoned and Remote Bullmoose Mine in NWT │ Kris Hojka, Steve Bundrock and Caitlin Moore ...................................................................................................................... 539 Applications of Geomorphology in Mine Reclamation │ Kenneth Myers ................................ 551 Bringing Recent Expertise from Fly Ash Tailings Pond Closures to Mine Tailings │ Paul Schmall and Dale Evans ...................................................................................................... 561 Circular Economy: Usage of Iron-Nickel Slag as Construction Material for Rockfill Dikes │ Simone Sousa, João Paulo Rodrigues, Henrique Guerzoni, Juliano Ferreira and Carlos Lima ......................................................................................................................... 575 Path to Safe Closure – A Case Study and Lessons Learned │ Sam Abbaszadeh, Clint Strachan, Cassandra Hall, Steven Siemoneit and Tamara Johndrow ................................. 587 Chapter 6: Design .................................................................................................................... 599 The Integration of Geological and Hydrogeological Models to Support Design and Operations of a Tailings Facility │ Madeleine Sauvé, Brienna Shaw, Shelby DeMars and Daryl Dufault ................................................................................................................................. 601 Design Elements and Closure Considerations of a Tailings Dam Separated from an Upstream Lake │ Tanya Walkenbach, Craig Schuettpelz and Kimberly Vander Vis ........................ 613 Integrated Waste Storage Facility Design: Considerations for Co-disposal of Waste Rock and Tailings for a New Mine │ Tyler Dixon, Ajitha Wanninayake and Milan Thiaga ............. 627 Engineering Geological Models and a Digital Modelling Solution for Tailings Management │ Tristan Jónsson Menzies and Yusuf Simjee ........................................................................ 641

vi

Transitioning from Upstream Raising to Downstream Raising on Two TSFs at the Tarkwa Open-pit Mine in Ghana │ Johan Boshoff, Louise McNab, Nathaniel Asifu Mensah and Sifiso Dlamini ............................................................................................................... 653 Alternate Approach for the Development of Dry Stack Tailings Facilities │ Luciano Piciacchia and Antonio Vides ............................................................................................................... 667 Seismic Liquefaction Potential Assessment by Artificial Neural Networks │ Marjan Oboudi and Rafael Dávila ................................................................................................................ 679 Soil Arching as a Means to Resist Internal Erosion under High Seepage Gradients │ Isaac A. Jeldes, Nicholas R. Brink, Matthew J. Pauly, Efrain A. Rondinel and Jesus H. Seda ... 689 Extending the Habitat of Viper Filtration: Implementing the Technology in the Americas │ Oliver Whatnall, Ricardo Santander, Stuart Wilson and Sam Caldwell ............................. 703 Operational Liquor Management, Drainage Improvements and Influence on Buttress Design │ Dale Hone ............................................................................................................ 715 Coke as a Filter in Oil Sands Mining Dams │ Pathma Wedage, Christopher Fortier, Anthony Burnett and Weidong Li ...................................................................................................... 727 Chapter 7: Environment and Water Issues .......................................................................... 743 Characterization of Potentially Acid-Generating Tailings for Filtered Tailings Storage Design │ Allison Surrette, Gideon Lambiv Dzemua, Hendrik Falck, Karsyn Beatty, Charlotte Gibson and Heather Jamieson ............................................................................................. 745 Beyond Foundation Seepage: Hydrogeology, Seepage and GISTM │ Jonathan Keizer ........... 759 Method for Evaluating Impacts of Climate Change on Snowmelt and Rain-on-Snow Events │ Stephen Clark and Nicole Whitmore ................................................................................... 769 Selection of Climate Change-Informed Design Storm Events for a Tailings Storage Facility at Red Dog Operations, Alaska, USA │ Bridget Eckhardt, Tenaya Brown and Tyler Oester ................................................................................................................................... 781 Leveraging Exploration Assay and Environmental Data for Mine Waste Planning: Developing Geochemical Block Models for Environmentally Complex Mine Waste │ Kristin Salzsauler, Tom Meuzelaar, Scott Davidson, Emily Sportsman, Ross Hammett, Greg Warren, Brent Murphy and Elizabeth Miller .............................................................................................. 793 Sediment Yields at Mine Sites: Case Study of Vale’s Carajás Iron Mine, Brazil │ Adolfo Correa, Jose Vasquez, Vinícius Sucupira, Mardon Mendes, and Deni Souza ..................... 807 Innovative Passive Treatment of Selenium in Mine-Influenced Brine Using Solar-Powered Buoyant Photocatalysts │ Aldrich Ngan, Aaron Bleasdale-Pollowy, Tim Leshuk and Frank Gu .............................................................................................................................. 819 A Review of Microbial Induced Carbonate Precipitation Treatment in the Remediation of Diverse Mine Tailings │ Sarah M. Miles ........................................................................... 829

vii

Molecular Biological Tools for Monitoring Selenium Reduction and Metabolic Functions in Microbial Communities │ Larissa Smith, Asma Rahman, Melody Vachon, Philip Dennis, Jeff Roberts, Silvia Mancini, Andrew Holmes and Rachel James .............. 841 Experimental Evaluation of Acid Mine Drainage Potential for Cemented Paste Backfill │ Mohammad Shafaet Jamil, Michael Marsh and Stephen Butt ............................................ 853 Holistic Nitrogen Management throughout the Project Life Cycle │ Rachael Mutsaerts and David Kratochvil ................................................................................................................. 867 The Impact of Climate Change on Extreme Events for Operation and Closure of Tailings Facilities │ Marc Olivier Trottier, Kathryn Franklin, Julio Portocarrero, Daryl Dufault and Robert Millar ................................................................................................................ 881 Chapter 8: Geotechnical Issues .............................................................................................. 889 Analysis of the Compaction Behaviour of Iron Ore Filtered Tailings on Dry Stacks │ Ana Luisa Cezar Rissoli, Géssica Soares Pereira, Daniel Bastos Ferreira and Anselmo José Coelho Mendes ................................................................................................................................ 891 Static Liquefaction of Nearly-Saturated Oil Sand Tailings │ Abouzar Sadrekarimi and Farshad Zehforoosh ............................................................................................................. 905 Life of Mine Tailings Consolidation Model │ Arturo Rodriguez, David Williams, Sebastian Quintero and Engels Trejo .................................................................................................. 917 Tailings Behaviour Interpretation at High Stresses │ Alfonso Cerna-Diaz, Thomas Barham, Osmar Charca, Jason Hilgers, Lisa Yenne, Ranjiv Gupta and Mike Waldron ................... 933 Earthquake-induced Deformation Analysis of a TSF Undergoing Tailings Reprocessing │ Paola Torres, Jorge Macedo and Solange Paihua ................................................................ 947 Seismic Assessments for Decharacterization Projects in Brazil │ José Ccotohuanca, Raphael Viola and Enzo Silva Dias ................................................................................................... 959 Muon Radiography as a Novel Tool to Characterize Tailing Storage Facilities │ Tancredi Botto, Ricardo Repenning, Claudio Rocha and Paolo Farina ............................................. 973 Settlement Evaluation of a Tailings Dam and its Application in Performance-based Design │ Pooya A. Sheykhloo, Braden Error, Lisa Yenne and Richard Davidson ............................ 985 Geotechnical Characterization and Investigation of an Iron Ore Tailings Dry Stack │ Victor Bretas, Nathalia Sena, Fernanda Gavioli, Filipe Costa, Breno Castilho and Miguel Villalobos ............................................................................................................................ 993 Chapter 9: Governance and Management .......................................................................... 1001 Tailings Facility Failures – Lessons for Accountable Executives and Consulting Engineers │ Michael Davies, John Lupo and Caius Priscu ................................................................... 1003

viii

Organizational Structures for Tailings Management, Applying Learnings from Other Industries │ Jarrad Coffey and Taylor Welch ................................................................... 1017 The Willingness of Young Engineers to Assume the Role of EOR for TSFs │ Mitchell Prince and Jeremy Boswell ............................................................................................... 1031 ALARP for Safety Management of Tailings Storage Facilities │ Yi Zhu and Marnie Pascoe ................................................................................................................................ 1041 Process Safety Approach for Reviewing Critical Controls on Tailings Storage Facilities │ George Afriyie and Kimberly Finke Morrison .................................................................. 1055 Critical Review of Distribution and Causes of TSF Failure Using Publicly Disclosed Information │ Cristina Vulpe, Josephine Liantono, Wei Liu and Andy Fourie ............... 1069 The Role of Corporate Governance: How to Develop a Global Tailings Division │ Louise McNab, Johan Boshoff, Lucas Scampoli, Maguire Walsh and Fabiana Cirillo ............... 1081 A Fundamental Shift in the Development of Operating, Maintenance and Surveillance Manuals – a Case Study│ Louise McNab, Li-Bonné Lotter, Stephen Joseph, Johan Boshoff and Nathaniel Asifu Mensah ..................................................................... 1091 Chapter 10: Liners and Supportive Technologies .............................................................. 1103 Conversion of a Downstream Raised Embankment from Clay Core to Bituminous Geomembrane Liner │ Wes Herweynen, Rob Longey and Bill Oats .............................. 1105 Mine Waste Storage Facility Liner Design and Testing – A Case Study from the Cerro Corona Mine │ Raquel Borja, Jorge Duenas, Richard Ordonez and Josh Rogers ........... 1117 High Capacity Pressure Filters for Tailings Dewatering: Slurry Characteristics and Their Impact on Process Optimization │ Francesco Kaswalder, Andrea Grosso, Nicola Mirko Finocchiaro, Andrea Pezzi and Luca Zanoni .................................................................... 1129 Assessing Compression Wave Velocity in Tailings – a Case Study │ Thomas Barham, Alfonso Cerna-Diaz, Raguvind Gounder and Kanyembo Katapa .................................... 1141 Deconstruction of an Upstream Raised Tailings Storage Facility: Project Design and Execution │ Darryl Godley and Gastón Quaglia .............................................................. 1155 Chapter 11: Risk Management ............................................................................................ 1167 Developing Recovery Plans for Catastrophic Tailings Dam Failure Events │ Caius Priscu, Doina Priscu and Janis Shandro ........................................................................................ 1169 Pilbara Case Study: Tailings Dams’ Failure Probability Quantification, Worlds’ Benchmarks and ALARP │ Cesar Oboni and Franco Oboni ................................................................ 1187 A Discussion on Using Numbers to Demonstrate ALARP │ Scott Gover, Peter Chapman and Jordan Ribbons ........................................................................................................... 1199

ix

Using Statistics and Judgment to Generate Screening-level Estimates of Tailings Dam Risk │ Michael Porter, Brad Russell and Bryan Watts .................................................... 1217 Case Study: Approach to Determining the Risk Mitigation Priority of a Historic TSF in North America │ Andy Rudy and Jeff Coffin ............................................................................. 1231 Credible Failure Modes – Summary of 2021 and 2023 Workshops │ Andy Small, Angela Küpper, Tamara Johndrow and Mohammad (Mamun) Al-Mamun .................................. 1245 The Tailings Safety Case – An Example │ Jiri Herza, Jarrad Coffey and Ryan Singh ........... 1257 Risk-informed Weighting and Communicating Uncertainties for Tailings MAAs │ Kate Patterson, Jamie Engman, Laura Wishart, Len Murray and James Penman ..................... 1277 A Comprehensive Risk Management Strategy │ Michael James, Michel Julien, Jessica Huza, Edouard Masengo and Thomas Lepine ................................................................... 1289 Did Someone Say ALARP? If So, How Do We Get There? │ Malcolm Barker, Manoj Laxman and Jayamini Methiwala ...................................................................................... 1301 Application of Monte Carlo Simulation to the Probability Assessment in FMEA │ Geinfranco Villalta and Raquel Borja .................................................................................................. 1317 Operational Control and Trigger Action Response Plan for a Tailings Storage Facility │ Alejandro Calvo, Manuel Cervantes, Laura Moreno and Diego Cobos ........................... 1337 Extending a Mining Company’s Risk Assessment Framework to Tailings Facilities │ Andy Small, Johanna Barbaran and Ogechi Mary Ileme ............................................................ 1347 Chapter 12: Site Investigation .............................................................................................. 1361 Balabag Tailings Storage Facility – Successful Construction Monitoring and Supervision │ Farzad Daliri, Rob Longey, David Brett and Cliff James ................................................. 1363 Monitoring of Iron Tailings Saturation Using Water Moisture Sensors │ Luciano Souza, Thatyane Gonçalves, Evelyn Santos, José Ccotohuanca, João Paulo Silva, Andy Small and Frank Pereira ............................................................................................................... 1373 Back-Analysis of Earthquake-triggered Las Palmas Tailings Dam Failure and Runout Processes │ Shielan Liu and Harvey McLeod .................................................................. 1385 Improvement in the Recovery of Undisturbed Samples in Soils with High Resistance to Penetration Using the Pitcher Sampler │ Sabrina Rocha, Jeanne Castro, Rodrigo Fonseca and Filipe Costa ................................................................................................... 1399 Deep Drive® – a Remotely Operated Heavy CPT System with Continuous Pushing for Access Restricted Locations │ Ray Wood, Frank Pereira, Larissa Rezende, Ana Leal, Alessander Kormann, Barbara Motta and Paul Roach ...................................................... 1409 Deep CPTs to Characterize Mine Tailings at Pinto Valley │ Kelly Cabal, Timothy Boyd and Daniel Servigna .......................................................................................................... 1419

x

Determination of the Degree of Saturation above the Water Table from CPTu Probing in Tailings │ E. Rust and M. Rust ......................................................................................... 1429 An Assessment of the Accuracy of Existing Methods to Interpret Partially Drained CPT Data in Mine Tailings │ Mason Ghafghazi, Wei Liu, James Sharp and Michael Etezad ......... 1445 Brittleness of Iron Ore Tailings – Fact or Artifact, a Case Study │ N.J. Vermeulen and A. Archer ........................................................................................................................... 1457 Characterization of In-Situ State Parameter through Tube Measurement and CPTu Methods in Partially Saturated Tailings │ Ezra Coyle, Bobby Otieno and David Reid .................. 1469 Seismic Safety of Upstream Tailings Dams: A Risk-Informed Case │ Manuel Monroy, Yuan Li and Alfonso Cordero ........................................................................................... 1479 Establishing a Site-Specific Standard of Practice for Field Vane Shear Testing in Mine Tailings │ Jason W. Harvey, Arielle A. Hogan, Dafar N. Obeidat, Iván A. Contreras and Shane A. Kelly .................................................................................................................. 1493 Partial Drainage Effects during Vane Shear Tests, with an Emphasis on the Measurement of Remoulded Strengths │ David Reid, Caleb Rodriguez, Andy Fourie and Bandana Tiwari ................................................................................................................................ 1505 Tailings State from Basic Laboratory Tests and SCPT Data │ Mahmoud N. Hussien and Jamel Sgaoula .................................................................................................................... 1517 Use of CPT with Dual Pore-Water Pressure Filter Elements in Characterization of Mine Tailings │ Iván A. Contreras, Jason W. Harvey and Dafar N. Obeidat ............................ 1531 Analysis of Vibrational Waves Induced by Machinery in Tailings Dams │ Ricardo Cabette Ramos, Thatyane Martins Gonçalves, Gino Calderon Vizcarra and Frank M. S. Pereira ............................................................................................................ 1543 Evaluation of Liquefied Strength of Uncompacted Tailings Sand Using Cone Penetration Test at an Oil Sands Tailings Facility │ Ying Zhang, Ayman H. Abusaid, Gordon W. Pollock, Yvonne (Yirao) Qiu, Ryan Moore and Jason Rhee ............................................ 1557 Chapter 13: Supportive Technologies ................................................................................. 1571 Hydraulic Dewatered Stacking – Developing Strategies for Brownfield Applications at Mogalakwena, South Africa │ Murray McGregor, Phil Newman and Andrea López ..... 1573 Successes with Hydraulic Dewatered Stacking at the El Soldado Demonstration Facility │ Phil Newman, Mark Bruton, Andrea López, Jose Burgos and James Purrington ............ 1587 Case Study of Tailings Deposition Planning and Monitoring Using Frequent Drone-Based Photogrammetry Surveys for a TSF │ Laura Nugent, Ranjiv Gupta, Robert Snow, Jason Hilgers and Julia Hawn ..................................................................................................... 1601 Application and Learnings of GISTM Management Optimization Tools │ Karen Bechard, Rachel James, Kristin Salzsauler and Chao Han ............................................................... 1613

xi

Nature-based Biocementation Approach for Mine Tailings Stabilization │ Md Al Imran, Susan Anne Baldwin, Julian McGreevy and Bill Burton .................................................. 1625 Soda Ash Leaching of Cantung Mine Tailings for Tungsten Recovery and Desulfurization │ Terry C. Cheng, Ceferino Soriano, Derek Smith, Heather Jamieson and Hendrik Falck ............................................................................................................. 1639 Investigation of Solids Content Influence on High Density Slurries Niobium Fine Tailings Disposal │ Francisco Gregianin Testa, Marcos Antônio Lemos Júnior, Renata Monteiro Furtado and Arthur Pinto Chaves ...................................................................................... 1651 Design and Development of a Dam Breach Detection System │ Ignacio García Schmidt, Erik Ketilson and Darryl Godley .............................................................................................. 1663 Full-Scale Demonstration of Somerset Sub325 Dewatering Centrifuge: Discussion of Testing Procedures and Results from Multiple Locations, Applications, and Industries │ Mike Barish, James C. Fisher II, Tony Toney and Dave Osborne ............................................. 1671 Slurry Trenching and Soil Mixing Applications for Tailings Dam Construction and Improvement │ Nathan Coughenour, Daniel Ruffing and Hubert Guimont .................... 1687 Chapter 14: Surveillance and Monitoring .......................................................................... 1697 The Deployment of the Internet of Things for Tailings Monitoring │ Vincent Le Borgne and Adam Dulmage .................................................................................................................. 1699 Historical Topographic Surveys of Jagersfontein Tailings Dam Produced from Archive Satellite Images │ Sam Rivet and Matt Nishiyama .......................................................... 1711 Advanced Analytics Applied to InSAR Dam Monitoring Program │ Luciano M. de Assis, Nicholas A.C. Marino, Gabriel G. Magalhães, Karina Pereira and Lirielly Vitorugo ...... 1739 InSAR Monitoring at the Cerro Corona TSF │ Ed Sage, Josh Rogers, Rachel Holley, Hayley Larkin, Narayanee Vummidi, Johan Boshoff, Jimmy Fiestas, Richard Ordonez, Jorge Dueñas and Brandon Pastor .............................................................................................. 1755 Comprehensive Monitoring of Tailings Dams: Navigating the Limitations and Blind Spots of Interferometric Synthetic Aperture Radar (InSAR) Technology │ Michael Davis and Nick Linton ....................................................................................................................... 1767 Fiber Optic Distributed Sensing System for Monitoring of Tailings Storage Facilities │ Daniele Inaudi, Roberto Walder, Régis Blin and Olivier Dinh ......................................... 1779 Deterministic and Statistical Analysis in the Definition of Triggered Action Response Plans in Tailings Dams │ Alfredo João Carvalho Nunes, Francesco Cavalieri, Henrique Lopes dos Santos Lopes, André Pereira Lima and Raphael Santos Rodrigues .................................. 1793 Interpretation of Vibration-Induced Pressure Generation in Experimental Excavation of an Iron Tailings Reservoir │ Thatyane Martins Gonçalves, Deborah A. Perotti, Giacomo Re, Tarcísio B. Celestino, Frank M.S. Pereira and Ricardo Cabette Ramos ........................... 1805

xii

Spectral Index Monitoring for Temporal Changes in Mining Areas │ Prabin Acharya and Fangzhou Liu ..................................................................................................................... 1817 Chapter 15: Sustainability Issues ......................................................................................... 1829 Climate Change Resilience Assessments for Tailings Storage Facilities │ Leila Ang and Nigel Moon ....................................................................................................................... 1831 An Approach to Undertaking a Tailings-related Human Rights Impact Assessment in Ghana │ Louise McNab, Robert Siaw, Maud Ofosua Ofori, Florence Ansere-Bioh, Ashleigh Shelton, Nathaniel Asifu Mensah and Johan Boshoff ....................................................... 1845 Study of Mechanical Properties of One-Part Geopolymers Made from Iron Ore Tailings │ R.K.R. Silva, F.S. de Faria, M.A. Longhi, P.F.F. Martins and F.S. Lameiras .................. 1859 New Directions for Tailings Management │ Priscilla P. Nelson ............................................. 1871 Chapter 16: Author Index ...................................................................................................... 1885 Author Index ............................................................................................................................. 1887

xiii

ORGANIZERS Norman B. Keevil Institute of Mining Engineering (NBKIME), University of British Columbia (UBC) A leader in mining education, research, and industry partnerships, UBC’s Department of Mining Engineering has faculty that teaches and mentors graduate students and staff to undertake research in all aspects of mining in order to study and improve the industry for future generations. One of North America’s largest and most advanced centres for mining engineering education and research, the UBC NBKIME is known for being a small, close-knit family. The department is exemplified by the dedication of the faculty and staff who provide a dynamic, hands-on learning experience for both undergraduate and graduate students, combined with a diverse and inclusive educational and research environment, open to everyone interested in this field. Gifts from alumni, corporations, foundations, students, parents and other friends assist the NBKIME in conducting leading edge research, providing outstanding education and contributing to social and economic development. For more information, visit the NBKIME website at http://mining.ubc.ca

C3 Alliance Corp. At C3 Alliance Corp., our mission is to create stronger communities through innovative and inclusive solutions for resource development. We understand that the sharing of information and ideas is vital to gaining insights and developing relationships and that community members, business, and government representatives can all benefit from meeting to share stories and knowledge. C3’s expert team of event planners specialize in providing project management, strategic knowledge and expertise for planning conferences and networking opportunities. We benefit from a vast network of professional contacts to organize exceptional events that connect people and knowledge across resources sectors, communities and government. For more information, visit the C3 website at http://www.c3alliancecorp.ca/

xiv

COMMITTEES Conference Chair Caius Priscu, UBC NBKIME, and Priscu and Associates Consulting Engineers Inc.

Organizing Committee Caius Priscu, UBC NBKIME, and Priscu and Associates Consulting Engineers Inc. Michael Davies, Davies Mining Consultancy Ltd. Chris Anderson, Teck Resources Ltd. Dirk Van Zyl, Professor Emeritus, UBC NBKIME Daryl Hockley, SRK Consulting Inc. Kate Patterson, Klohn Crippen Berger Grant Mitchell, BGC Engineering Robyn Gaebel, WSP Masaki Miyoshi, Teck Resources Ltd. Daniela Uhlenbruck, Klohn Crippen Berger Gaston Quaglia, SRK Consulting Inc. Bryan Bale, Teck Resources Ltd.

Conference Secretariat Sarah Weber, C3 Alliance Corp. Amrita Budhwar, C3 Alliance Corp. Meryl Nelson, C3 Alliance Corp. Nicole Vanderfleet, C3 Alliance Corp. Melissa McRitchie, C3 Alliance Corp. Bronagh Furlong, C3 Alliance Corp.

Technical Review Committee The Organizing Committee expresses its appreciation and sincere gratitude to the following reviewers of the technical content, who helped to ensure the high quality of this publication. Jason Hilgers, AECOM Janis Shandro, Arrowsmith Gold Inc. Hernan Cifuentes, ATC Williams Ali Hooshiar, Ausenco Kurt Schimpke, Barr Engineering Tim Bekhuys, Bekhuys & Associates Alan Chou, BGC Engineering Inc. Andrea Lougheed, BGC Engineering Inc. Andrea Walsh, BGC Engineering Inc.

Reza Moghaddam, Lundin Mining Carlo Cooper, Minebridge Software Inc. Michael Nahir, Parsons Caius Priscu, UBC NBKIME, and Priscu and Associates Consulting Engineers Inc. Doina Priscu, Priscu and Associates Consulting Engineers Inc. Dirk van Zyl, Professor Emeritus, UBC NBKIME xv

Caroline Bates, BGC Engineering Inc. Chang-Gyun (CJ) Jeong, BGC Engineering Inc. Jack Seto, BGC Engineering Inc. Mike Belfry, BGC Engineering Inc. Renata Wood, BGC Engineering Inc. Dean Korri, Cliffs Technology Group David Slack, ConeTec Dallas McGowan, ConeTec Cassandra Hall, FMI Georgia Lysay, FMI Andrew Holmes, Geosyntec Consultants Inc. Daniel LaPorte, Geosyntec Consultants Inc. Kristin Salzsauler, Geosyntec Consultants Inc. Silvia Mancini, Geosyntec Consultants Inc. Shane Kelly, Gregg Drilling Jen Durocher, Klohn Krippen Berger Joe Quinn, Klohn Krippen Berger Kate Patterson, Klohn Krippen Berger Andy Small, Klohn Krippen Berger Len Murray, Klohn Krippen Berger Rick Friedel, Klohn Krippen Berger Daniel Ruane, Knight Piesold Consulting

Stephen Bell, Putzmeister Andres Barrero, SRK Consulting Inc. Darryl Godley, SRK Consulting Inc. Daryl Hockley, SRK Consulting Inc. Erick Lino, SRK Consulting Inc. Gastón Quaglia, SRK Consulting Inc. Ignacio Cueto, SRK Consulting Inc. Brian Gray, Stantec Josh Rogers, Stantec Bryan Bale, Teck Resources Ltd Chris Kennedy, Teck Resources Ltd Jennifer Brash, Teck Resources Ltd Scott Martens, Teck Resources Ltd Chris Johns, Tetra Tech Randal Osicki, Thurber Engineering Ltd. Irwin Wilesky, WSP Lowell Wade, WSP Martyn Willan, WSP Michael Etezad, WSP Neeltje Slingerland, WSP Rui Couto, WSP Sean Wells, WSP Carlos Iturralde, Yamana Gold

Advisory Committee Jason Hilgers, AECOM Nick Raugh, AGRU America Marie-Josee Banwell, TRE ALTAMIRA Inc. Ali Hooshiar, Ausenco Kurt Schimpke, Barr Engineering Alan Chou, BGC Engineering Inc. Chris Bareither, Colorado State University David Slack, ConeTec Dallas McGowan, ConeTec Kevin Schraden, Diemme Filtration Tamara Johndrow, Freeport-McMoRan Cassandra Hall, Freeport-McMoRan Shane Kelly, Gregg Drilling Rick Friedel, Klohn Crippen Berger

Rick Friedel, Klohn Crippen Berger Daniel Ruane, Knight Piesold Carlo Cooper, Minebridge Software Todd Wisdom, Paterson & Cooke Stephen Bell, Putzmeister Ljiljana Josic, AtkinsRéalis Josh Rogers, Stantec Brian Gray, Stantec Chris Johns, Tetra Tech Randal Osicki, Thurber Engineering Ltd. André Jabir Assumpção, TPF Lucas Barros Duarte, Vale Martyn Willan, WSP

xvi

THE ORGANIZERS GRATEFULLY ACKNOWLEDGE THE FOLLOWING CONFERENCE SPONSORS AND EXHIBITORS:

SPONSORS

xvii

SPONSORS

xviii

EXHIBITORS AECOM Afitex-Texel Geosynthetics AGAT Laboratories AGRU America, Inc. Alfa Laval AtkinsRéalis Axter Coletanche Inc. Barr Engineering & Environmental Science Canada Ltd. Bauer Foundations Beadedstream inc. BGC Engineering Cambridge Insitu Campbell Scientific Canary Systems Carpi Tech CEMI/MICA CGG ConeTec Copperstone Technologies Ltd Diemme Filtration SRL EcoMister Evaporator Enviro Integration Strategies Envirosuite Esdat Software by Escis Freeport-McMoRan GBG Group GEOKON Geosense Ltd Geosyntec Consultants, International Inc. Geotech Drilling – Gregg Drilling GHD Limited GKM Consultants Graymont Hatch Ltd. HUESKER Inc. Institute of Mine Seismology

Kamengo Keller Klohn Crippen Berger Ltd Knight Piesold Ltd Layfield Geosynthetics Lodige Process Technology, Inc. LSI Lastem Matec Industries SPA MENARD CANADA Micronics Engineered Filtration Group MineBridge Software Inc. Nanometrics Parsons Paterson & Cooke PhotoSat PURE PVC Sheetpiles System Putzmeister Resource Engineering Consultants ROCTEST LTD RSL Membranes / waterStrider SECURE Energy Silixa Ltd Solmax International Inc. Somerset International SRK Consulting (Canada) Inc. Stantec Suncor Energy Inc. Teck Resources Ltd Terra Insights Tetra Tech TPF TRE Altamira Inc. Valmet LTD Veolia Water Technologies & Solutions Willowstick Technologies LLC Wood Canada Ltd. WSP

xix

“Sharing knowledge is teaching others, and in doing so, we teach ourselves.”

Foreword The success of the Tailings and Mine Waste conference series is measured by the quality of technical contributions, the support it receives, and how well the event is organized. This informative conference would not be possible without quality technical papers, experienced keynote speakers and panelists, supportive sponsors and exhibitors, and an amazing team of volunteers and organizers. The international community rose to the occasion once again for the Tailings and Mine Waste 2023 conference, held in beautiful Vancouver, BC, Canada. The 2023 Organizing Committee acknowledges the contributions of this year’s authors for their high quality, practical, and innovative papers. The Technical Committee had the difficult task of selecting the final papers from a list of 300+ abstract submissions, a record so far for this conference series. We thank all the technical reviewers, those responsible for the preparation and production of these proceedings, as well as all those involved with the organization of this conference. The support of the various committees, companies and individuals representing them is invaluable and greatly appreciated. The Organizing Committee strongly believes that only through the sharing of ideas, high-quality debate and from learning from both successful and challenging past experiences, we can advance knowledge and improve in this critically important field of practice. Again, thank you to all speakers, participants, and presenters for sharing such valuable information and expertise. You have contributed to the great success of Tailings and Mine Waste 2023 in Vancouver. We look forward to seeing you all again at the next conference in this international series. Caius Priscu, Ph.D, P.Eng Conference Chair Tailings and Mine Waste 2023 Vancouver, BC, Canada

xx

Chapter One

Analysis

1

2

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Seismic Evaluation of a Tailings Dam Using Uncoupled and Fully Coupled Soil Constitutive Models Zana Karimi, AECOM, USA Pooya A. Sheykhloo, AECOM, USA Lisa Yenne, AECOM, USA Ron Hickman, Freeport-McMoRan Inc., USA Sam Saunders, Freeport-McMoRan Inc., USA Masood H. Kafash, AECOM, USA Amin Gheibi, AECOM, USA

Abstract The Henderson Tailings Storage Facility (TSF) is an active facility located near Parshall, Colorado, consisting of two dams – 1 Dam and 3 Dam. The upstream method of construction has been used for tailings deposition since the mid-1970s. This paper presents the results of the seismic evaluation of 1 Dam using multiple constitutive modeling methodologies: an uncoupled Mohr-Coulomb approach and a fully coupled critical state-compatible PM4Sand/Silt approach. The analyses were completed to study the seismic response of the tailings embankment under the Maximum Design Earthquake (MDE) with a return period of 10,000 years. The numerical analyses showed that the results of the fully coupled, effective stress PM4 models were generally consistent with the uncoupled Mohr-Coulomb models. The PM4 models have the capability to estimate the generation of excess pore water pressure and onset of soil liquefaction during the application of the input ground motion. Subsequent zones of tailings materials that were prone to soil liquefaction or strength loss were identified based on two criteria: 1) excess pore water pressure ratio; and 2) shear strain in the PM4 modeling. This study provides valuable insights into the methods used to estimate seismic response of the Henderson TSF and highlights the importance of using properly calibrated advanced constitutive modeling methodologies to capture the complex response of tailings materials under seismic loading. The fully coupled models were capable of capturing the hysteretic soil response, including stress-strain response and accumulation of plastic shear strains, providing confidence in the predicted modes of deformation and informing the design and management of the facility.

3

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Tailings are the by-products generated during mining activities. They are usually transported in a slurry form to a containment area known as an impoundment, typically enclosed by embankments called tailings dams. These embankments are constructed using different methods such as upstream, downstream, or centerline techniques, and are raised in multiple stages or lifts. They consist of various soil types, with differing layer orientations and thicknesses, as well as zones containing materials ranging from fine to coarse, which vary in density. Throughout the construction process and the operational lifespan of a tailings dam, the layers are exposed to different loading conditions, including potential seismic activity. Evaluation of the long-term performance of a tailings embankment and its stability under static and dynamic loading is performed using different numerical methods to confirm its safety during and after mining operations. This paper presents the results of the seismic evaluation of the 1 Dam embankment at the Henderson Tailings Storage Facility (TSF) using two soil constitutive modeling methodologies including an uncoupled Mohr-Coulomb approach and a fully coupled critical state-compatible PM4Sand/Silt approach under the Maximum Design Earthquake (MDE). The numerical analyses indicated that the results from the fully coupled, effective stress PM4 models were generally in agreement with the uncoupled Mohr-Coulomb models. The fully coupled models more accurately accounted for hysteretic soil responses, including non-linear stress-strain response and the subyield accumulation of shear strains. Additionally, the PM4 models had the capability to estimate the generation of excess pore water pressure and subsequent onset (timing) of soil liquefaction during the application of input ground motion. Evaluation of these parameters allowed for a more realistic estimation of anticipated deformation patterns, as well as identification of zones of tailings materials that were prone to soil liquefaction or strength loss.

Site overview The Henderson Mill is located in Grand County, Colorado near the summit of Ute Pass. The Mill is owned and operated by Climax Molybdenum Company, a subsidiary of Freeport-McMoRan Inc. Henderson Mill operates a single tailings storage facility composed of two active tailings dams as shown on Figure 1. The largest dam, known as 1 Dam, initially provided primary containment. A smaller dam, located south of the main dam and known as 3 Dam, has combined with 1 Dam, resulting in a singular crest formed by the two embankments. The Henderson tailings impoundment is being raised using the upstream method of dam construction. 1 Dam and 3 Dam currently impound tailings composed of hydraulically deposited sand, silt, and clay size particles.

4

SEISMIC EVALUATION OF A TAILINGS DAM USING UNCOUPLED AND FULLY COUPLED SOIL CONSTITUTIVE MODELS

Figure 1: Henderson Mill facilities (Source: Google Earth, September 2019)

Numerical dynamic deformation methodology Dynamic deformation modeling was performed using the computer code FLAC - Fast Lagrangian Analysis of Continua, Version 8.0 (Itasca, 2019). This version of FLAC is a two-dimensional, explicit finite difference computer program for performing soil-structure interaction analysis, capable of estimating behaviour at large strains, and it was used to perform the seismic deformation analysis. Analyses were completed under the MDE using two constitutive models: 1) an uncoupled Mohr-Coulomb model; and 2) a fully coupled critical state-compatible PM4Sand/Silt model. Uncoupled Mohr-Coulomb is a linear-elastic perfectly plastic model that does not account for generation of excess pore pressure during cyclic loading. In this model, the onset of soil liquefaction or cyclic softening is implemented into the model at a certain time step during the ground shaking through manual reduction in material shear strength and stiffness. The Mohr-Coulomb model has been widely used in industry by practitioners for many years. It provides a good baseline comparison with advanced and complex models such as PM4Sand and PM4Silt. Advanced, fully coupled, effective-stress constitutive models PM4Silt and PM4Sand (Boulanger and Ziotopoulou, 2019) explicitly incorporate pore pressure generation and subsequent estimation of timing (onset) of soil liquefication during the application of ground shaking. These models simulate nonlinearity in soil response, in addition to capturing the accumulation of hysteretic shear strain and subsequent displacements. These advanced constitutive models provide a more reliable estimate of deformations (Montgomery and Abbaszadeh, 2017). The FLAC deformation analyses include the following steps: 1. A turn-on gravity analysis to evaluate the initial stresses before an input earthquake motion is applied. At the end of the analysis, the calculated stresses (horizontal, vertical, and shear stresses) should satisfy

5

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA the horizontal and vertical force equilibrium conditions and strain compatibility conditions should be satisfied, as well. 2. Dynamic analysis is performed to evaluate the response and deformation of the embankment from gravity and the input earthquake motion. In this case, the dynamic stresses are added to the static stresses, and the earth structure is allowed to deform (translate, rotate, compress, and expand). 3. For the Mohr-Coulomb model, the residual strengths and reduced stiffness are assigned after the first peak of acceleration of the input ground motion to implement the liquefied or cyclically softened state of some targeted zones of materials. These zones are identified through a two-dimensional equivalentlinear, drained analysis in the Quad4 platform (Hudson et al., 1993). The timing of liquefaction in the Mohr-Coulomb model is also examined to evaluate possible base isolation effects due to the early assignment of post-earthquake properties to the liquefied zones. A post-earthquake analysis is performed to evaluate the deformations of the embankment under gravity loading alone, following the input earthquake shaking. For PM4 analyses, during the post-shaking (i.e., post-earthquake) stage, the postearthquake material properties such as residual shear strength and stiffness are assigned to the region(s) of materials that are identified as potentially liquefiable or cyclic softened based on: 1) excess pore pressure ratio (Ru) generation; or 2) developed peak shear strain (g) level criteria. These two postearthquake criteria are discussed in more details in the following sections. 4. During this post-earthquake stage of analysis, the input earthquake motion is stopped, but the gravity load is maintained to evaluate deformations that may be induced by re-adjustment of stresses and strains developed during earthquake shaking (e.g., Naesgaard and Byrne, 2007; Perlea and Beaty, 2010). The gravity force is applied until a final, static equilibrium is achieved. Alternatively, if the structure is “unstable” under post-earthquake loading, a static force imbalance is maintained, and the structure will continue to deform, indicating that a “flow failure” of the impoundment may be considered a likely outcome.

Model geometry, material properties, and pore pressure condition A representative two-dimensional study cross section used in analyses is shown on Figure 2. This section has the maximum height of tailings of 265 feet (ft). This section has a starter dam and consists of up to 250 ft of coarse tailings overlying a layer (approximately 10 to 20 ft) of fine tailings, which sits on native foundation materials. There is also a continuous thin layer of fine tailings beneath the entire section, correlating well with the depositional history. The estimated pore pressure conditions and phreatic surface are shown on Figure 2. Extensive field investigation programs have been completed at the Henderson tailings dams. The field investigations have included seismic cone penetration test (sCPT) soundings, drill holes, in-situ vane shear

6

SEISMIC EVALUATION OF A TAILINGS DAM USING UNCOUPLED AND FULLY COUPLED SOIL CONSTITUTIVE MODELS testing, and geophysical surveys. Multiple laboratory testing programs have been completed on samples of tailings and foundation materials during the last few decades. Relatively undisturbed Shelby tube samples from the drilling investigations, as well as disturbed samples have been used to assess engineering properties and strength testing. Laboratory testing has included static, cyclic, and post-cyclic direct simple shear (DSS) testing; consolidation testing; static, cyclic, and post-cyclic consolidated isotropic undrained triaxial testing with pore pressure measurements (CIU’); static consolidated isotropic drained triaxial testing (CID); index testing; and permeability testing.

Figure 2: Study section geometry and phreatic surface Advanced materials characterization through field measurements and laboratory data was performed for each material type. Coarse tailings were characterized as dilative sand-like material with cyclic mobility, and fine tailings as ductile contractive clay-like material with a slight reduction in shear strength at shear strains greater than 10 to 15 percent. The starter dam is characterized as well-compacted sand-like materials with high cyclic and static shear resistance. The upper foundation is characterized as dilative clay-like materials with no potential for strain softening. The lower foundation is characterized as dilative sand-like materials with high cyclic and static resistance.

Single-element calibrations PM4Silt PM4Silt is a critical-state-based and stress ratio-controlled bounding surface plasticity model for lowplasticity silts and clays that exhibit stress history and normalized, undrained shear strength in geotechnical earthquake engineering applications (Boulanger and Ziotopoulou, 2019). The bounding and dilation stress ratios (Mb and Md) are functions of the state parameter (ξ), so that they converge to the critical state stress ratio (M) as the soil is sheared to critical state (ξ = 0). Four primary parameters exist for model calibration and 20 secondary parameters may be modified from their default values. The four primary input parameters are undrained shear strength at critical state (Su,cs) or undrained shear strength ratio at critical state (Su,cs/σ‘vc),

7

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA shear modulus coefficient (Go), contraction rate parameter (hpo), and a post-strong-shaking shear strength reduction factor (Fsu). Fine tailings and upper foundation materials are modeled with PM4Silt. The main purpose of modeling these two materials with a PM4Silt model is to incorporate the hysteretic deformations and subyield accumulation of strain during the analyses and evaluate its potential for liquefaction or cyclic softening. Other purposes of calibration include capturing rate of excess pore pressure build-up, straincompatible shear modulus reduction and damping curves, as well as monotonic responses. Single-element calibration analyses are performed to calibrate the primary and selected secondary input parameters for the model against the laboratory testing.

PM4Sand PM4Sand is a stress-ratio controlled, critical state-compatible, bounding surface plasticity model for sand and other purely non-plastic granular soils that was developed to approximate stress-strain responses of specific importance to geotechnical earthquake engineering and liquefaction problems (Boulanger and Ziotopoulou, 2017). Three primary parameters exist for model calibration and 16 secondary parameters may be modified from their default values. The three primary input parameters are: (1) the shear modulus coefficient, Go, which should be calibrated to the estimated or measured in-situ Vs; (2) an apparent relative density (DR) that affects the peak drained and undrained strengths and the rate of strain accumulation during cyclic loading; and (3) the hpo, which is used to calibrate the model to the estimated in situ cyclic resistance ratio (CRR). Coarse tailings and starter dam materials were modeled using PM4Sand. Assessments of laboratory testing results indicate possible sample disturbance due to the sand-like, low-plasticity nature of the materials. Therefore, the main emphasis was put on the field data including the sCPT signatures and shear wave velocity measurements during the calibration of this material (lab-based calibration was investigated and considered unsuitable for the PM4Sand calibration). The purpose of modeling this layer with the PM4Sand model was to incorporate the hysteretic responses for this type of material including cyclic mobility, stress-strain response, and rate of excess pore pressure generation. Other purposes of calibration included capturing strain-compatible shear modulus reduction and damping curves. Single-element calibration analyses were performed to calibrate the primary and selected secondary input parameters for the model. Primary parameters, N1,60 and Go, were selected, based on field measurements, and the primary parameter hpo was calibrated against cyclic resistance response (Idriss and Boulanger, 2008).

8

SEISMIC EVALUATION OF A TAILINGS DAM USING UNCOUPLED AND FULLY COUPLED SOIL CONSTITUTIVE MODELS

Dynamic deformation analysis results Mohr-Coulomb model Dynamic deformation using the Mohr-Coulomb soil constitutive model for the study section and configuration shown on Figure 2 was completed under the design ground motion (spectrally matched to the mean 10,000-year return period hazard). Contours of shear strain (e) and vectors of displacement are shown on Figure 3. Results of dynamic deformation analysis show that tailings material located near the toe and behind the starter dam moves downstream with a maximum vector of about 3.9 ft. The liquefied portion of the fine tailings layer behind the starter dam (extending about 300 ft upstream) dominates the deformation pattern as can be seen from the contours of shear strains and distortion in the meshes in this zone. This mode of movement leads to a minimal crest settlement and transition of about 0.1 and 0.4 ft, respectively as shown on Figure 4.

Figure 3: Mohr-Coulomb model; contours of shear strain and vectors of displacement Maximum predicted settlement is about 1.8 ft at the mid-slope. Maximum transitional movement occurs at the toe and behind the starter dam, which is about 3.5 ft downstream with about 1.5 ft of heave. Time histories of horizontal and vertical displacements show that deformations arrest at the end of shaking and the embankment is stable with tolerable deformations. The resultant mode of deformation in this simulation is inclined to follow the softened zone with assignment of post-earthquake properties within the fine tailings located near the embankment toe.

9

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 4: Mohr-Coulomb model; contours of horizontal and vertical displacements

PM4Sand/Silt model Dynamic deformation analyses for the study section and configuration shown on Figure 2 were completed using fully coupled effective stress dynamic analyses incorporating both PM4Sand and PM4Silt models following the approach previously described above. In these analyses, at the end of motion, excess pore pressure and peak shear strains within each element were examined against the triggering criteria, discussed below, to identify region(s) of materials prone to liquefaction or cyclic softening and, therefore, reducing the strength and stiffness to the post-earthquake material properties. Then, the model continued for an additional 10 to 15 seconds under gravity loading alone to assess its stability. It was assumed that the material reaches a state of sustained soil liquefaction if it generates great enough excess pore water pressures to exceed a threshold value at the end of shaking. This threshold value was set to an excess pore water pressure ratio (Ru) of 0.7 or greater at the end of application of ground motion. Similarly, it was assumed that the material reaches a state of sustained soil liquefaction or cyclic softening if the experienced peak shear stain (g) level exceeds a threshold value at the end of shaking. This threshold value was set to a peak shear strain (g ) level of 2.5% or greater at the end of application of ground motion. The shear strain criteria was set well below the point at which strain softening was observed in lab testing (>10%) and is considered a conservatism in the modeling method. If any of these thresholds were reached, the material was assigned residual post-earthquake properties and the analyses were continued for an additional 15 seconds under gravity loads. A summary of the dynamic deformation analyses results using the seismic design events (MDE) is provided below.

10

SEISMIC EVALUATION OF A TAILINGS DAM USING UNCOUPLED AND FULLY COUPLED SOIL CONSTITUTIVE MODELS Contours of shear strain and displacement vectors are shown in Figure 5. The model captures the nonlinearity and hysteretic accumulation of cyclic shear strain in addition to the pore pressure response. The results show that fine tailings material accumulated shear strains exceeding 5%, which extended up to 700 ft upstream of the starter dam. This leads to a block movement within the dam with about 4.4 ft of vector displacement. Figure 6 shows the contours of horizontal and vertical displacements. The maximum horizontal movement of the dam is 4.0 ft and occurs at the mid slope, while the horizontal movement at the crest is about 2.3 ft. Maximum vertical movement (i.e., settlement) is about 1.2 ft and occurs at the crest. A maximum heave of 0.8 ft is estimated behind the starter dam. Figure 7 shows the contours of excess pore pressure ratio (Ru) at the end of ground shaking. This figure shows that the Ru is below the threshold value of 0.7 in both coarse and fine tailings materials. Results from this simulation led to a more realistic estimate of the mode of deformation by capturing the sub-yield accumulation of shear strains as well as the generation of excess pore pressure and subsequent onset of liquefaction or cyclic softening.

Figure 5: PM4Sand/Silt model; contours of shear strain and vectors of displacement Horizontal and vertical displacement time histories of selected recording nodes, during and after application of the ground shaking, are shown in Figure 8, which includes history points corresponding to those shown on Figure 7; positive displacements are downstream (or heave) and negative displacements are upstream (or settlement). As shown on Figure 8, after the ground motion application (i.e., 45 seconds), the post-earthquake properties are applied to the liquefied or cyclic softened materials based on pore pressure ratio or shear strain criteria, which is driven by the strain accumulated within a zone of fine tailings

11

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA in this case. This leads to a sudden increase in displacement as can be seen on displacement time histories; however, deformations are stable at the end of analysis, indicating tolerable deformation.

Figure 6: PM4Sand/Silt model; contours of horizontal and vertical displacements

Conclusions Dynamic deformation analyses were completed using both uncoupled Mohr-Coulomb and fully coupled PM4Sand/Silt models. The comparison of the results showed that the Mohr-Coulomb model using equivalent-linear site response and liquefaction triggering analysis results can predict different displacement mode compared to PM4 models in this case. In this model, the timing of liquefaction onset was selected such that it would yield to the largest resultant deformation. The Mohr-Coulomb model started to show increasing deformation due to the yielding of zones of materials that were predefined as liquefied with post-earthquake residual shear strength and stiffness near the toe of the embankment. This model is a linear-elastic perfectly plastic model that does not account for the sub-yield accumulation of shear strains. Therefore, shear strains mainly accumulate in the zones that have reached their assigned shear strength (e.g., in the predefined liquefied zones). The fully coupled, effective stress, PM4Sand/Silt models were capable of capturing the hysteretic soil response including stress-strain response and accumulation of plastic shear strains, which in turn resulted

12

SEISMIC EVALUATION OF A TAILINGS DAM USING UNCOUPLED AND FULLY COUPLED SOIL CONSTITUTIVE MODELS in a more realistic estimate of predicted modes of deformation. Both the Mohr-Coulomb and PM4 models showed tolerable deformation.

Figure 7: PM4Sand/Silt model; contours of generated excess pore pressure ratio (Ru)

Figure 8: PM4Sand/Silt Model; displacement time histories of selected points

13

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

References Beaty, M. and P.M. Byrne. 1998. An effective stress model for predicting liquefaction behaviour of sand. In P. Dakoulas, M. Yegian, and R. Holtz, eds, American Society of Civil Engineers, Geotechnical Special Publication, pp. 766–777. Boulanger, R.W. and K. Ziotopoulou. 2017. PM4Sand: A Sand Plasticity Model for Earthquake Engineering Applications, Version 3.1. Report No. UCD/CGM-17/01, Center for Geotechnical Modeling, Department of Civil and Environmental Engineering, University of California, Davis. Boulanger, R.W. and K. Ziotopoulou. 2019. A constitutive model for clays and plastic silts in plane-strain earthquake engineering applications. Soil Dynamics and Earthquake Engineering 127(9): 105832. DOI:10.1016/j.soildyn.2019.105832 Hudson, M., I.M. Idriss and M. Beikae. 1993. User’s Manual for QUAD4MU: A Computer Program to Evaluate Seismic Response of Soil Structures Using Finite Element Procedures Incorporating a Compliant Base. Center for Geotechnical Modeling, Department of Civil and Environmental Engineering, University of California, Davis. Idriss, I.M. and R.W. Boulanger. 2008. Soil liquefaction during earthquakes. Earthquake Engineering Research Institute, MNO-12. Itasca Consulting Group, Inc. (Itasca). 2019. FLAC – Fast Lagrangian Analysis of Continua. Version 8.0. Minneapolis, MN: Itasca. Montgomery, J. and S. Abbaszadeh. 2017. Comparison of two constitutive models for simulating the effects of liquefaction on embankment dams. 37th Annual USSD Conference and Exhibition, Anaheim, CA. Naesgaard, E. and P.M. Byrne. 2007. Flow liquefaction simulation using a combined effective stress – total stress model. 60th Canadian Geotechnical Conference, Canadian Geotechnical Society, Ottawa, Ontario. Perlea, V.G. and Beaty, M.H. 2010. Corps of Engineers Practice in the Evaluation of Seismic Deformation of Embankment Dams.

14

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Seismic Deformation Assessment of Tailings Dams in Subduction Zones Jorge Macedo, Georgia Institute of Technology, USA Jonathan Bray, University of California Berkeley, USA Chenying Liu, Georgia Institute of Technology, USA

Abstract The seismic performance assessment of tailings dams requires an estimate of the seismically-induced slope displacement. Modified Newmark-type analyses are useful to develop an initial estimate of the seismic slope displacement. Most simplified seismic deformation assessment procedures were developed for shallow crustal tectonic settings. Few procedures are available for subduction zone interface earthquakes, and no procedures are available for intraslab events. This study uses the comprehensive NGA-Sub ground motion database to formulate new seismic slope displacement models for subduction zone interface and intraslab earthquakes using a mixed random variable regression formulation. The models are formulated in terms of the sliding mass properties (its yield coefficient and fundamental period), the spectral acceleration at the degraded period of the tailing dam, earthquake magnitude, and peak ground velocity. Seismic slope displacement scales differently in interface and intraslab settings. This study advances the performancebased seismic design of tailings dams affected by subduction zone earthquakes.

Introduction Seismic slope displacement procedures that utilize a modified Newmark-type sliding block model may be used to assess the potential seismic performance of tailings dams. Even in cases when advanced nonlinear effective stress analyses will be employed, the modified Newmark-type sliding block analyses provide a preliminary estimate of the likely range of the seismic slope displacement (D) of a tailings dam. Accordingly, several seismic slope procedures have been developed for use in engineering practice (e.g., Bray and Travasarou, 2007; Rathje and Antonakos, 2011; Bray and Macedo, 2019). However, few seismic slope displacement procedures exist for subduction zone earthquakes, and no seismic slope procedure specifically addresses intraslab earthquakes. Subduction intraslab seismic sources play a significant role in tailings dam seismic design, particularly in regions like the South American Andes. Figure 1 illustrates the probabilistic seismic hazard assessment

15

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA (PSHA) results for a Peruvian site, where intraslab events dominate the seismic hazard. Due to these types of tectonic settings, seismic slope displacement models for subduction intraslab tectonic settings are required. Moreover, given the scarcity of seismic slope displacement models in subduction interface tectonic settings, additional models for this tectonic setting are also warranted to better account for epistemic uncertainties.

Figure 1: Probabilistic seismic hazard assessment of a site in the Peruvian Andes: (a) Location, and (b) PGA hazard curves deaggregated by tectonic settings with intraslab events governing the seismic hazard Recent advances in the Pacific Earthquake Engineering Research (PEER) Center NGA-Sub ground motion project have provided an extensive collection of ground motion records for interface and intraslab subduction zone earthquakes. This presents an opportunity to address the gap in seismic slope displacement models for intraslab subduction earthquakes and improve upon existing models for interface subduction zone earthquakes. This paper introduces new performance-based semi-empirical models that estimate the seismic displacement of tailings dams experiencing shear deformation during subduction zone earthquakes. The models follow the framework of Bray and Travasarou (2007) and can be used in probabilistic, deterministic, and pseudo-probabilistic approaches. The study summarizes relevant previous research, presents ground motion databases for subduction zone earthquakes, and describes the seismic slope displacement models. Notably, the models demonstrate different scaling of D in interface and intraslab settings with recommendations provided for their application in engineering practice.

Ground motion database This study utilizes the recently developed NGA-Sub ground motion database (Bozorgnia and Stewart, 2020) from the PEER Center for seismic slope displacement analyses. The full NGA-Sub database contains

16

SEISMIC DEFORMATION ASSESSMENT OF TAILINGS DAMS IN SUBDUCTION ZONES 71,343 three-component recordings obtained from 1,883 subduction zone earthquakes worldwide between 1937 and 2016. Ground motion recordings for this study were selected based on the criteria outlined in Kuehn et al. (2020). The selection process narrowed down the dataset to 6,240 two-component horizontal ground motion recordings from 113 interface earthquakes with moment magnitudes (Mw) ranging from 4.8 to 9.1, and 8,299 two-component ground motion recordings from 123 intraslab earthquakes with Mw ranging from 4.0 to 7.8. Figure 2 illustrates the distributions of earthquake moment magnitude and closest sourceto-site rupture distance (Rrup) for the interface and intraslab earthquake recordings used in this study. The ground motion database employed in this study is significantly more comprehensive compared to previous studies conducted in subduction zone settings (e.g., Bray et al., 2018).

Figure 2: Moment magnitude (Mw) and closest source-to-site distance (Rrup) distribution for the subduction zone ground motion recordings used in this study for (a) interface and (b) intraslab earthquakes

Seismic slope displacement sliding block model The fully coupled, nonlinear, and deformable stick-slip one-dimensional (1D) sliding block model developed by Rathje and Bray (2000) with improvements implemented by Bray et al. (2018) (BMT18 model) is employed to calculate seismic slope displacement. The model captures the dynamic response of the deformable sliding mass using an equivalent linear viscoelastic modal analysis and strain-dependent material properties. Previous studies (e.g., Rathje and Bray, 2000; Wartman et al., 2003; Bray and Travasarou, 2007; Bray et al., 2018) have validated its ability to estimate earthquake-induced slope displacement. Ground motion records are applied to the model's base, considering 5%-damped elastic acceleration response spectral value (Sa(1.3Ts)), peak ground velocity (PGV), and moment magnitude (Mw). Two-component ground motions are combined, considering their maximum D value calculated from each polarity, and the average D values for each horizontal component are used. This approach aligns with

17

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA engineering practice. The final database used to develop the seismic slope displacement models contains 1,510,093 and 1,855,516 calculated seismic slope displacement values for the interface and intraslab earthquake, respectively.

Estimation of seismic slope displacement The BMT18 model has proven to be robust, and its estimations compare well to observations from field case histories for interface subduction zone earthquakes (Bray et al., 2018). However, this study takes advantage of the new NGA-Sub database that provides nearly 6 times more interface subduction zone earthquake records than the amount used to develop the BMT18 model. Thus, an updated interface earthquake seismic slope displacement estimation model is developed in this study.

Figure 3: Residuals (i.e., ln(Ddata) – ln(Destimated)) between the calculated seismic slope displacement for intraslab earthquakes and the estimated values from the BMT18 interface earthquake model

18

SEISMIC DEFORMATION ASSESSMENT OF TAILINGS DAMS IN SUBDUCTION ZONES The BMT18 model was not intended for intraslab subduction zone earthquakes and should not be used for this purpose. However, due to the lack of an intraslab-specific model, engineers have employed the BMT18 interface model to estimate D for intraslab earthquakes. To assess its suitability, the BMT18 model's results are compared to newly generated D values from intraslab records. Figure 3 reveals significant biases in the residuals of the BMT18 model when applied to seismic slope displacements calculated using the NGA-Sub-based intraslab earthquake database. These discrepancies highlight the need for an intraslab-specific seismic slope displacement model, which is the primary motivation for this study. Negligible calculated seismic slope displacement values (i.e., smaller than 0.5 cm) are of little engineering interest and can be considered as “zero” for practical purposes. Bray and Macedo (2019) present an in-depth discussion of the benefits and practical implications of using a mixed random variable regression model to estimate seismic slope displacement. In addition to not letting order of magnitude differences in negligible displacements affect the uncertainty of the estimate, the estimate of the probability of negligible (i.e., “zero”) displacement can be used as a screening analysis. It can also be combined with the estimate of ‘non-zero” seismic slope displacement to capture the overall uncertainty of the estimate at meaningful calculated seismic slope displacements. The mixed random variable model has the form of Equation 1: ((( 𝑓" (𝑑) = 𝑃(𝛿(𝑑 − 0.5) + (1 − 𝑃()𝑓 " (𝑑)

(Equation 1)

Where 𝑓" (𝑑) is the probability density function (PDF) of D, 𝛿(𝑑 − 0.5) is the Dirac delta function that gives a value of 1 when d = 0.5 and 0 otherwise, 𝑃( is the probability of a negligible D (i.e., D ≤ 0.5), and ((( 𝑓" (𝑑) is the PDF of D for D > 0.5. The total PDF 𝑓" (𝑑) has a finite probability mass to account for negligible D values for D ≤ 0.5 (i.e., for “zero” displacement) and has a PDF for D > 0.5 that models the distribution of non-negligible D values. Consistent with the mixed random variable characterization of D, a model is first developed for estimating the probability of negligible displacements (i.e., P(D = 0)), which is expected to depend on the slope’s strength and stiffness and the seismic demand, which can be represented by 𝑘1 , 𝑇3 , and 𝑆5 (1.3𝑇3 ), respectively (Bray et al., 2018). As found by Bray and Travasarou (2007), Bray et al. (2018), and Bray and Macedo (2019), the optimal individual intensity measure of earthquake shaking in terms of efficiency, sufficiency, and availability in robust ground motion models (GMMs) was found to be 𝑆5 (1.3𝑇3 ). A logistic regression (Hosmer et al., 2013) is used to develop a model that estimates P(D = 0) as a function of 𝑘1 , 𝑇3 , and 𝑆5 (1.3𝑇3 ) as shown in Equation 2.

19

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA ?

9(":;)

𝑙𝑛 80 µm. The volume percentages of each particle species are calculated and illustrated in Figure 11. (It should be stressed here that the volume percentage of each particle species is different from the concept of the local solids content of each particle species as the latter incorporates the information of the total solids content of all species.) As shown in Figure 11, the volume percentage of particles with sizes larger than 80 µm decreased significantly in the hindered settling zone due to segregation, as reflected by the dark blue curves. (Dashed lines represent the original volume percentage.) These particles accumulated at the bottom of the settling column during the test. The second largest particle species, represented by the red curve, also exhibited segregation and concentration in the height range of 5 to 9 cm. This result aligns with the illustration of the differential settling behaviour in Figure 1. Due to the

163

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA segregation of these two large particle species, the particles with sizes smaller than 40 µm demonstrated high volume percentages in the hindered settling zone. When focusing on the influence of the suspension’s pH value on the differential settling behaviour, the particle species volume percentage profile is effective in illustrating the changing trend. Taking the second largest particle species (40 to 80 µm) as an example, an increase in the suspension’s pH from 6 to 12 resulted in a less evident concentration of this species at a height of 5 to 9 cm. The general volume percentage of this species in the hindered settling zone approached the initial value when the pH was 12. These observations indicate the inhibition of segregation of this particle species under the influence of pH condition.

Figure 11: Volume percentage profiles of five particle species

Influence of pH condition on the aggregation of the particles In accordance with the description of the differential settling model, the settling velocity of the interface reflects the movement of the smallest particle species in the suspension. Since increasing the pH accelerated

164

INFLUENCE OF THE SUSPENSION CHEMICAL PROPERTY ON THE DIFFERENTIAL SETTLING BEHAVIOUR OF COPPER TAILINGS SUSPENSION the settling of the interface in our tests, the result highlights an important fact: the pH condition of the suspension changed the settling velocity of the fine particles. Analyzing the solids content profiles, a minor difference in the overall solids content of the hindered settling zone was noticed when the pH was increased from 6 to 10. Moreover, a significant increase was observed as the pH was further increased to 12. Taking into account the solids content data, it can be concluded that the change in the settling velocity is attributed to the alteration in the aggregation degree of the fine particles present in the suspension. By modifying the aggregation degree, larger clusters or flocs, which serve as the basic settling units in the liquid phase, were formed. As a result, these larger clusters settled at higher velocities, even with similar or slightly higher solids content. Considering the overall settling behaviour of the suspension, increasing the pH value of the suspension resulted in a decrease in the relative settling velocity of the particle species in the copper tailings, and it inhibited the segregation of coarse particles. Since the settling velocity of the fine was increased, it remains uncertain how the settling velocity of the coarse particle species is influenced by the pH condition. To address this question, a potential solution lies in performing a force balance analysis on the largest particle species, as previously suggested by Li and van Zyl (2022; 2023).

Conclusion and recommendations In this study, the differential settling behaviour of a copper tailings suspension has been investigated by analyzing the influence of pH conditions. The results revealed two significant findings regarding the sedimentation of the suspensions. Firstly, altering the pH value had a direct impact on the settling velocity of the supernatant-suspension interface. More specifically, increasing the pH value from 6 to 12 enhanced the settling velocity of fine particle species by promoting the aggregation degree of the suspension. Secondly, the pH condition of the suspension influenced the differential settling behaviour. It is observed that increasing the pH value suppressed the segregation phenomenon of coarse particles. This suggests that changes in the aggregation degree of the suspension can significantly affect the settling behaviour. Further experimental validation is recommended to confirm and expand upon these observations for various types of tailings. In future studies, it would be beneficial to investigate other chemical properties of the suspension, such as conductivity and the addition of flocculants. These factors may have additional and more significant effects on the differential settling behaviour of tailings suspension. Overall, this research contributes to a better understanding of the differential settling behaviour of copper tailings suspensions and highlights the importance of pH control. The findings can serve as a foundation for further studies and assist in the development of efficient settling strategies for tailings management.

165

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

References Concha, F., Lee, C.H. and Austin, L. 1992. Settling velocities of particulate systems. Part 8. Batch sedimentation of polydispersed suspensions of spheres. International Journal of Mineral Processing 35(3–4): 159–175. Fitch, B. 1962. Sedimentation process fundamentals. Transactions of the Society of Mining Engineering of Amie, 223(2): 129–137. Greenspan, H.P. and Ungarish, M. 1982. On hindered settling of particles of different sizes. International Journal of Multiphase Flow 8(6): 587–604. Hanson, G. 1985. Modeling of the settling of thick slurries. PhD thesis, Iowa State University. Kynch, G.J. 1952. A theory of sedimentation. Transactions of the Faraday Society 48(2): 166–176. Li, Y. and van Zyl, D. 2022. Hindered settling of flocculated multi-sized particle suspension, part I: Segregation mechanism of non-flocculated particles. Powder Technology 407, 117683. Li, Y. and van Zyl, D. 2023. Hindered settling of flocculated multi-sized particle suspension, part II: Experimental study of particle segregation based on copper tailings suspension. Powder Technology 415, 118154. Lockett, M.J. and Bassoon, K.S. 1979. Sedimentation of binary particle mixture. Powder Technology 24(1): 1–7. Maclver, M.R., Hamza, H. and Pawlik, M. 2021. Effect of suspension conductivity and fines concentration on coarse particle settling in oil sands tailings. The Canadian Journal of Chemical Engineering 99(9): 2024–2034. Mirza, S. and Richardson, J.F. 1979. Sedimentation of suspensions of particles of two or more sizes. Chemical Engineering Science 34(4): 477–454. Tiller, F.M. and Khatib, Z. 1984. The theory of sediment volumes of compressible, particulate structures. Journal of Colloid and Interface Science 100(1): 55–67.

166

Chapter Two

Application of GISTM

167

168

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Utilized Approaches in the Application of GISTM to Legacy Tailing Facilities Matthew S. Gore, Geosyntec Consultants, USA Ali Nasseri-Moghaddam, Geosyntec Consultants, Canada

Abstract The international mining and related consulting industries are currently amid a major shift in approach to managing tailings as they adjust to the Global Industry Standard on Tailings Management (GISTM). The GISTM’s clear and admirable intent is “striving for zero harm to human health and the environment.” Prior to GISTM, most of the mining companies would have claimed that they operated in a safe manner and that their tailings storage facilities (TSFs) were safe. However, recent failures have provided evidence to the contrary. Consequently, GISTM has shifted the mining industry into debates on numerous topics, including the meaning of safe and associated risks, a robust plan, ALARP (“As Low as Reasonably Practicable”), what is minimum GISTM conformance, how highly technical matter should be communicated with the public without raising unnecessary concerns, and so on. To be compliant, mining companies allocated significant resources to the process while continuing to strive to be a successful business industry. One particularly challenging scenario is the implementation of GISTM to legacy (closed) facilities, which is the focus of this paper. The challenge presents itself in many ways, from identifying and performing gap analyses on facilities that have limited to no data, to attempting to assess decades old methods/construction practices, a form of presentism, to implementing modern standards. This paper will present several examples of legacy facilities that are in the process of being changed to achieve GISTM conformance. The paper will discuss some of the difficulties for each site and share utilized approaches to debated conditions. This paper intends to educate those less familiar with GISTM at closed legacy sites, and to be a catalyst for discussion among peers.

Introduction The Global Industry Standard on Tailings Management (GISTM) requires mining companies to be responsible for the management of a tailings storage facility (TSF) from planning to operations to postclosure. The responsibility stems from the overarching goal of “striving for zero harm to human health and the environment.” This requirement may seem straightforward for a mining company to achieve, if the

169

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA company designs, operates, and closes the TSF, and maintains ownership with management responsibilities throughout the entire lifecycle. However, this GISTM requirement leads to unique and sometimes tricky situations for a mining company when the TSF is a legacy facility that joins the company’s portfolio of assets through an acquisition. In these situations, there are numerous conditions that make GISTM conformance a difficult and complicated pathway to navigate. This paper will discuss the authors’ definition of a legacy facility, experiences and challenges encountered while working on legacy facilities, examples shared to illustrated experiences, utilized approaches to addressing the examples, and a conclusion. It is noted that aspects of the discussed situations may be captured during formal/structured risk assessment procedures. However, in general some of these topics may fall beyond the scope of risk assessments and may be overlooked. One of the objectives of this paper is to emphasize the important role of engineering judgment and experience in managing project risks.

What is a legacy facility? Existing guidelines are either quiet on legacy facilities or provide limited information on the topic. In this paper, a legacy facility is defined as “a facility that has joined a company’s asset portfolio through acquisition where management and liability for the facility are transferred to the acquirer. In many cases, the acquirer receives the facility in some state of closure, where they never had operated the TSF or had input on the design.” This paper will discuss experiences working on legacy facilities, some of the issues that arise, and utilized approaches, where available. Names and locations of the sites will not be shared due to client confidentiality.

Challenges with legacy facilities Limited to no information/data Acquisitions of legacy TSFs can result in responsibilities for facilities with limited to no data. Legacy facilities typically do not provide mining companies revenue, resulting in a “minimalist” mindset, where the asset management and owner leadership approach is to minimize the costs associated with maintenance and eventual post-closure or relinquishment to the greatest extent possible. Because of this fact, legacy facilities are set as a lower priority for allocating capital expenditure compared to operating mining facilities with a company portfolio. To provide perspective, Table 1 provides the authors’ reflections on some of the experiences of limited data when dealing with legacy facilities. Note that this list is not exhaustive.

170

UTILIZED APPROACHES IN THE APPLICATION OF GISTM TO LEGACY TAILING FACILITIES Table 1: Experiences of limited data for legacy facilities Topic

Details and descriptions

Dams and appurtenant

Original designs and design basis information, asbuilts, deviance records, geotechnical data, historical performance records, historical monitoring information

Waste Information

Boundaries and locations, operation and deposition records, tailings properties

Loading conditions

Seismic data, previous preloading conditions

Water

Groundwater, stormwater, and/or flood data, controls, drainage gallery (galleries), ground improvement conditions

Environment

Environmental and ecological Impacts: geochemical data, historical environmental/ecological impacts, species at risk

Facility history

Operations, lift schedule/increments, original design intent, engineer intentions, changes to processing, changes to tailings placement, changes to tailings saturation, etc.

There are numerous issues related to having limited to no information. To start, owners and asset management teams cannot fear or change what they do not know about their sites. Surprises are typically not good for the mining and engineering industries. GISTM requires an understanding of every element of the facility to provide proper safety and to limit risk. Lack of information can lead to a significant level of reflection and work to obtain missing information and data. However, there are some types of information for understanding a site’s design that are difficult to replicate, replace, or create after a site has reached legacy status, including as-builts, construction information, and/or the site history. This type of missing information can lead to the misunderstanding of the intent of TSF features, the TSF characteristics or observed behaviour, or even affect the dam break analysis and consequence classification. Lastly, lack of information can lead to an expensive site investigation or multiple rounds of investigation to properly address the requirements of GISTM and fill in gaps, leading to significant costs for the owners of these legacy facilities.

Historical significance Legacy facilities tend to be at least a few decades old (i.e., 50 years or older) and can have local, regional, and/or national importance depending on the communities’ connections to mining. There can be interested parties like historical societies, archeological communities, mining enthusiasts, and government agencies with regulations that can lead to preservation needs. Legacy facilities can also have connections to first

171

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA nations and indigenous people, specific communities called out by GISTM for protection and consideration. Historical conditions present issues to owners in several ways. Land, structures, and/or features of the facility may have value to the surrounding communities and stakeholders. In some cases, local, regional, and/or national regulations can lead to limitations or preventions for the facility owner for management or remediation means due to protections. Additionally, regulations due to historical significance can lead to added responsibilities and costs for site management. Stakeholder engagement can also be difficult, depending on site-specific conditions, scenarios, opinions, and desires. Consider recent situations presented in the public media forum where indigenous or first-nations community members have emotional reactions towards the mining community. There are also historical examples where an elevated level of emotion can be felt from community members to the preservation of features like valued buildings, railroads, historical pathways/travel ways, and monuments.

Social and disclosure impacts Legacy facilities, due to their history and changes over time, can have positive or adverse impacts on the surrounding society. A positive impact can be remining tailings for extraction of the originally desired product or extraction of new products now in high demand. There are several sites that the authors have worked with where changes in technology have led to evaluating potential for additional extraction of gold and copper. Or a tailings pile is evaluated for rare earth metals to meet the needs of the new green initiatives emerging around the world. The challenge with this positive can be how to approach management of the site under GISTM with the dictated deadlines while managing the different internal stakeholder interests. There is also the issue of transferring the management of the site from a closed condition to a new, active facility. Some examples of negative impact from the social perspective include (i) the surrounding neighbors not knowing a tailings pile is in their backyard and (ii) mine development may have not considered ancestral, indigenous, or first-nations perspectives or heritage conditions at the start of the mine. For (i), disclosure of a legacy tailings facility to a community that was not aware of its existence can lead to a host of issues related to environmental impact, local health (i.e. – previous recreational areas deemed unsafe due to water quality or dust), loss of recreational areas, real estate values, and insurance, to name a few. Living near a legacy facility also has the potential to raise the community concern/fear for their personal situations. For (ii), heritage protections for ancestral, indigenous, and first nations peoples have become more inclusive and extensive, leading to better education among these people and the mining community. This outreach is a positive, but the general negative history and/or site-specific conditions can lead to a negative shadow from the broader society over the owner of a legacy facility that impacted these people in the past or is currently impacting the community.

172

UTILIZED APPROACHES IN THE APPLICATION OF GISTM TO LEGACY TAILING FACILITIES Disclosure can be a very delicate task for owners, especially given the various impacts a TSF may have on stakeholders. Consider the impact of disclosing a legacy site to the community as an extreme dam that has the potential to hurt or kill many in the community, which the community did not know existed, or having to try to walk that disclosure back when a subsequent review indicates that the risk of the dam to the community is not as high as first disclosed. Undoing such a situation would be difficult to impossible, even with robust, state of practice engineering assessment results, since the initial disclosure would already have impacted the community and trust would have been damaged. It is a scenario that the authors have discussed with several owners when considering the negative impact a legacy site could have for them and the surrounding community.

Land ownership issues and non-GISTM conforming agencies Legacy facilities can be discovered to be owned by one entity, but safety and environmental responsibilities fall to another entity. There are also cases where a site controlling agency or owner does not adhere to GISTM. Government agencies, for example, have their own standards, regulations, and guidance that are independent and can be considerably different from GISTM1. An example would be the US Nuclear Regulatory Commission. In this case, their regulations for the preferred method of maintaining tailings to protect human life and the environment from radionuclides from mill tailings can differ from GISTM. Additionally, GISTM conformance is a voluntary situation where ICMM members have agreed to adhere to the standard. Not all mining companies adhere to the guidance from ICMM or GISTM. The mining industry knows that some mining companies do not adhere to GISTM. For context, here are some examples of land ownership issues to consider: • Land ownership by a private or government entity but safety and/or environmental responsibility held by a mining company. This example can be in the form of an Environmental Protection Agency issuance of CERCLA Superfund status for public land that used to be a mine site. • Land ownership and/or mine operations may be held by one mining company, but some element of the GISTM may be held by another mining company. This second responsible mining company may have operated the mine in the past, or may never have operated at the mine, receiving responsibility through acquisition. This example can also relate to joint ventures where one company/owner may hold responsibility for environmental cleanup of legacy structures while the other company/owner is continuing to mine the asset and holds responsibility for the TSF. 1

Government or local authorities may have less, or more strict requirements compared to GISTM. In the case of less

restrictive requirements, a GISTM participating company will have to implement GISTM and defend their choices in the regulatory review. For more restrictive requirements, a GISTM participating company will have to adhere to the authority’s additional requirements.

173

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Mine or waste facilities may be owned by a mining company, but the facility may be surrounded by land controlled by another entity, sometimes government-controlled entities that define their own environmental controls and standards. • Government or public entity has established access to privately owned, closed mining land under non-contractual agreements. • Government may control land and not subscribe to GISTM. This example can relate to land controlled by the federal government (e.g., US Forestry or US Bureau of Land Management land) where the government restricts the access, and a private mining company does not have the authority to force access to the site or force implementation of GISTM at a site under their control. Some of these situations can lead to control, regulation, and/or stakeholder interests that may not align with GISTM. These situations can also lead to complicated relationships and approaches to addressing legacy facility situations. Along with these difficulties, a mining company may find itself dealing with resistance to work at the site (think access issues for investigation or monitoring instrumentation) or approval of certain types of remediation or closure activities. These situations can even see a complete limitation on instituting GISTM conformance requirements.

Deposition considerations Legacy facilities may have been deposited in a manner that is not considered current state of practice. These types of deposition can include: • Slurry deposition into natural bodies of water (such as an ocean or river). • Slurry deposition downslope with limited to no containment (such as an alluvial fan down a mountain slope). • Side or end dump onto angle of repose sloped stacks. Assessing these conditions under the microscope of presentism can lead to questions about the operator’s intent at the time of construction. There can be criticism of the former and current owner’s actions in the handling of this waste, given today’s state of practice techniques, and prescriptive deadlines may be leading to pressure to address the issues. However, as with all the paper’s discussion points, legacy issues do not typically follow prescriptive solutions. Each situation and solution will be unique.

Land use change There have been experiences where an entire TSF or a volume of tailings have been placed as structural fill for several reasons, including mine expansion or foundation material for structures or infrastructure. Tailings were considered cheap and locally available building materials. There have also been situations where unidentified TSFs or tailings piles were subjected to construction through or over the tailings. In both

174

UTILIZED APPROACHES IN THE APPLICATION OF GISTM TO LEGACY TAILING FACILITIES situations, the tailings provided some benefit to a structure or infrastructure. However, despite the benefit, there could be issues related to GISTM conformance and potential environmental impact that may result in reclamation/GISTM activities contrasting or conflicting with the current conditions.

Examples To clarify the situation, this paper provides examples in the following sections, along with utilized approaches to address the issues.

Example 1: “Start from scratch” A client acquired a TSF constructed in the early 19th century that had the designation of “active closed” with no design documents, as-built drawings, or history of construction stages.

Utilized approaches for Example 1 GISTM requires significant work be performed to provide the accountable individuals, that is the Asset Manager (AM), the Engineer of Record (EoR), and the Accountable Executive (AE), with a level of confidence with the safety and risks associated with the TSF in meeting conformance requirements. For this example, a substantial site investigation and monitoring program was developed and was being implemented to gather missing geotechnical, hydraulic/hydrogeologic, geochemical, and other data to develop a Design Basis Report (DBR), geotechnical assessments, hydrology and hydraulics (H&H) assessments, seismic/liquefaction assessments, and other required documents. The management and engineering teams had struggled with limited time for completion (GISTM set completion deadlines), limited resources (drillers, contractors, and other resources were spread thin due to many mining and engineering companies needing similar work worldwide), and client budgetary constraints, given that legacy TSFs are not profit-generating facilities. In addition, the age of the facility and local ties to mining had the management team balancing the GISTM requirements with the potential of preserving elements of the old mine, like a historical railroad and supporting buildings. These structures were considered culturally historical with a local historical society wanting preservation. In addition, the management team continued to struggle with gaps that were difficult to fill, like a lack of design plans, construction records/as-builts, and construction quality control/quality assurance (CQA). In this case and others, the best outcome that can be achieved is to indicate that no adverse issues were found during recent investigation work while the owner moves ahead with site management alongside a perpetual gap. GISTM provides a good framework for risk management for sites with this type of limited information. A potential failure mode assessment (PFMA) developed a list of credible failure modes that will be assessed further by a scheduled semi-quantitative risk assessment (SQRA). These risk assessments

175

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA are supported by the development of an Operation, Maintenance, and Surveillance Manual (OMS), Emergency Response Plan (ERP), related Trigger Action Response Plans (TARPs), and a monitoring program. All of these are being developed to manage the risks at the site in a comprehensive manner. The owner has required these elements of the risk management to be considerably robust to counter the unknowns and lack of information for this site. Additionally, the roles of the EoR and senior independent technical reviewer (SITR) have provided the owner with multiple levels of professional engineering judgement in assessing the gaps at the site and developing a good approach to minimizing risk.

Example 2: “Surprise – you own it!” A client was made aware of a 90-year-old TSF through government assessment of responsible parties for environmental impacts from the site. It turns out there are many of these types of sites around the world. This example is a site closed by a mining company that was eventually acquired by the client. Once closed, the previous mining company relinquished the land to a local, private entity. The surrounding land around the TSF is a government protected area. Environmental impacts from the TSF were observed and responsibility was discovered to be on the client.

Utilized approaches for Example 2 One of the important considerations for this site was balancing the desire of the government protecting the surrounding land with the GISTM conformance. The process has included discussions about the “As Low as Reasonably Practicable” (ALARP) determination of the site with comparative balancing between the GISTM requirements and the government regulations/standards. One example of this work is the client needing to balance the desire of the government entity for natural and esthetically pleasing streams/creeks around the TSF and the GISTM requiring stormwater control designed for a given flood return period event based on consequence classification. The current, natural-looking streams’ design was part of the accepted Record of Decision (ROD) agreed upon prior to GISTM to meet environmental requirements by the government agency. The situation is quite complicated when considering the GISTM requirements of meeting a 1 in 10,000 rain event for safe closure. Protection against such an event at this site would require considerable armouring or unnatural appearing structures to control the water at that level. The ongoing assessment is considering where the site rests currently in terms of being ALARP. This case is a good example of a complicated balancing exercise between the desires of the impacted stakeholders, risk management for the mining company, and achieving conformance with GISTM. Risk management for this site included a PFMA and a company-developed quantitative risk assessment that are leading to probabilistic assessment of the failure modes.

176

UTILIZED APPROACHES IN THE APPLICATION OF GISTM TO LEGACY TAILING FACILITIES This site has multiple levels of professionals assisting in management that include the EoR team and Independent Technical Review Board (ITRB). Under the direction of the ITRB, the EoR team is undertaking advanced deformation modelling of the TSF to evaluate the potential failure modes, including erosion potential for the 1 in 10,000 rain event to understand the potential impact to the TSF stability and surrounding environment. This process leads to considerable professional engineering judgement collaboration between the owner engineers, EoR team, and ITRB. It is interesting to note that the regulating government standards dictated minimum requirements, cleanup methodology, and design review acceptance/rejection power for remediation, irrespective of the GISTM standard. There are some issues that relate between the government standards and GISTM. But again, there is a need for balancing the various requirements, standards, and desired design methods to be successful in managing this TSF. The key word for this example was balance: balancing multiple stakeholders, balancing multiple regulators and different standards, and balancing differing interests to accomplish a sustainably safe and well-maintained facility.

Example 3: “Oh, that is tailings?” A stack of tailings was discovered by a client through an exhaustive review of historical records for acquired assets. The history of the site included the local transportation agency building a roadway along an abandoned mine access road through the abandoned mine area and the tailings pile. The pile is a dry stack, and the regional area is semi-arid to arid, limiting water being introduced to the stack. The client had acquired the asset through acquisition for production at other mine sites and had developed limited management measures. In this situation, the tailings were side-dumped and end-dumped into a small mountain ravine with the tailings remaining in place after the facility was abandoned. The pile is considered dry stack with no observed or known artesian activities around or under the stack, and the regional area is semi-arid to arid, limiting water being introduced to the stack through precipitation.

Utilized approaches for Example 3 In this situation, the tailings were side-dumped and end-dumped into a stack on a mountain side slope, remaining in place after the mine operations ceased. There are several determinations needed for this stack upon discovery of the structure. The determinations include the volume/height of the tailings, history of the stack (if any additional information is available), geochemical conditions of the tailings, and potential environmental impact, if any. This pile also has social impact with neighboring residents using the pile as part of their personal ATV/off-road motorcycle playground. Parts of the abandoned mine, including the tailings stack, have shown evidence of being used as a dumping ground, leading to potential impacts by outside sources.

177

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA This example shows the complexity of dealing with an abandoned legacy facility with disjointed, inconsistent, and limited management activities. The client must determine if the tailings pile is a TSF under GISTM. Even if it is determined to not be a GISTM regulated TSF, the impact of the pile on the environment needs to be assessed. Additionally, stakeholders need to be engaged to discuss competing interests, including the roadway potentially using tailings as a foundation, the use of the area for recreational purposes, and the potential culturally historical significance of the abandoned mine. As a positive, GISTM has accelerated these discoveries for some clients and the conversations needed to determine proper maintenance and management moving forward. This example is early in the stages of determination, but the key is that the client and assigned engineer (potentially eventually reaching EoR status) need to develop a clear understanding of the situation on the geotechnical, social, and regulatory sides to build a proper methodology for minimizing risk and elevating safety of the pile to ALARP. Risk management will start with establishing the potential failure modes expected for the stack and environmental risk associated with the tailings.

Example 4: “Maybe, maybe not” A client had a legacy facility with a small dam at the property’s edge that retains water intermittently for annual agricultural and recreational purposes. Through a historical review, it was discovered that tailings may be within the upstream catchment of the dam due to depositional methods prior to construction of the containment berms at the identified TSF located a substantial distance upstream of the small dam. Like the previous example, there was a need to determine the existence of tailings and evaluate the potential environmental impacts if the tailings exist. In addition, this situation needed to be evaluated to establish if the dam is acting as a TSF or other type of structure, i.e., water storage facility (WSF).

Utilized approaches for Example 4 The work included a review of historical information dating back over 100 years and establishing the structure’s initial intent and the facility’s current intent. A site investigation was developed and performed to evaluate the potential existence of tailings upstream of the dam. The review determined that the original intent and current intent of the dam is to function as a WSF for agricultural purposes with a secondary function as recreation for the local community. The site investigation found tailings upstream of the dam, but it was not holding them in place. Instead, the tailings were discovered to be capped by organics and natural soil that had covered the tailings over a prolonged period and no tailings were exposed at the surface. Based on discovered information and the geotechnical investigation, the dam does not appear to be and was subsequently not classified as a TSF under GISTM. However, the client plans to finalize the potential environmental impact of and develop a formal management methodology for the tailings as part of the upstream TSF under GISTM conformance.

178

UTILIZED APPROACHES IN THE APPLICATION OF GISTM TO LEGACY TAILING FACILITIES Consulting with a client dealing with such a facility means that work includes history research and applying environmental/engineering knowledge while being patient with the client as they work through what they want to do with each unique situation. Risk management associated with the tailings is being addressed by the client by including them in the management of the upstream TSF, maintaining their commitment to GISTM’s requirement of zero harm to humans and the environment. Geotechnical and environmental risks will be included in the TSF PFMA and SQRA, an ongoing environmental impact assessment for the TSF is being adjusted to include assessment of the capped tailings, and a study is planned to evaluate the best management approach for the capped tailings. The situation has provided excellent opportunities for collaboration between the TSF EoR and client specialists to share experiences and knowledge during classification of the capped tailings and determining where to manage the tailings within the asset management system. This collaboration allowed for a broad spectrum of professional judgement to be applied to a complicated situation for the client.

Concluding remarks The intention of this paper was to share experiences with examples working on legacy facilities, hopefully providing some of the unique situations and solutions needed to address the conditions at these sites under the lens of GISTM. In some cases, work involves multiple stakeholders, regulators, owners, and interested parties, and thus it is necessary to have good understanding of multiple regulations, as well as communication among all parties, to determine the best approach for a given site. Sometimes a lack of knowledge or understanding has left a dangerous void that must be addressed. In these cases, there was a requirement for ongoing and extensive investigation and assessment work. The client and consultants need to be patient with the investigation process as deadlines loom, knowing that the goal is to provide a safe, minimal risk situation for involved parties and the community at large. And this point needs to be remembered for all sites involving tailings that are dealing with GISTM. Patience, communication, a proper balance, and the desire to have appropriate and ample information for making decisions are key to success in dealing with legacy facilities under the spotlight of GISTM.

References ICMM (International Council on Mining and Metals). 2020. Global Industry Standard on Tailings Management (GISTM).

179

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

180

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Deviance Accountability – A Register-Based Approach Martyn Bryan Willan, WSP Canada Inc., Canada Jennifer Brash, Teck Coal Ltd, Canada Andy Haynes, WSP Canada Inc., Canada

Abstract Deviance accountability reporting for tailings storage facilities was included as Requirement 6.5 of the Global Industry Standard on Tailings Management (GISTM) in 2020 (GTR, 2020) by the introduction of the Deviance Accountability Report (DAR). While this implies a formal, report-style document, specific details and guidance on the format and content of the DAR are limited, thereby allowing individual organizations to determine the approach best aligned with their own stewardship and governance systems. Teck Coal Limited (Teck Coal), in collaboration with WSP Canada Inc. (WSP), has developed a register-based Deviance Accountability system. This register-based approach aims to provide a practical solution to tracking, assessment, review, and recording acceptance of deviances (both individually and cumulatively) without the limitations associated with formal, report-style documents. In addition, the register-based approach minimizes the delay between identification of a deviance and the assessment of associated impacts, enabling the DAR to be a valuable tool in decision making for active mine sites. In this paper the authors will provide an overview of Teck Coal’s register-based approach to deviance accountability and review the identified benefits. Initial feedback on the implementation of the system following roll-out across Teck Coal’s operations will also be discussed.

Introduction The documentation and review of deviances, from design criteria and basis as well as operational procedures and stewardship standards, have been part of effective change management processes for the mining industry for many years. The outcomes of these reviews were often recorded in multiple documents (sometimes by different teams or authors) including annual facility performance reports (also known as dam safety inspections), construction verification/construction record reports, or as part of internal change management processes or regulatory requirements/submissions. In 2020, the Deviance Accountability Report (DAR) was incorporated as Requirement 6.5 of the GISTM (GTR, 2020). The GISTM provides limited guidance or specific details related to the format and

181

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA content of the DAR, and it is therefore left up to individual organizations to determine the best approach in line with their own stewardship and governance systems in consideration of published guidance (for example Brown et al. (2022). Many have interpreted that the DAR should be a formal, report-style document. However, formal, report-style documents are “static” documents, the content and outcomes of which are fixed and representative of the information available at the time the report is developed. Reports also take significant effort to compile, review, and issue, meaning that they are often immediately out of date. As such, formal, report-style DARs may present limitations for communicating and managing deviations in the dynamic nature of a changing environment, such as the mining industry, where the timely understanding of the impacts of proposed changes is essential to inform operational decisions. As part of their journey towards GISTM conformance, Teck Coal, in collaboration with WSP, has developed a register-based approach to deviance accountability. This register-based approach aims to provide a practical tool for tracking, reviewing, updating, and documenting acceptance of deviances (on an individual and cumulative level). The DAR register is a “living” document that consolidates pertinent information in a single, easily updated and searchable spreadsheet, allowing for timely recording of a deviance, assessment of its associated impacts on facility safety, and the implementation of any necessary risk mitigation or change management processes.

Teck Coal Deviance Accountability register The Teck Coal DAR register was developed with the intent to be easy to use and to require minimal administrative effort, while also providing useful input to site decision making. The DAR register provides a “live” record of deviances, and documents the assessment of associated individual and cumulative impacts. It also tracks actions related to the management of deviances, identifies any resulting risk impacts, and documents the review and acceptance for each deviance.

Deviations captured The ICMM Conformance Protocols, which accompany the GISTM, note that the DAR should address the cumulative impact of “material change” (ICMM, 2021). The GISTM (GTR, 2020) defines a material change as changes which are “important enough to merit attention or having an effective influence or bearing on the determination in question.” This effectively means that the determination of what constitutes a material change is up to the judgement of the Responsible Tailings Facility Engineer (RTFE) and the Engineer of Record (EoR), with consideration of any available corporate guidance. Teck Coal has provided the RTFEs and EoRs with preliminary guidance regarding types of deviations to capture. Examples include, but are not limited to:

182

DEVIANCE ACCOUNTABILITY – A REGISTER BASED APPROACH • Deviations from the design intent, criteria, or basis. • Changes to tailings production levels or water inputs. • Discovery of or changes to the condition of foundation units. • Deviations from the established and approved operation processes (such as pond levels, tailings deposition, etc.). The DAR register should also include those deviations recorded in other reports (such as construction record and annual facility performance reports). In this way, the DAR register represents a comprehensive record of deviations and an assessment of their cumulative impacts. However, it is not the intention of the DAR register to replace or supersede information contained within other reports. Where applicable, reference to key documents can be incorporated within the DAR register to indicate where more information may be available.

Roles and responsibilities The GISTM (Req. 6.5) has defined the following roles and responsibilities for key tailings stewardship personnel associated with the DAR register: • EoR: Prepares a periodic DAR. • Accountability Executive (AE): Approves the DAR. In addition, Teck Coal has defined the following roles and responsibilities associated with the DAR: • RTFE: Owns the DAR register and is responsible for identifying deviations, notifying the EoR, executing the associated procedures to update the DAR register, ensuring the register is maintained up-to-date, and for obtaining the required inputs, reviews and approvals. The RTFE is also responsible for communicating any resulting action items to those responsible for implementation. • Independent Tailing Review Board (ITRB): Reviews and provides opinions on the DAR register annually, at a minimum.

Register format The Teck Coal DAR register is spreadsheet-based, which allows for data filtering and sorting. The information captured within the Teck Coal DAR register includes: • ID: a unique identifier to allow for ease of tracking. The ID includes embedded details of the site and structure. • Date: date the deviance occurred. • Structure: a unique structure identifier, for facilities with multiple structures. • Category of deviance: the main category of the deviance, chosen from the following options:

183

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA •

construction

•

surveillance

•

operations

•

personnel

•

design

•

event

•

other.

• Description of deviation or event: description of the change that has occurred in comparison with the baseline condition/intent. • Documentation reference: reference to any documentation related to the deviance (studies, assessments, reports etc.) . •

Assessment of deviation – two columns provided for each deviance: •

Individual: commentary on the implications of the individual deviation on a shortterm basis. Any residual impacts that may remain after actions are taken should also be noted.

•

Cumulative: commentary on the implications of the deviation (as detailed for the individual assessment), assessed cumulatively with all previous deviations for the same structure.

• Failure mode impacted: identify the failure mode(s) impacted by the deviation. •

The inclusion of the potential failure mode(s) related to the deviances is considered important to enable the quick filtering of all deviances and enable the RTFE and/or EOR to assess individual or cumulative impacts of these changes by failure mode.

• Description of failure mode impact: brief description of the impact of the deviation on the selected failure mode(s). • Required actions: describe any associated actions identified to manage impacts related to the deviance (individual or cumulatively). Actions are assigned to a responsible individual and include timeline for implementation. • Actions completed: list any actions taken and note any not completed and why. • Reviews: separate columns are provided to record the review (initials and date) of each deviation by identified personnel (see “Roles and responsibilities” section)

Establishing the baseline for deviation Inherent in the identification of deviations is the adopted basis for comparison, i.e. what is being deviated from? As such, it is important to establish the baseline conditions for the DAR register. This is relatively

184

DEVIANCE ACCOUNTABILITY – A REGISTER BASED APPROACH straightforward for facilities that have up-to-date design, site characterization, and performance data. However, it can be a more complex decision when considering facilities with limited available background data and reporting (such as older facilities). The selection of the baseline condition for the DAR register must be clear and documented, including details of the date, design, condition, performance objectives etc. that the DAR register is capturing deviances from.

Update and review of the deviance accountability register Following the identification of a deviance, the RTFE and EoR complete an assessment of the impacts, after which they complete the register entry. Where additional engineering work is required to inform the DAR register assessment or identify the impact of the change, the RTFE and EoR will work in collaboration to complete these works. The DAR register is hosted on corporate SharePoint sites, which allows for electronic tracking of revisions/edits. On a mutually agreeable frequency (at least an annual basis, or as otherwise needed in the opinion of the RTFE or EoR), the EoR provides endorsement of the Deviance Accountability register via formal documentation (e.g. a technical memorandum/letter or similar). Following endorsement by the EoR, and on at least an annual basis, a copy of the Deviance Accountability register is provided to the ITRB for their review and opportunity to comment. If required, the RTFE will arrange for a meeting with the EoR and ITRB to discuss any identified concerns. The DAR register is also provided at least annually to the Accountable Executive for their approval (per GISTM Req. 6.5). The Deviance Accountability register should also be reviewed as part of Dam Safety Reviews (DSRs) to provide a third-party independent review of the potential impacts from individual or cumulative deviances and the actions taken to address these.

Initial feedback on Deviance Accountability register implementation The Teck Coal tailings team has provided feedback on their experiences implementing the DAR register that also provides valuable insight on the practicality of the system in an operating mining environment. This feedback is also valuable for others who may be considering implementing a similar system.

Implementation program The implementation of the DAR register in Teck Coal was accompanied by a standard procedure document describing the purpose of the DAR register, key information to be included, the format and use of the DAR register, procedures for maintaining, reviewing and updating, and roles and responsibilities. A standard register template was also provided in excel format, including examples of completed entries, for consistency of application and use at each of the Teck Coal sites. In addition to the procedure document

185

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA and template, training on the DAR register was provided to the Teck Coal tailings team prior to implementation. These resources are all available to the Teck Coal tailings team for their future reference.

Useability and benefits Feedback received from the RTFEs is that the DAR register is easy to use and is proving a straightforward and valuable addition to the tools available for managing change. The site tailings teams are utilizing the DAR register for identifying changes (proposed or existing), preparing preliminary entries to the DAR register, and informing the EoR. Where needed, the EoRs are also providing supplemental commentary, studies or documentation, which are filed and referenced directly in the DAR register under the associated deviance entry. The DAR register is also being used to record the consideration and rationale for accepting or rejecting proposed changes (i.e., used to assess the impact of potential changes, not just those that have been implemented). Having these records and details easily accessible is anticipated to help with future staff transitions as well as to prevent “normalization of deviation.” Recognizing the benefits of the DAR register approach in providing one location to record deviances and the associated impacts, necessary actions and reviews, is a concept that can be applied more broadly. The tailings teams have expanded their usage to select water management structures.

Register format As described above, the Teck Coal tailings team has utilized the DAR register for recording considerations associated with proposed deviations, as well as those that have been implemented. There are significant benefits to capturing the rationale as to why proposed deviations were not adopted, as these can then be used as part of future change assessment. As such, the Tailings Teams have suggested that a “status” column or similar be added to the DAR register template to note whether the deviation is proposed, rejected, inprogress or complete (for example). This will be implemented in the next revision of the Teck Coal DAR register template. The inclusion of the potential failure mode(s) related to the deviances is considered important to enable the quick filtering of all deviances that may individually or cumulatively constitute a material change, and enable the RTFE and/or EOR to assess such impacts.

Interaction with other components of the Action Tracking System TSF governance and review systems generate many action items, including AFPRs, DSRs, ITRBs, internal auditing and review processes, among others. These action items may be tracked in a variety of systems or registers depending on the purpose/objective of the action item. The Teck Coal DAR register includes tracking of all action items associated with deviances, which provides clear and direct linkage between

186

DEVIANCE ACCOUNTABILITY – A REGISTER BASED APPROACH changes and impacts in one location. The distinct purpose of each different action item tracking system should be clear to the users, to avoid duplication of work and inconsistency. However, it has been noted that the DAR register should not become the place to record all action items, deficiencies, opportunities for improvement, or recommendations, unless these represent a deviation to the facility or the stewardship and governance system. In such cases, duplication of an action may be the best option to ensure it is tracked both as part of site action item tracking systems as well as the DAR register.

Interaction with other elements of the change management system Assessments completed as part of the DAR register approach can provide information useful to the implementation of other components of the change management system already established at Teck Coal. However, depending on the other components in place at each Operation (mine site), there is potential for the DAR register to duplicate, in whole or in part, existing change management registers. Therefore, at some operations, the DAR register may be the single entirety of the TSF change management system, whereas for others it may serve as a summary or component of the overall system. The place of the DAR register within an organization’s broad system of change management should be thoughtfully considered to reduce duplication of work, where practicable, and the purpose and objectives of each component of the system should be understood by all parties involved.

Conclusion Standalone formal reports are a common form of communication within the business world and can be highly effective when used in the right context, such as when presenting information from a limited set of authors to address a specific objective based on data wholly available at the time the work was completed (i.e. a “snapshot in time”). However, as “static” documents, they cannot incorporate new or developing data/information without extensive effort, often in the form of rewriting and reissuing the report. This can often lead to reports quickly becoming obsolete, with the risk that the findings and outcomes are “filed and forgotten” or may even be outdated immediately upon issue. In addition, formal reports require significant professional effort (by both those preparing and those reviewing the document), which, in the context of tailings stewardship, can divert valuable expertise away from tasks that could have a more direct impact on the management of facility safety and operations. This can become particularly relevant for reports requiring frequent and extensive updates, such as those that may be needed to satisfy the objectives of the DAR register in an actively operating mining environment. Register-based systems have been used extensively within the tailings/mining industry, for example documenting of facility risks (risk register), and have proven to be highly useful tools for the efficient collating of key information, outcomes of discussions/workshops and, where required, referencing relevant

187

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA studies or other information. In addition, as they are usually in electronic format, they allow for straightforward updating and revision tracking and for the easy searching and sorting of information by key attributes/parameters. Registries can also be maintained “live” and can therefore incorporate new information, data, or changes in knowledge in a time effective manner. This is especially useful where conditions can change quickly (for example, within operating mines or in high staff turnover environments such as in tailings stewardship), or where understanding the potential impact deviances is critical to the business decision making process. Teck Coal’s register-based approach to the DAR has been working well, with site teams noting the ease of use and benefits of having a comprehensive and “live” record of deviation assessments. In addition, the register-based approach minimizes the delay between identification of a deviance and the assessment of associated impacts, enabling the DAR register to be a valuable tool in decision making for active mine sites. The authors believe the use of a DAR register is a beneficial addition to TSF stewardship and that benefits can be achieved by integrating with other complementary systems such as risk registers, action item trackers, and change management processes.

References Brown, T., C. Priscu and A. Küpper. 2022. Deviance accountability reports: considerations for a practical approach. In Proceedings of the 26th International Conference on Tailings and Mine Waste. Denver, CO: Colorado State University: 814–823. Global Tailings Review (GTR). 2020. Global Industry Standard on Tailings Management. Accessed 06/04/2023 at: https://globaltailingsreview.org/wp-content/uploads/2020/08/global-industry-standard_EN.pdf International Council on Mining and Metals (ICMM). 2021. Conformance Protocols: Global Industry Standard on Tailings Management. Accessed 16/06/2023 at: https://www.icmm.com/website/publications/pdfs/environmental-stewardship/2021/tailings_conformanceprotocols.pdf?cb=21097

188

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023 | Vancouver, Canada

GISTM – Defining “Substantial” Conformance: Gold Fields’ Substantial Implementation of the GISTM at the Cerro Corona Mine in Peru and Tarkwa Mine in Ghana Louise McNab, Gold Fields Ltd, Australia Johan Boshoff, Gold Fields Ltd, Australia Javier Gutierrez, Gold Fields Ltd, Peru Edwin Zegarra, Gold Fields Ltd, Peru Diana Honores, Gold Fields Ltd, Peru Nathaniel Asifu Mensa, Gold Fields Ltd, Ghana Fernando Rodriguez, RDZ Consulting, Peru Josh Rogers, Stantec, Peru

Abstract The Global Industry Standard on Tailings Management (GISTM) provides a framework for the safe and responsible management of tailings, aiming to prevent catastrophic failures and mitigate environmental and social impacts. The GISTM sets out criteria for “meet,” “partially meet,” and “does not meet” conformance with the Standard, but does not provide or define the term “substantial conformance.” This paper presents Gold Fields’ interpretation of the concept of substantial conformance and describes its journey towards achieving it at their Cerro Corona Mine in Peru and Tarkwa Mine in Ghana. This paper explains why Gold Fields rated many of their assets as “partially conformant” against several GISTM requirements in their Annual Tailings Disclosure Report. It also outlines the measures taken by the company to achieve substantial conformance. The benefits of achieving substantial conformance are highlighted, such as enhanced environmental and social performance and greater stakeholder trust. Finally, the paper concludes by highlighting the importance of transparent, ongoing collaboration and knowledge sharing between industry, academia, and regulators to improve tailings management practices globally.

Introduction Background on the Global Industry Standard on Tailings Management (GISTM) The safe and responsible management of tailings is of paramount importance in the mining industry to

189

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA prevent catastrophic failures and mitigate environmental and social impacts. In response to recent incidents and the growing need for a comprehensive standard, the Global Industry Standard on Tailings Management (GISTM) was introduced in 2020. This paper aims to provide an in-depth analysis of Gold Fields’ substantial implementation of the GISTM at their Cerro Corona Mine in Peru and Tarkwa Mine in Ghana. The GISTM serves as a governance and management standard that sets the guidelines for tailings management practices. It is crucial to note that the GISTM does not introduce any novel technical requirements, but instead focuses on providing a comprehensive framework for effective management. However, it brings significant changes in terms of key appointments, roles within the company, reporting, and monitoring requirements. To fully conform with the GISTM, mining companies must adhere to over 220 deliverables, which are outlined across 15 Principles. These principles cover a wide range of aspects related to tailings management. Notably, the requirements of the GISTM amount to a total of 77, each consisting of multiple requirements that must be met (a total of 219 requirement “parts”). It is important to emphasize that the GISTM is not solely centred around engineering disciplines. Approximately 40% of the requirements pertain to engineering, while 40% focus on governance and management activities, and the remaining 30% pertain to environmental and social disciplines. This paper will shine a light on Gold Fields’ interpretation of the GISTM’s concept of substantial conformance, which is not explicitly defined within the standard. Gold Fields has undertaken a comprehensive journey toward achieving substantial conformance, as evidenced by the recently released Annual Tailings Disclosure Reports for the Tarkwa and Cerro Corona mine, where many assets were rated as “partially conformant.” By examining the measures taken by Gold Fields to achieve substantial conformance, this paper aims to provide insights into the benefits of enhanced environmental and social performance and the cultivation of greater stakeholder trust. Furthermore, this research underscores the importance of transparent collaboration and knowledge sharing among industry stakeholders, academia, and regulators. Through such collaborations, the mining industry can continuously improve its tailings management practices globally, ensuring the safety and sustainability of mining operations.

Purpose of the paper The purpose of this paper is to provide insight into Gold Fields’ implementation of the Global Industry Standard on Tailings Management (GISTM) at the Cerro Corona Mine in Peru and Tarkwa Mine in Ghana. By focusing on Gold Fields’ experiences, we aim to shed light on the concept of substantial conformance and its practical application.

190

GISTM – DEFINING “SUBSTANTIAL” CONFORMANCE: GOLD FIELDS’ SUBSTANTIAL IMPLEMENTATION OF THE GISTM AT THE CERRO CORONA MINE IN PERU AND TARKWA MINE IN GHANA This paper seeks to explain why Gold Fields rated some of their assets as “partially conformant” against various GISTM requirements in their Annual Tailings Disclosure Report. We will delve into the challenges and limitations faced by Gold Fields during their journey toward achieving substantial conformance, providing valuable insights into the complexities of tailings management implementation.

GISTM implementation Gold Fields’ journey As a member of the International Council on Mining and Metals (ICMM), Gold Fields made a commitment to implementing the Global Industry Standard on Tailings Management across all our tailings facilities. Over the past three years, our Global Tailings Management team has dedicated significant efforts to confirm that our four tailings facilities with the highest consequence classification achieve conformance by August 2023. This Endeavor has been accompanied by an ambitious and challenging timeline, reflecting the high bar set by the Standard. We take pride in witnessing the transformative improvements that have occurred in tailings management, governance, and monitoring throughout the industry. It is gratifying to see the collective progress and the positive impact we are making as responsible mining practitioners. Throughout this journey, we have highly valued the collaboration with stakeholders and peers, recognizing the immense value of openly sharing our experiences, expertise, and lessons learned. By engaging in this collaborative approach, we have fostered a culture of continuous improvement, collectively working towards enhancing tailings management practices worldwide. The exchange of knowledge and the synergy created through collaboration have been instrumental in driving meaningful change, ensuring the safety, and minimizing the environmental and social impacts associated with tailings management.

Achieving “substantial conformance” The commitment made by the International Council on Mining and Metals (ICMM) at the launch of the Global Industry Standard on Tailings Management (GISTM) in August 2020, to achieve full conformance within three years for tailings facilities with “extreme” or “very high” consequence classifications, was an ambitious yet necessary target. This commitment acted as a catalyst, driving immediate and sustained action by companies to work towards and maintain full conformance with the Standard. Gold Fields, among others, has made significant progress in this regard, which we have termed as “substantial conformance.” However, the reality of progressing implementation while addressing site-specific challenges within the given three-year timeframe has proven to be highly challenging. It is not feasible to progress all 77 requirements simultaneously, as certain requirements must wait until others have been progressed or

191

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA completed. For example, developing breach analyses relies on establishing credible failure modes, which, in turn, is necessary to determine consequence classifications. Various practical technical challenges exist across different jurisdictions. These challenges include limited availability of reliable testing facilities and long lead times for obtaining analytical results. Additionally, there is a relatively small pool of deep expertise in tailings management within the consulting community, which both members and other companies rely on for their expertise. For Gold Fields, the term “substantial conformance” carries significant importance. It enables a more balanced evaluation of their adherence to the GISTM requirements, distinguishing performance from the term “partially meets.” Some organizations may have adopted the term “Meets with a Plan” as an alternative. It is crucial to consider the broader context and the journey towards conformance when evaluating performance in tailings management. Assigning a simplistic percentage score to multifaceted aspects, such as human rights due diligence assessments, can oversimplify the evaluation process. Gold Fields recognizes the limitations of a linear approach and emphasizes the demonstration of substantial conformance.

Defining “substantial conformance” However, it is essential to clarify the concept of substantial conformance. Within Gold Fields’ interpretation, substantial conformance acknowledges the efforts made and progress achieved, even if certain elements are still in the process of being fully executed or updated. For example, a mining operation may have completed a site characterization report within the past 3 to 5 years but chooses to update it based on new findings from a recent study. While a site characterization report technically exists, the question arises as to whether the company should classify itself as “meets” due to its existence or as “partially meets” because it is not fully executed at the time of disclosure. Defining substantial conformance allows for accurate and transparent assessment of performance.

Disclosing substantial conformance The Gold Fields’ Annual Tailings Disclosure Report In line with its commitment to the disclosure requirements of the Global Industry Standard on Tailings Management (GISTM), Gold Fields has undertaken the task of preparing an Annual Tailings Disclosure Report for each of its global operations. These reports provide a comprehensive overview of the mining operations and their Tailings Storage Facilities (TSFs), with a particular focus on those ranked with an Extreme Consequence classification, such as the Cerro Corona and Tarkwa operations, whose reports were disclosed in August 2023.

192

GISTM – DEFINING “SUBSTANTIAL” CONFORMANCE: GOLD FIELDS’ SUBSTANTIAL IMPLEMENTATION OF THE GISTM AT THE CERRO CORONA MINE IN PERU AND TARKWA MINE IN GHANA The Annual Tailings Disclosure Report consists of three distinct sections. Part 1 offers an overview of the mining operation and its TSFs, providing essential context for the subsequent sections. Part 2 comprises a Plain language summary, which includes all the information required by Principle 15 of the GISTM. This summary aims to present the key details in a clear and accessible manner for a broader audience. The most extensive section, Part 3, is a detailed self-assessment report. This section contains a single A4 page dedicated to each of the 219 requirement parts of the GISTM. Each page presents the specific GISTM requirement, its corresponding criteria, a justification for Gold Fields’ self-assessment rating, and a self-assessment rating itself. This report style, spanning approximately 300 pages, offers a comprehensive account of Gold Fields’ approach, methodology, and the work undertaken to achieve substantial conformance. Initially, the team prepared a simple table using a traffic light system to indicate compliance with each principle. However, considering the “partially meets” status of certain requirements, the team recognized the need to showcase their commitment and the substantial effort involved in reaching conformance. Thus, the decision was made to adopt the detailed self-assessment report format, ensuring transparency and thoroughness in their assessment. The Annual Tailings Disclosure report was a collaborative effort led by the Global Tailings Management Team, who worked closely with the respective mining sites. To ensure the accuracy and credibility of the report, a third-party consultant was appointed to verify the content. The scope of the verification was to confirm the validity of Gold Fields’ self-assessment, rather than conducting an independent audit. Independent reviews and audits are already conducted by Gold Fields’ engineer of record partner, independent reviewers, independent technical review board, and governance and management systems reviewers. Prior to the official disclosure deadline, the self-assessment verifiers from the third-party consultant visited the sites and reviewed the content of the Annual Tailings Disclosure report. Their preliminary feedback provided valuable insights and bolstered confidence in Gold Fields’ self-assessment process. As a final step, the third-party verifiers issued a Statement of Assurance, confirming the accuracy and integrity of the report. This Statement of Assurance has been published on Gold Fields’ website, further reinforcing the transparency and accountability of the company’s tailings management practices.

Why assets were rated as “partially conformant” despite substantial conformance The “partially conformant” rating assigned to GISTM requirements in Gold Fields’ Annual Tailings Disclosure Report can be attributed to several factors. Particularly for Tailings Storage Facilities (TSFs) with Extreme Consequence Classifications, the presence of approximately 1,000 project-affected people

193

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA downstream necessitates meaningful engagement and a comprehensive approach. Gold Fields’ teams have previously engaged these communities on various topics such as earthquakes, fire, cyanide, and explosions. However, initiating conversations specifically about tailings storage facilities poses unique challenges. It requires a gradual and ongoing process of building trust and understanding, acknowledging that it is a journey rather than a one-time interaction. To maintain a high standard for themselves, Gold Fields recognized the need to integrate the findings of comprehensive human rights due diligence assessments into their site characterization reports and emergency preparedness and response plans. Although this integration is not a direct requirement of the GISTM, Gold Fields understands that true effectiveness and impactful change can only be achieved by ensuring that this crucial information becomes an integral part of their systems and processes. Simply fulfilling a checklist or storing the information on a shelf without integration and action contradicts Gold Fields’ perspective. By acknowledging the importance of integrating human rights considerations into their operations, Gold Fields aims to go beyond mere compliance. They strive to create tangible and meaningful improvements in their tailings management practices, actively seeking to address the concerns and needs of project-affected communities downstream. This commitment to holistic and inclusive engagement is a key factor in the rating of assets as “partially conformant,” reflecting Gold Fields’ dedication to continuously advancing their tailings management approach.

Gold Fields’ journey towards substantial conformance at the Cerro Corona and Tarkwa mines Cerro Corona The Cerro Corona mine has a single, extreme consequence classification TSF, approximately 156 m in elevation, constructed at an altitude of 3,400 m above sea level. In August 2023, the Gold Fields Cerro Corona tailings stewardship team conducted an assessment of the TSF, determining its conformance level to be 85% in accordance with the requirements outlined in the Global Industry Standard on Tailings Management (GISTM). It is worth noting that all material dam safety requirements have been thoroughly addressed and resolved. However, it is important to delve into the significance of this assessment and what it truly signifies in terms of the facility’s compliance and performance. To clarify, it indicates that all critical dam safety requirements have been addressed, confirming the safety and stability of the TSF. Furthermore, it signifies that 87.5% of the 219 individual “parts” of each requirement of the GISTM have been successfully implemented and are in conformance.

194

GISTM – DEFINING “SUBSTANTIAL” CONFORMANCE: GOLD FIELDS’ SUBSTANTIAL IMPLEMENTATION OF THE GISTM AT THE CERRO CORONA MINE IN PERU AND TARKWA MINE IN GHANA Achieving this level of conformance has been a monumental effort, with the Cerro Corona team completing an impressive total of 161 of 219 criteria projects and deliverables within a timeframe of approximately three years.

Tarkwa The Tarkwa mine has four operating TSFs, three of which have been assigned an Extreme to Very High Consequence Classification. In August 2023, the Gold Fields Tarkwa tailings stewardship team conducted an assessment of the TSF 1, 2 and 3 at the Tarkwa mine, determining its conformance level to be 79% in accordance with the requirements outlined in the Global Industry Standard on Tailings Management (GISTM). It is worth noting that all material dam safety requirements have been thoroughly addressed and resolved. To clarify, it indicates that all critical dam safety requirements have been addressed, confirming the safety and stability of the TSF. Furthermore, it signifies that 79 % of the 219 individual “parts” of each requirement of the GISTM have been successfully implemented and are in conformance. Achieving this level of conformance has been a monumental effort, with the Cerro Corona team completing an impressive total of 238 of 301 criteria projects deliverables within a timeframe of approximately three years.

Measures taken by Gold Fields The measures taken by Gold Fields encompassed the establishment of robust governance and management mechanisms. This included the creation of a comprehensive Tailings Management Policy, the appointment of key roles responsible for tailings management, and the development of a Tailings Management Framework, Management Standard, Incident Guideline, and RACI (Responsible, Accountable, Consulted, and Informed) matrices. At the operational level, the Cerro Corona and Tarkwa Tailings Stewardship teams were formed from the outset, working collaboratively with the Global Tailings Management team to address all aspects of the GISTM requirements. To promote technical expertise and adherence to best practices, an Engineer of Record firm was appointed for each operation, to lead technical studies and TSF design components. Recognizing the importance of independent review and validation, Gold Fields engaged third-party consultants at both operations to conduct comprehensive reviews and audits of various aspects of the TSF management process. Throughout the journey towards substantial conformance, the Cerro Corona Tailings Stewardship team actively engaged internal stakeholders, including over 450 contractors who are members of the downstream community. Robust, multi-level training was conducted to enhance awareness and understanding of tailings management practices and ensure alignment with the established standards.

195

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Additionally, the team began the process of consulting with local non-governmental organizations (NGOs) and the Defensa Civil to disseminate technical information and foster dialogue with the communities surrounding the mine. This proactive engagement aimed to promote transparency, foster understanding, and address any concerns or questions raised by the community stakeholders. The Tarkwa Tailings Stewardship team simultaneously commenced their community engagement program, nominating local volunteers to receive dedicated tailings training. The team has worked closely with community Chiefs, internal and external stakeholders to communicate messaging about tailings management. Gold Fields’ commitment to achieving substantial conformance at the Cerro Corona and Tarkwa Mine has involved comprehensive measures spanning governance, technical expertise, independent validation, stakeholder engagement, and community collaboration. Through these efforts, Gold Fields is dedicated to implementing best practices in tailings management and ensuring the safety, environmental responsibility, and positive social impact of its operations.

Challenges faced During the journey towards achieving substantial conformance, Gold Fields encountered several significant challenges. These challenges encompassed various aspects, including language barriers, extensive learning requirements, cultural transformation, tight timeframes, global coordination, community engagement, and interdisciplinary skill development. One of the challenges faced was the need to work across two languages, English and Spanish, at the Cerro Corona mine. Operating in bilingual environments necessitated effective communication and translation to ensure clear understanding and collaboration among team members. Learning the requirements of the Global Industry Standard on Tailings Management (GISTM) inside out proved to be a demanding task. The comprehensive understanding of the standard was crucial to implementing the necessary measures for conformance. In addition to completing individual projects, Gold Fields had to create a new culture and way of working within the organization. This involved not only meeting specific project goals but also fostering a broader shift in mindset and work practices to align with the principles and requirements of the GISTM. The timeframe for establishing the new culture, governance and management framework, and closing out all 219 requirement parts was particularly challenging. The process required efficient planning, coordination, and implementation of multiple initiatives within a limited timeframe. Conducting meetings across multiple time zones added another layer of complexity to the process. Coordinating schedules and ensuring effective participation from team members in different regions demanded careful organization and adaptability.

196

GISTM – DEFINING “SUBSTANTIAL” CONFORMANCE: GOLD FIELDS’ SUBSTANTIAL IMPLEMENTATION OF THE GISTM AT THE CERRO CORONA MINE IN PERU AND TARKWA MINE IN GHANA As the Cerro Corona and Tarkwa TSF’s have an Extreme Consequence Classification, educating more than 1,000 people on the risks associated with tailings management at both operations presented a considerable challenge. This task required the development and implementation of robust training programs to enhance awareness and understanding of tailings-related risks among employees and contractors. Establishing strategies to connect and engage with communities without causing unnecessary alarm was a crucial challenge. Balancing effective communication and community involvement while addressing concerns and building trust required careful planning and sensitivity. It was not feasible to progress all 77 requirements of the GISTM simultaneously, as certain requirements could not commence until others had been progressed or completed. For example, developing breach analyses relies on establishing credible failure modes, which, in turn, is necessary to determine consequence classifications. Various practical technical challenges exist across different jurisdictions. These challenges include limited availability of reliable testing facilities and long lead times for obtaining analytical results. Furthermore, the journey towards substantial conformance required engineers to develop interdisciplinary skills. For example, understanding and conducting human rights due diligence assessments, as well as recognizing the importance of collaboration with social practitioners and environmental professionals, expanded the skillset beyond traditional engineering disciplines. Overcoming these challenges demanded resilience, adaptability, and a collaborative approach. Gold Fields tackled these obstacles head-on, implementing strategies and initiatives to ensure the successful progression towards substantial conformance and the enhancement of their tailings management practices.

Benefits of achieving substantial conformance Substantial conformance to the Global Industry Standard on Tailings Management (GISTM) brings forth numerous benefits for Gold Fields and their tailings management practices. These benefits encompass the integration of interdisciplinary teams, enhanced design and engineering approaches, the integration of human rights, social and environmental considerations, fostering a culture of sharing and openness, assurance of facility integrity, technological advancements, global collaboration, and valuable learning experiences from independent technical reviewers. One significant benefit is the integration of interdisciplinary and international teams. Substantial conformance encourages collaboration and breaks down silos, ensuring that tailings design and engineering no longer operate in isolation. The involvement of diverse expertise from various disciplines allows for a more comprehensive and holistic approach to tailings management, incorporating considerations beyond traditional engineering aspects.

197

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA By achieving substantial conformance, Gold Fields successfully embeds human rights, social, and environmental considerations into the design and operation of their facilities. This integration ensures that ethical and sustainable practices are upheld, fostering a positive impact on local communities and the environment. The Annual General Meetings (AGMs) play a pivotal role in creating a culture of sharing and openness within Gold Fields. These gatherings provide a platform for knowledge exchange, facilitating easier collaboration and fostering a sense of collective responsibility towards tailings management. The AGMs serve as opportunities to learn from each other’s experiences, thereby promoting continuous improvement and best practices, conducting an AGM in two languages was an enjoyable and engaging experience. Substantial conformance offers reassurance that the facility is built on sound engineering principles and is designed to meet the highest standards of safety and environmental performance. This assurance not only instills confidence within Gold Fields but also among stakeholders and the wider community. Technological advancements in the field of TSF monitoring became an integral part of substantial conformance. The adoption of innovative monitoring technologies enhanced the efficiency and effectiveness of monitoring practices, enabling real-time data analysis and proactive risk management. This integration of technology further strengthens Gold Fields’ commitment to continuous improvement and informed decision-making. Achieving substantial conformance opens up opportunities for Gold Fields to work globally, collaborating with international teams and leading professionals in the field of tailings management. This global engagement facilitates the exchange of ideas, sharing of best practices, and exposure to diverse perspectives, enriching the overall knowledge and expertise of the organization. Participating in quarterly working groups provided our teams with a sense of belonging to a global initiative, enabling them to exchange ideas and insights across continents. These collaborative sessions fostered the establishment of international relationships with peers from diverse backgrounds, transcending borders and facilitating valuable cross-cultural learning experiences. Furthermore, Gold Fields gains valuable insights and learning experiences through engagement with independent technical review board members and independent reviewers. These experts provide objective assessments and guidance, ensuring rigorous scrutiny of tailings management practices. This collaboration promotes accountability, transparency, and continuous improvement within Gold Fields’ operations. Overall, the benefits of substantial conformance extend beyond compliance with the GISTM, enabling Gold Fields to advance their tailings management practices, enhance environmental and social performance, and contribute to the overall sustainability of their operations.

198

GISTM – DEFINING “SUBSTANTIAL” CONFORMANCE: GOLD FIELDS’ SUBSTANTIAL IMPLEMENTATION OF THE GISTM AT THE CERRO CORONA MINE IN PERU AND TARKWA MINE IN GHANA

Conclusion In conclusion, the concept of substantial conformance within the Global Industry Standard on Tailings Management (GISTM) holds significant benefits for companies like Gold Fields and the broader mining industry. Throughout this paper, we have highlighted key points related to substantial conformance and its implications for tailings management practices. The benefits of substantial conformance include the integration of interdisciplinary teams; the embedding of human rights and environmental considerations into design; the fostering of a culture of sharing and openness through AGMs; the assurance of facility integrity, technological advancements, and global collaboration; and the valuable learning experiences gained from independent technical reviewers. However, the journey towards substantial conformance is not without its challenges. Gold Fields encountered obstacles such as language barriers, extensive learning requirements, cultural transformation, tight timeframes, global coordination, community engagement, and interdisciplinary skill development. However, these challenges were met with determination, resilience, and a collaborative approach. The significance of substantial conformance lies in its ability to drive positive change and elevate tailings management practices to new heights. It promotes a comprehensive and holistic approach that goes beyond mere compliance. By integrating diverse disciplines, considering social and environmental factors, and fostering a culture of transparency and learning, substantial conformance contributes to improved safety, sustainability, and stakeholder trust. As we look to the future, it is imperative to maintain the momentum and continue striving for excellence in global tailings management practices. We call upon industry stakeholders, academia, and regulators to collaborate, share knowledge, and actively work towards refining and implementing best practices. By engaging in ongoing dialogue, learning from each other’s experiences, and embracing innovative approaches, we can collectively enhance the safety, environmental responsibility, and social performance of tailings management.

References International Council on Mining and Metals (ICMM). (2020a). Global Industry Standard on Tailings Management. International Council on Mining and Metals. International Council on Mining and Metals (ICMM). (2020b). Global Industry Standard on Tailings Management Good Practice Guide. International Council on Mining and Metals. International Council on Mining and Metals (ICMM). (2020c). Global Industry Standard on Tailings Management Conformance Protocols. International Council on Mining and Metals.

199

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

200

Engineers of Record, Professional Registration, and the Mining Industry in Mexico Jesus E. Romero, WSP USA, USA

Abstract As we prepare to comply with the Global Industry Standard on Tailings Management (GISTM), there are significant challenges related to Engineers of Record, professional registration, and the mining industry in Mexico. The shortage of qualified Engineers of Record will be a pressing concern for the next 20 years. The scarcity of Engineers of Record has led mining companies to rely on a limited number of consulting firms that offer these specialized services. However, even these companies face a shortage of senior technical leaders and require more professionals, who need extensive mentoring, training, education, and credentials. In Mexico, current regulations do not mandate an Engineer of Record as the responsible party for any tailings storage facility, making it complex and politically sensitive to modify these regulations. Waiting for regulatory changes could take decades, which is why it is crucial to invest in the development of professionals who can meet the requirements of the GISTM. Addressing the shortage of qualified engineers requires immediate action to provide mentoring, training, and education opportunities. Engaging the tailings and mine waste community is essential for advancing provisions in the registration processes, allowing Mexican engineers to meet the professional criteria outlined in the GISTM through similar evaluation and testing procedures as licensed engineers in Canada or the US. Collaboration with academia is necessary to explore new education delivery methods for training, mentoring, and educational opportunities for future Engineers of Record in Mexico. The COVID-19 pandemic has demonstrated the viability of remote learning. By garnering support from the tailings and mine waste community and pursuing collaboration with academia, innovative learning methods can be developed to create new training opportunities for the next generation of Engineers of Record in Mexico.

Introduction The Global Industry Standard on Tailings Management (GISTM) is an initiative aimed at enhancing safety, environmental stewardship, and social responsibility in the mining industry. It was developed following the devastating tailings dam failure at the Samarco mine in Brazil in 2015 and subsequent incidents that highlighted the need for improved tailings management practices.

201

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The GISTM is a collaboration between the International Council on Mining and Metals (ICMM), the United Nations Environment Programme (UNEP), and the Principles for Responsible Investment (PRI). It provides a comprehensive set of guidelines and requirements for the safe management of tailings facilities throughout their life cycle, from design and construction to operation, closure, and post-closure. Some key aspects of the GISTM in the mining industry include the following. The GISTM places a strong emphasis on preventing catastrophic failures of tailings storage facilities (TSFs). It promotes the adoption of risk-based approaches to ensure the safety of workers, communities, and the environment. The standard encourages the use of best practices to minimize the environmental impact of tailings management. It includes measures to reduce water consumption, prevent pollution, and promote the rehabilitation and reclamation of TSFs to restore the affected ecosystems. The GISTM emphasizes the importance of transparency, accountability, and stakeholder engagement. It encourages mining companies to disclose relevant information about their tailings facilities and engage with affected communities and other stakeholders. The GISTM aims to become a widely recognized and applied standard in the mining industry. Its adoption by mining companies and regulatory bodies helps to create a consistent framework for tailings management practices globally, raising the overall standard of safety and environmental performance. By promoting responsible tailings management, the GISTM aims to prevent catastrophic incidents, protect the environment, and improve the overall sustainability of the mining industry. It provides a framework for continuous improvement and serves as a guide for companies seeking to ensure the safe and responsible management of tailings throughout their operations. In summary, effective tailings management is essential for minimizing environmental impacts, protecting water resources, mitigating risks, engaging with communities, complying with regulations, and promoting the long-term sustainability of the mining industry. By adopting responsible practices, mining companies can minimize their environmental footprint and contribute to the well-being of local communities and ecosystems.

Importance of Engineers of Record in the mining industry Engineers of Record (EoR) play a critical role in ensuring safety and compliance in engineering projects. An Engineer of Record is a licensed professional engineer or a team of engineers who take responsibility for the design, analysis, and documentation of engineering projects. They are accountable for the technical accuracy, integrity, and compliance of the project design with applicable codes, standards, and regulations. Engineers of Record conduct thorough analysis and calculations to ensure that the design meets safety requirements, functional specifications, and performance objectives. Engineers of Record have a crucial role in ensuring compliance with applicable codes, regulations, and industry standards. They stay updated with local, national, and international codes and regulations related to their specific field of expertise.

202

ENGINEERS OF RECORD, PROFESSIONAL REGISTRATION, AND THE MINING INDUSTRY IN MEXICO Potential risks associated with the project are evaluated by these professionals and they develop strategies to mitigate them. They conduct risk assessments to identify potential hazards and implement measures to minimize or eliminate risks. This includes designing safety features, implementing appropriate materials and construction techniques, and considering factors such as seismic activity, fire safety, and environmental impact. Engineers of Record maintain comprehensive documentation of the project design, including drawings, calculations, specifications, and reports. These records serve as a reference for future assessments, inspections, and audits. Proper documentation helps ensure accountability, transparency, and traceability of the engineering decisions made throughout the project lifecycle. Engineers of Record often provide construction oversight to ensure that the project is implemented according to the approved design. They review construction plans, inspect the worksite, and monitor the construction process to ensure compliance with the design intent. They collaborate with contractors, subcontractors, and other stakeholders to address any issues that arise during construction and ensure that quality standards are maintained. As the responsible professionals, Engineers of Record carry a certain level of professional liability. They are accountable for the accuracy of their designs and the safety of the project. In case of any failures or incidents, they may be held legally and ethically responsible for any deviations from accepted engineering practices or failure to meet applicable codes and regulations. The role of Engineers of Record is crucial in ensuring that engineering projects are designed, executed, and completed in a safe, compliant, and responsible manner. Their expertise, attention to detail, and adherence to professional ethics help safeguard public safety, protect the environment, and uphold the integrity of the engineering profession.

Current challenges and uncertainties in Mexico regarding Engineers of Record and professional registration This is an overview of some challenges and uncertainties related to Engineers of Record (EoR) and professional registration in Mexico. • The process of professional registration for engineers in Mexico is linked to receipt of a bachelor’s degree, under the current regulation the Department of Education (Secretaria de Educación Publica) issues a professional certificate to practice the profession for which a bachelor's degree was obtained. There is no requirement to start the formal training of a professional engineer through a qualifying examination. Also, there is no requirement for experience in the field of practice and under the supervision of a licensed professional. • Current regulation in Mexico does not require that an Engineer of Record oversees the design, construction, and operation of tailings facilities. In general, ensuring compliance with professional

203

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA registration requirements and the proper appointment of Engineers of Record can be challenging. The current situation presents a lack of enforcement mechanisms or monitoring systems to ensure that projects have qualified and registered professionals overseeing them. • Encouraging ongoing professional development and continuing education among engineers is essential for maintaining high standards of competence and staying updated with the latest industry practices. There is limited availability and accessibility of training programs, continuing education opportunities, and professional development resources for engineers in Mexico that are interested in becoming an Engineer of Record. • With the rapid advancement of technology and the emergence of new fields within engineering, such as artificial intelligence, renewable energy, and sustainable infrastructure, there may be uncertainties regarding the inclusion of these areas in professional registration frameworks. Updating and adapting the registration process to account for these evolving fields can be a challenge. Developing consensus among the relevant professional engineering organizations, regulatory bodies, and legal authorities in Mexico could provide the legal and administrative framework to identify a path moving forward in selecting an existing organization to carry on the responsibilities of a regulating body that administers the qualifying exams and the governance of the licensed professionals just as it is expected in the United States of America and in Canada.

Current state of professional registration in Mexico Overview of existing regulations and their limitations In Mexico, all professions are regulated at the federal level, there is no technical boards of registration at the state level that are regulating the governance of specific engineering licensed professionals. There is no specific requirement for an Engineer of Record for TSFs in Mexico. While all professions are regulated, the lack of explicit regulations mandating an Engineer of Record for TSFs may contribute to uncertainties regarding accountability, oversight, and adherence to engineering best practices in tailings management. In Mexico, general regulations and codes related to engineering, construction, and environmental management apply to TSFs. Some of the relevant regulations and guidelines include: • NOM-141-SEMARNAT-2003: This establishes the procedure to characterize the tailings, as well as the specifications and criteria for the characterization and preparation of the site, project, construction, operation, and post-operation of tailings dams. • Environmental Regulations: Environmental regulations in Mexico, such as the General Law on Ecological Balance and Environmental Protection (Ley General del Equilibrio Ecológico y la

204

ENGINEERS OF RECORD, PROFESSIONAL REGISTRATION, AND THE MINING INDUSTRY IN MEXICO Protección al Ambiente), set standards for environmental impact assessments, waste management, and protection of natural resources. These regulations indirectly affect TSF management and may require environmental permits and compliance with environmental standards. The lack of specific regulations requiring an Engineer of Record for TSFs in Mexico may present limitations and challenges, including: • Without explicit requirements for an Engineer of Record, there may be uncertainties regarding oversight, accountability, and the application of engineering best practices in the design, construction, and operation of TSFs. This can raise concerns about the safety, stability, and environmental performance of these facilities. • The absence of consistent regulations across states in Mexico may lead to variations in the requirements, standards, and oversight practices related to TSFs. Inconsistencies can create challenges in ensuring consistent levels of engineering expertise and compliance with best practices. • While there may be industry standards and guidelines available, the lack of explicit regulations for an Engineer of Record can create challenges in achieving widespread adoption and enforcement of these standards. This can impact the overall quality, safety, and environmental performance of TSFs.

Political and regulatory complexities in modifying existing regulations to require Engineers of Record for TSFs Examination of the challenges and barriers in amending regulations Amending existing regulations can involve navigating various stakeholder interests. Different industry sectors, professional organizations, government agencies, and advocacy groups may have diverse perspectives and priorities. Balancing these interests and reaching consensus on regulatory changes can be challenging. Additionally, regulatory changes often require political support and engagement from government officials and legislators. Political dynamics, including competing agendas, party affiliations, and electoral considerations, can impact the willingness and ability to initiate and support regulatory amendments. Amending regulations may require allocating resources for research, analysis, and stakeholder consultations. Government agencies responsible for regulatory oversight may face limitations in terms of staffing, expertise, and financial resources. Resource constraints can impede the speed and effectiveness of the regulatory amendment process. Quite often modifying existing regulations typically involves administrative processes, such as drafting new legislation, conducting impact assessments, soliciting public

205

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA input, and ensuring legal and procedural compliance. These processes can be time-consuming and require coordination among multiple government agencies and departments. Also, regulatory amendments must adhere to legal frameworks, constitutional provisions, and established procedures. Ensuring compliance with existing laws, regulations, and judicial precedents is essential. Legal challenges or interpretations of statutes can complicate the process of amending regulations.

Potential implications of political controversies surrounding regulatory changes Political controversies surrounding regulatory changes can lead to delays or the stalling of proposed amendments. Political debates and disagreements can impede the progress of regulatory reforms, leaving existing regulations in place despite recognized shortcomings. Political controversies can introduce uncertainty and inconsistency in the regulatory environment. Proposed changes may be met with resistance or face legal challenges, creating a lack of clarity about the future direction of regulations. Uncertainty can impact businesses, investors, and other stakeholders, leading to cautious decision-making and potential economic consequences. Addressing political controversies and navigating the complexities of regulatory amendments requires transparent and inclusive processes, stakeholder engagement, evidence-based decision-making, and a focus on the public interest. It is essential to balance competing perspectives and ensure that regulatory changes are well-founded, transparent, and effectively communicated to maintain trust and credibility in the regulatory system.

Collaboration with Canadian professional organizations The role of collaboration in addressing the shortage of Engineers of Record Collaborating with Canadian professional organizations for the management and regulation of the engineering practice could allow for the development of the principles and foundation of an organization in Mexico that could issue a professional license equivalent to those from the following organizations: • Engineers and Geoscientists British Columbia • Association of Professional Engineers and Geoscientists of Alberta (APEGA) • Engineers Geoscientists Manitoba • Engineers and Geoscientists New Brunswick • Association of Professional Engineers and Geoscientists of Saskatchewan (APEGS) • Engineers Nova Scotia • Engineers PEI • Engineers Yukon

206

ENGINEERS OF RECORD, PROFESSIONAL REGISTRATION, AND THE MINING INDUSTRY IN MEXICO • Northwest Territories and Nunavut Association of Professional Engineers and Geoscientists (NAPEG) • Ordre des ingénieurs du Québec (OIQ) • Professional Engineers and Geoscientists of Newfoundland and Labrador (PEGNL) • Professional Engineers Ontario (PEO) The role of collaboration in addressing the shortage of Engineers of Record can be discussed in the following context: • It can facilitate the transfer of knowledge and best practices related to EoR requirements, standards, and oversight processes. This knowledge transfer can help enhance the understanding and implementation of effective EoR practices in regions facing shortages. • International partnerships can support capacity building efforts by providing training, workshops, and professional development opportunities for engineers and regulatory professionals. Leveraging the expertise of Canadian professional organizations can contribute to strengthening the skills and competencies of professionals involved in overseeing TSFs and dam safety. • It can also offer technical support and guidance in areas such as geotechnical engineering, risk assessment, and tailings management. This can assist in addressing technical challenges, improving engineering practices, and ensuring compliance with international standards. Canadian professional organizations have extensive experience and expertise in engineering, particularly in the mining sector. Leveraging their knowledge and resources can have several benefits: • Canadian professional organizations have developed well-established standards, guidelines, and codes of practice for engineering disciplines, including geotechnical engineering and tailings management. Collaborating with these organizations can help in adopting and adapting these standards to local contexts, thereby enhancing the quality and safety of engineering projects. • Canadian professional organizations can create networking and mentoring opportunities for engineers and regulatory professionals in regions facing shortages. These connections can foster professional development, knowledge sharing, and collaboration among peers, further strengthening the expertise and capabilities of the workforce. • They can facilitate industry partnerships and engagement. These partnerships can support knowledge transfer, research collaborations, and joint initiatives aimed at addressing the shortage of EoRs. Industry participation can help identify emerging trends, innovative solutions, and technological advancements that can be applied to improve engineering practices.

207

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA By fostering collaboration and leveraging the expertise and experience of Canadian professional organizations, regions facing a shortage of EoRs can benefit from international knowledge exchange, capacity building, technical support, and the adoption of best practices. Such collaboration can contribute to addressing the shortage and enhancing the overall quality, safety, and sustainability of engineering projects in the mining industry.

Potential benefits of collaboration with Canadian professional organizations Collaborating with Canadian professional organizations can provide access to training programs, mentoring initiatives, and educational resources. These organizations often offer professional development courses, workshops, and seminars on various engineering disciplines, including geotechnical engineering and tailings management. Access to such resources can help address skill gaps, improve technical competencies, and enhance the qualifications of professionals involved in overseeing TSFs. Canadian professional organizations have established training programs specifically tailored to the needs of Engineers of Record and professionals involved in dam safety. These programs can provide comprehensive training on topics such as risk assessment, design principles, regulatory compliance, and emergency response planning. These efforts can facilitate mentorship programs, pairing experienced professionals with those seeking guidance and support. Mentoring relationships can foster knowledge transfer, career development, and the cultivation of best practices. Canadian professional organizations often publish technical guidelines, manuals, and research papers related to geotechnical engineering, tailings management, and dam safety. Access to these resources can contribute to the professional development and continuous learning of engineers and regulatory professionals. Creating opportunities for cross-border cooperation and mutual learning can lead to collaborations that usually yield benefits, including: • Enabling the exchange of knowledge, experiences, and lessons learned between professionals in different regions. Canadian professional organizations can share their expertise and insights on effective engineering practices, regulatory frameworks, and industry trends. • This collaboration allows for the adoption of best practices and is refined within the Canadian engineering community. These best practices may include innovative approaches to geotechnical engineering, risk management, and environmental stewardship, which can be applied in tailings management and dam safety initiatives. • Efforts of collaboration with Canadian professional organizations can foster research collaborations and joint projects. These collaborations can contribute to the development of new technologies, engineering solutions, and scientific advancements in tailings management. Shared research

208

ENGINEERS OF RECORD, PROFESSIONAL REGISTRATION, AND THE MINING INDUSTRY IN MEXICO endeavors can lead to improved practices, increased safety, and enhanced environmental sustainability. • Collaborative partnerships provide opportunities for professionals to engage in peer networking, attend conferences, and participate in technical workshops. These events allow for the exchange of ideas, discussions on emerging challenges, and the establishment of professional connections. Participation in Canadian professional organization conferences and events can broaden perspectives and facilitate the transfer of knowledge across borders. By collaborating with Canadian professional organizations, regions facing challenges related to Engineers of Record and dam safety can benefit from access to training, mentoring, and educational resources. They can also tap into opportunities for cross-border collaboration, knowledge sharing, and mutual learning, ultimately leading to improved engineering practices, enhanced regulatory frameworks, and increased professional capacity in tailings management and dam safety.

Engaging the tailings and mine waste community Advancing provisions in the registration process for engineers in Mexico To advance provisions in the registration process for engineers in Mexico, it is important to ensure alignment with international standards and best practices. This can be achieved by reviewing and incorporating relevant guidelines, codes, and regulations from reputable international bodies such as the International Professional Engineers Agreement (IPEA), International Federation of Consulting Engineers (FIDIC), or relevant standards set by Canadian or American professional engineering organizations. Aligning with these standards can help establish a robust and internationally recognized registration process for engineers in Mexico. • The registration process should include a thorough evaluation of professional competence, ensuring that engineers possess the necessary qualifications, skills, and knowledge to carry out their responsibilities effectively. This may involve assessing academic qualifications, work experience, technical competencies, and adherence to professional ethics and standards. • Emphasizing the importance of continuing professional development (CPD) is crucial to ensure that registered engineers stay updated with the latest advancements, technologies, and best practices in their field. Implementing CPD requirements as part of the registration process can help maintain high professional standards and ensure ongoing professional growth.

209

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Promoting equal opportunities and evaluation procedures as in Canada or the US To promote equal opportunities and fair evaluation procedures in the professional registration process, it is important to establish transparent and objective evaluation criteria and processes. This ensures that all applicants are assessed based on their qualifications and abilities, regardless of their background or affiliations. • The evaluation criteria should be clearly defined, measurable, and directly related to the required competencies and qualifications for professional registration. This can include academic qualifications, professional experience, technical skills, ethical standards, and adherence to relevant codes and regulations. • The registration process should be transparent, with clear guidelines and procedures that are accessible to all applicants. This includes providing information on the required documentation, evaluation methods, timelines, and appeals processes. Transparent processes help build trust, eliminate biases, and ensure fairness in evaluating applicants. • Establishing independent evaluation bodies or regulatory authorities can help ensure impartiality in the registration process. These bodies should be composed of qualified professionals with expertise in the relevant engineering disciplines and should operate independently from any external influence or pressure. •

Implementing an appeals mechanism allows applicants to challenge evaluation decisions that they believe are unfair or unjust. This ensures a fair and transparent process and provides a means for applicants to seek recourse if they believe their evaluation was not conducted in accordance with the established criteria and procedures. Promoting equal opportunities and fair evaluation procedures in the registration process for engineers

in Mexico helps create a level playing field, encourages diversity, and ensures that qualified professionals can participate and contribute to the industry.

Collaboration with academia for alternative education delivery There are several ways of exploring innovative learning methods for training Engineers of Record, here are a few worth pointing out.

Assessing the potential of technology-driven learning platforms • Online courses and webinars can be developed in collaboration with academia to provide flexible learning opportunities for EoRs. They can be designed to accommodate self-paced learning, allowing professionals to access the content at their convenience.

210

ENGINEERS OF RECORD, PROFESSIONAL REGISTRATION, AND THE MINING INDUSTRY IN MEXICO • Virtual simulations and laboratories can be utilized to create realistic scenarios for EoRs to practice decision-making, problem-solving, and risk assessment. • Technology-driven learning platforms can incorporate interactive tools, multimedia resources, and visualizations to enhance understanding and engagement.

Incorporating practical experience and case studies • Collaboration with mining companies and industry professionals can facilitate the incorporation of practical experience into the training of EoRs. This can involve industry-led workshops, site visits, and collaborative projects that expose EoRs to real-world challenges and best practices. • Academia can develop case studies and problem-based learning approaches that simulate real-life situations encountered by EoRs. These case studies can involve analyzing and solving complex engineering problems related to tailings management, dam safety, and regulatory compliance. • Establishing internship and co-op programs in collaboration with mining companies allows EoRs to gain practical experience in the field. These programs provide opportunities to apply theoretical knowledge in real-world settings and learn from experienced professionals. Collaboration with academia for alternative education delivery provides an opportunity to explore innovative learning methods for training Engineers of Record. Technology-driven learning platforms can enhance accessibility and engagement, while incorporating practical experience and case studies can bridge the gap between theory and practice. By leveraging these collaborative approaches, the training of EoRs can be enriched, ensuring that professionals are equipped with the necessary knowledge, skills, and practical experience to effectively oversee tailings storage facilities and ensure dam safety.

Conclusion In conclusion, the shortage of qualified Engineers of Record (EoRs) in the mining industry poses significant challenges to the effective management of tailings storage facilities (TSFs) and to ensuring dam safety. However, through strategic collaborations and proactive measures, these challenges can be addressed to meet future demands and enhance environmental and social sustainability in the mining sector. Collaboration with Canadian professional organizations offers valuable opportunities to tackle the shortage of EoRs. Through international partnerships and knowledge exchange, regions facing shortages can benefit from training, mentoring, and educational resources provided by Canadian professional organizations. Leveraging the expertise and experience of these organizations will help to ensure alignment with international standards and best practices, enhancing the professional registration process in regions such as Mexico. The promotion of equal opportunities and fair evaluation criteria can create a level playing field and foster diversity in the profession.

211

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Furthermore, collaboration with academia plays a vital role in addressing the shortage of EoRs. Exploring innovative learning methods, such as technology-driven platforms and practical experience incorporation, allows for effective training and development of EoRs. Through online courses, virtual simulations, and interactive tools, professionals can acquire the necessary skills and knowledge to navigate complex challenges in tailings management and dam safety. Incorporating practical experience, case studies, and industry partnerships provides hands-on learning opportunities, bridging the gap between theory and practice. Overall, through collaborative efforts, the mining industry can address the shortage of qualified EoRs, enhance tailings management practices, and ensure the safety and compliance of TSFs. By leveraging the knowledge, expertise, and resources of Canadian professional organizations, promoting equal opportunities, and collaborating with academia, the industry can build a competent workforce, embrace innovation, and achieve environmental and social sustainability in the mining sector.

References Camara de Diputados de Mexico. 2023. Ley General del Equilibrio Ecológico y la Protección al Ambiente (General Law of Ecological Balance and Environmental Protection). Accessed 5/26/2023 at https://www.diputados.gob.mx/LeyesBiblio/ref/lgeepa.htm International Council on Mining and Metals, United Nations Environment Programme (UNEP) and Principles for Responsible Investment (PRI). 2020. The Global Industry Standard on Tailings Management. Retrieved from https://globaltailingsreview.org/global-industry-standard/ International Professional Engineers Agreement. 2023. International Engineering Alliance. Accessed 6/15/2023 at https://www.ieagreements.org/agreements/ipea/ Secretaria de Educacion Publica. 2023. Expedición de cédula profesional electrónica. Accessed 5/4/2023 at: https://www.gob.mx/tramites/ficha/expedicion-de-cedula-profesional-electronica/SEP6534 Secretaria de Medio Ambiente Recursos Naturales. 2003. Norma Oficial Mexicana NOM-141-SEMARNAT-2003. Accessed 5/15/2023 at https://www.profepa.gob.mx/innovaportal/file/1317/1/nom-141-semarnat-2003.pdf

212

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Effective EoR Succession Planning Recommendations for Implementation of GISTM Madeline R. Sova, WSP USA Inc., USA Hülya Salihoğlu Ertürk, WSP USA Inc., USA Christopher N. Hatton, WSP USA Inc., USA

Abstract The Global Industry Standard for Tailings Management (GISTM or the Standard) was a call-to-action for tailings storage facility (TSF) operations on a global scale. This compliance standard has paved the way for change in tailings management practices and established the need for sourcing competent individuals to fulfill the roles outlined, including the engineer of record (EoR). Transferring a facility’s design, construction, and operation knowledge from the current EoR to the prospective EoR can be complex, but it is essential for sustainability and consistency in safe TSF operation. Requirement 9.5 of the GISTM addresses change management regarding the EoR position. With a global shortage of suitable and qualified professionals, dedicated development planning for EoRs and implementation of long-term succession planning provide a practical pathway through EoR transitions. Effective succession planning can facilitate early identification of high potential successors, encourage periodic review of key roles and positions, and provide actionable career development frameworks. Structured communication of the succession plan to the relevant parties supports a transparent and organized transition. Key supporting positions that may be identified in an EoR succession plan include the Deputy EoR, for long-term planning and development of engineers through mentoring and project-specific experience, and the Alternative EoR concept, which may be beneficial for short-term or critical transition periods. A survey was administered to consultant engineers to gather data on EoR transitions. One survey finding indicated that approximately 30% of respondents do not want to pursue the role of EoR during their career. Succession planning may encourage developing engineers to fill the EoR role through thoughtful mentorship, written and oral communication practice, and real-world experience to gain the confidence needed to assume the position. Recommendations for developing an EoR succession plan are provided based on qualitative and quantitative findings from survey results and succession planning literature review.

213

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM

Introduction The mining industry, and more specifically the tailings and mine waste industry, has evolved significantly over the past three years with the launch of GISTM in August 2020 by the International Council on Mining and Metals (ICMM), the United Nations Environmental Programme (UNEP), and the Principles for Responsible Investment (PRI). The industry has long been working towards environmental stewardship and due diligence measures to protect the natural environment and promote safe-working operations. The GISTM standard is organized around six broad topic areas providing a framework and guide for mining companies to achieve safe tailings facility management. The standard applies to current and future TSFs. Timeframes are included within the Standard for ICMM members suggesting compliance by August 2023 for operating facilities classified as high- or extreme-consequence hazard and compliance by August 2025 for closed facilities and/or facilities classified as low- to moderate-consequence hazard. Mine Owners, regulators, and engineering consultants have been reporting on facilities; however, with the new Standard, they now report to a level not yet before required or defined. Because the typical lifecycle of a TSF can span 10 to 100 years, facilities often undergo progressive construction while maintaining active operations. The reporting under the Standard is an additional requirement to the typical construction documentation and operation activities, including required public disclosures and increased stakeholder engagement. Engaging in facility compliance with the Standard has been a large undertaking for the industry. It has required increased responsibility for engineers and facility Owners to produce synthesized and thoughtful assessments of the design intent for TSFs worldwide. Due to the history and longevity of many of these facilities, proper knowledge transfer and succession planning must be clearly defined and periodically reviewed for continuing operations at or above the standard of practice that GISTM has aimed to achieve.

EoR definition and responsibilities The EoR supports the TSF facility Owner by providing guidance and recommendation for operational practices, facility design, construction or design modifications, closure design and execution, and postclosure monitoring and maintenance. The definition provided by GISTM for the role of the EoR is “The qualified engineering firm responsible for confirming that the tailings facility is designed, constructed, and decommissioned with appropriate concern for the integrity of the facility and that it aligns with and meets applicable regulations, statutes, guidelines, codes, and standards. The EoR may delegate responsibility but not accountability. In some highly regulated jurisdictions, notably Japan, the role of EoR is undertaken by the responsible regulatory authorities” (GISTM, 2020).

214

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM Requirements, under the GISTM, establish an approach for safe operations toward the goal of zero harm to the environment. The Standard subdivides the six topic areas into 15 principles and 77 auditable requirements, including topics ranging from design and operations to emergency response and public disclosures. Topic IV: Management and Governance, Principle 9 of the Standard addresses the EoR and states, “… appoint and empower an Engineer of Record” (GISTM, 2020). Five requirements under this principle summarize, in broad terms, the requirements of the EoR. As stated under Requirement 9.1, the mining facility is to “engage an engineering firm with expertise and experience in the design and construction of tailings facilities of comparable complexity to provide EoR services for operating the tailings facility and closed facilities with ‘High,’ ‘Very High,’ and ‘Extreme’ consequence classification, that are in the active closure phase.” The mining operator may appoint an external firm or an in-house engineer with a similar level of expertise and experience as EoR. Supporting guidance documentation developed by ICMM, the Geoprofessional Business Association (GBA), the Canadian Dam Association (CDA), and the Mining Association of Canada (MAC) provide important underpinning detail to support interpretations and definition of the EoR and the responsibilities of the EoR team. Some of these responsibilities include assuring that the facility design meets the applicable regulations, codes, and guidelines; confirming that the facility is constructed and operated consistent with the design intent; and supporting operations throughout the facility lifespan. The ICMM Tailings Management Good Practice Guide, for example, states that it is up to the facility’s Owner to determine which EoR model aligns with their needs and the required qualifications and competencies of the EoR (ICMM, 2021). The multi-disciplinary nature of tailings facility management often requires the EoR to establish a supporting team composed of technical experts and competent staff members. This is the inherent nature of the EoR role; it is highly dependent upon the support of a quality team. EoRs will scale this multi-disciplinary team and assess the requirements based on the facility’s season of development, complexity, and operational needs. The guidance documents also define the roles and responsibilities of the facility Owner/operator, the accountable executive (AE), and the responsible tailings facility engineer (RTFE). These roles are likely to be held by the Owner’s representative(s) and are external in most cases (but not all) to the EoR team. This study does not address these external roles.

Roles for EoR succession planning The assignment of an EoR must be carefully considered by facility Owners and key stakeholders. The EoR should be technically competent to make design, construction, and operating decisions and requires essential soft skills such as emotional intelligence and maturity, practical communication skills, leadership qualities, problem-solving skills, and the ability to influence others to reinforce the successful management

215

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM of the facility. The EoR should “have experience and knowledge commensurate with the risk management requirements for the facility” (MAC, 2017). Sustainability is an essential element of the role. It would be unwise for a mine Owner to assume that the selected EoR will remain in place over the facility’s lifespan. Thus, it is prudent that “succession planning to support the transition from one EoR to another should be in-place” (GBA, 2019). The Deputy EoR (DEoR), as defined by the GBA, “would assist the EoR by serving as second-incommand (if the EoR is temporarily unavailable) and fulfill the duties of the EoR if the EoR is no longer able to perform them…Though not widely adopted, the designation of a Deputy EoR is considered an emerging approach to assist with succession planning” (GBA, 2019). Lack of comprehensive documentation for the role of DEoR may imply that it is a position that can vary on a project-by-project basis due to complexity, internal EoR team framework and demands, or the facility Owner’s operational and governance preferences. The authors would like to propose use of a temporary role designated as the Alternative EoR (AEoR). Born out of the necessity to maintain quality and consistency in the EoR engagement when resources are dramatically limited. Although this role is not cited in the literature, it is another interim successorship role designated only for special circumstances and is intended to be temporary (less than a year). An AEoR may be best described as a qualified engineer who may assume the role of EoR when an immediate position replacement is required. For example, when the existing EoR retires or resigns from their duties. In this example, the DEoR may not yet have the competencies to assume the position; the AEoR role provides practical and necessary temporary support to the project with the assistance of the DEoR and EoR team. The ultimate goal is to enhance the viability and success of the EoR.

Recognized labour shortage for GISTM implementation Research conducted and presented in the paper “Characterizing tailings professional labor demand” by Spencer et al. concludes that the tailings facilities worldwide (estimated at approximately 17,000 facilities) will require an estimated 12,900 to 18,900 full-time equivalent (FTE) professionals for safe and sustainable tailings management under the standard set by GISTM. The estimated FTEs for the role of EoR range from approximately 2,300 to 3,600 to satisfy the requirements under GISTM, see Figure 1 below. FTE positions are likely to be grossly underestimated due to variances in estimated facilities worldwide and the differing complexity of projects. The FTE estimates have also been generalized to quantify the specific labour resource. It is essential to realize that one FTE designation may require the additional support of multiple people ranging from project management professionals to administrative resources. Therefore, the succession planning of the EoR role for TSFs is crucial for keeping talented professionals in the pipeline but may also, more times than not, extend beyond the individual to a larger support group.

216

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM

Figure 1: Estimates of tailings labour demand for the minimum estimated 17,000 tailings facilities worldwide (Spencer et al., 2022)

Succession planning The recognized and projected shortage of competent engineers to fill the positions needed for the industry to satisfy the requirements of GISTM now makes succession planning an essential aspect of the EoR position. Morrison et al. (2017) presented results from an EoR workshop sponsored by the GBA in 2017. They found that “the process for transitioning from one EoR to another was identified as a concern by several participants. This concern applies whether a planned handover of responsibilities from the design phase to the construction and operational phases exist or when a change in the EoR due to contractual or other business reasons is contemplated.” Due to the nature and lifecycle of a TSF, change management for the role of EoR is a critical aspect of the business to support continuity and responsibility of safe tailings management. According to the National Institutes of Health (NIH), poor succession planning may lead to a loss of critical knowledge and time which can significantly influence productivity and cause disruptions to the workflow (NIH, 2021). Succession planning has proven successful by providing continuity for essential roles, improving strategy alignment and satisfaction, and efficiently identifying high-potential role candidates. When done proactively, succession planning facilitates early identification of high potential successors, encourages periodic, and sometimes a necessary, review of key roles and positions, and serves as a platform for developing actionable career development frameworks. Completing a transition-and-succession plan assists in identifying skillset gaps within the current team and supports the growth and education of the successor. A typical succession plan will include: •

a definition of the role and associated responsibilities, documentation of the short-term and longterm scenario plans;

217

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM •

a candidate evaluation for succession competency;

•

methods for communication of the plan to relevant parties; and,

•

additional involvement of management and leadership team with the review process (NIH, 2021). In addition, as emphasized in the ICMM Good Practice Guide, change in the role of EoR should also

include the documentation of TSF design intent and knowledge transfer, including documents, historical context, and supporting work completed (ICMM, 2021). Succession planning for developing engineers should be encouraged early in an individual’s career through thoughtful mentorship, practice in written and oral communications, and real-world experience to gain the confidence needed to assume a position. Communication of the succession plan to the relevant parties is mandatory, and practice supports a transparent and organized transition. Identified successors for key roles should be evaluated on a project-by-project basis. Succession planning should also include an evaluation of the transition timeframe component. Setting milestones and transition schedules during and after assuming the role will increase accountability in the transition process and the facilitators of the transition. A successful succession plan depends upon the mature engagement and selfless contributions of the EoR and the individual seeking to become an EoR. The position of EoR is distinguished and carries great responsibility and challenge. It requires an underlying drive, an element of purpose, and an understanding of the greater goal. It is not a reward for good work or years served, but an esteemed position recognized for hard work, personal sacrifice, and selfless achievement. The EoR is a leader that makes everyone around them better.

Survey and results The framework for an EoR, as outlined above, was evaluated against a community of potential and existing EoRs. The goal was to understand how industry professionals view the EoR role and what challenges the position faces within the industry. A survey was administered to approximately 125 consultants in the mine waste industry to gather feedback on the opinions of the role of EoR competencies and succession planning. The questionnaire comprised 25 questions grouped into four main categories: demographics, the role of EoR and DEoR, core competencies, and people and soft skills. The survey response rate was approximately 34%, yielding an 8% confidence interval. Survey participants ranged in experience level and included practitioners with experience as an EoR and/or DEoR and those without. It is recognized that the roles and responsibilities of the EoR and/or DEoR may need to be clarified to some professionals, especially those new to the industry. Therefore, the level of understanding of the role was also assessed. More notably, the role of DEoR is unclear and not well documented or defined in GISTM

218

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM or other supporting guidance documents and is often subject to the Owner or EoR teams’ interpretation. Participants were requested to rank their understanding of the specific roles (i.e., EoR and DEoR) on a scale of 1 (= no understanding) to 5 (= complete understanding). To assume the role of EoR, technical and non-technical competencies must be assessed to fill this key position with a thoughtful and thorough review. Respondents were asked to rank a variety of technical competencies needed for a successful EoR. These competencies include geotechnical background, risk management, contract terms, site experience, and operational knowledge. The survey also inquired how best to foster and develop the soft skills (i.e., non-technical skills) needed to perform the role. Those intangible skills such as emotional intelligence, emotional maturity, problem-solving, communications, organization, leadership, teamwork, or creativity revolve around personal relationships and are needed to facilitate better collaboration.

Quantitative results Respondents were categorized into five groups based on their demographic level of experience (Fig. 2). 28% of the participants served in an EoR position for 0 to 10 years, and 21% served as DEoR. 51% of respondents were not serving in an EoR or DEoR position, with 30% not interested in the role as part of their career development or path.

Not an EoR/Deputy EoR and not a prospective EoR/Deputy EoR.

17%

Not an EoR/Deputy EoR but prospective EoR/Deputy EoR.

30%

11% Currently serving in a Deputy EoR position. Currently serving in an EoR position (0-5 years' experience in the role).

21%

21%

Currently serving in an EoR position (5+ years' experience in role).

Figure 2: Questionnaire demographics The average understanding of the roles of the EoR and the DEoR, on a scale of 1 (= no understanding) to 5 (= complete understanding), was at 3.7 for the EoR role and 3.5 for the DEoR role, which is an indication that there is a need for clarification of EoR and DEoR role in industry documents. Based on the data collected, respondents with 0 to 5 years of EoR experience hold the position of EoR for two TSFs on average. For participants with 5 to 10 years of EoR experience, this number was three, on average. The average number of TSFs per DEoR was one.

219

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM More than half of the respondents (58%) indicate that a succession plan is in place, to some degree, on the TSFs that they were working on. Communication of a succession plan to the internal EoR team and the Owner occurs 74% of the time. It has been inferred from open-ended responses to survey questions and follow-up discussions with respondent EoRs that the transition plans have been developed on a project-byproject basis. It is recommended that a standard approach to transition planning through the development of a template would be an effective means of encouraging current EoRs to consider their short-term and long-term succession plans. A standard template is expected to facilitate plan visibility, communication, and collaboration. More than 70% of respondents stated that the EoR should have at least 10 years of experience in the mine-waste industry before assuming the role (Fig. 3). Only 2% think that 0 to 5 years of experience is sufficient to assume an EoR role.

0-5 years

6-9 years

10+ years

2%

26% 72%

Figure 3: Respondents’ opinion about the number of years of experience before assuming the role of EoR 89% of the questionnaire responses indicated that, depending on the complexity of the project, the TSF, and the experience of the EoR successor, the transition timeframe for assuming the role of EoR may last between six months to three years. Respondents also commented that retaining the outgoing EoR for “on-call” services for a period after the completion of the transition may be a beneficial addendum to the transition plan. Setting milestones and a transition schedule during and at the onset of assuming the EoR role will increase accountability by the EoR and the EoR team.

Qualitative results Qualitative feedback from the participants was grouped into four categories: core competency areas for EoR successors, soft-skill development, EoR mentoring, and EoR-Owner interactions. Additional feedback

220

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM and “lessons learned” from the existing, surveyed EoRs were also gathered from open-ended questions, contributing to the recommendations for improving the management of change and transition. Questionnaire results indicated that an incoming EoR should have experience with core competency areas such as facility site visits, Owner interactions, facility-design modeling and analyses, and fieldwork, including construction quality assurance (CQA), and exploratory, and data-monitoring activities. Respondents identified required experience that includes: • practice as an EoR or DEoR for other facilities; • knowledge of risk assessment and management; • participation in independent technical review board (ITRB) meetings; and, • communication with contractors. Competencies that pertain to a successor’s soft skills include open communication (oral and written) and active listening, Owner trust, application of input from multiple groups and collaboration, courage, adaptability, and consideration of challenges from multiple perspectives (e.g., the Owner, the Regulator, the Designer, the Contractor). Respondents’ suggestions for developing soft skills generally indicate that soft skills are learned from practice and experience, especially during interactions with the Owner and Owner’s representatives, and EoR mentors. Formal programs such as Toastmasters or professionalleadership courses may also enhance professional rapport and soft skills. Feedback suggests that the DEoR, or identified successor for the position of EoR, should build trust by communicating regularly with the Owner’s representatives via e-mail, phone call, or in person. 89% of respondents indicated that it is appropriate to communicate periodically, every month to every three months, varying in frequency based on the project complexity or current operational activities. In-person visits to the TSF may also support fostering these relationships.

Succession plan framework Recommendations for a succession plan framework have been collated and generalized based on the survey respondents’ feedback and from literature research. The physical process of documenting a succession plan has been shown to improve team culture and motivate individuals to pursue defined career path objectives. To its merit, succession planning also facilitates transitions between positions or roles with a more precise and concise direction. It is recommended that a succession plan template include details in the following categories: projectspecific details and brief historical context, definitions of the critical project roles and roles that are identified for successor candidates, scenarios for role transitioning, qualifications for the role, and

221

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM evaluation of the successor’s competency, periodic reviews of the succession plan and frequency of communications, and identification of additional management support members. A succession plan should include the role definitions (terms of reference) for the EoR, the DEoR, and potentially the AEoR. The role definition will outline the project's everyday responsibilities, duties, and commitments based on size and complexity. The interpretation and responsibilities of the DEoR may vary widely to accommodate a project and EoRs needs. It is important to detail this role and document the responsibilities so that project needs and requirements are met. In cases where short-term succession planning may be helpful, the designation of the AEoR and their responsibilities will allow for a direct reference for the knowledge transfer needed to facilitate a smooth transition. Each succession plan should include a standard set of competencies, agreed upon by the EoR team and the Owner, that must be met to fulfill the role of EoR. Additional competencies may be required if the project is substantially more complex or has unique aspects that could be more easily generalized. The evaluation of such competencies for the identified successors should be included in the succession plan. This evaluation will facilitate identifying knowledge gaps and areas a successor may thoughtfully target. Both technical and non-technical factors should be assessed in the competency evaluations. Succession plans should be completed with input from the EoR team and the Owner and be available and accessible to all relevant parties. The communication and review of the plan are just as important for the EoR team as it is for the Owner to be engaged to diminish the possibility of misperceptions of the roles and responsibilities of the EoR and the DEoR. It is recommended that succession plans be reviewed annually and updated as necessary. More complex projects may require succession plans to be updated more frequently.

Application scenarios Succession plans help address both short-term and long-term transitions and transitions of the EoR role internally and externally. Change can happen fast, and when teams need to prepare to transfer the skills and knowledge needed to fulfill the key roles and positions, especially that of an EoR, lack of pre-planning can be detrimental to the facility and teams supporting. Planning for short-term and long-term succession creates a roadway for identifying those who can fill positions quickly and those who may require further development or preparations to step into the role with competence.

Internal transitions Generally, it is relatively straightforward to manage internal transitions to a company (i.e., passing the role of EoR to a colleague or co-worker). Often, when a transition is needed, such as an EoR retiring, there is time for the position to be handed off and knowledge to be transferred. However, even an internal transition

222

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM can present challenges if inadequately planned. In cases where a more short-term replacement may be needed, such as an EoR leaving their place of employment, it may be beneficial to identify an AEoR. An AEoR may be an experienced engineer who can quickly fill the role of EoR for a short term or provide oversight and support to a DEoR during their development. In both cases, for long-term or short-term transition, it is beneficial to assign the DEoR as a support role for the EoR.

External transitions An EoR role may be transferred externally from one consulting firm to another or from a consulting firm to the Owner. Throughout the external transition process, both the previous EoR and the prospective EoR should be engaged to transfer project knowledge and history from the design phase through the construction and operation phases. An assessment of the potential risks of such a transition should be carefully considered and addressed promptly. The external EoR transition process can be more challenging than the internal transition because the Owner drives the process. Once the Owner selects the prospective EoR, the previous EoR has less control over the technical and non-technical evaluation of the successor EoR. At this point, the transition focuses more on handing over the responsibilities from one EoR to another. GISTM does not mention a specific timeframe for a healthy transition of the role; however, it is usually acknowledged by the industry that six months to a year of overlap period between the previous and successor EoR is appropriate.

Summary In an industry that requires a skilled workforce to uphold the safety of people and the environment, the role of EoR for TSFs comes with a weight of responsibility and commitment. With the necessary competencies needed to fill the role of EoR, the benefits of succession planning have never been more apparent. It is easy for companies to get busy with their day-to-day management of tasks; however, without adequately documenting and identifying potential successors for critical roles, the detriments to a company and our industry could be significant. Knowledge loss, gaps in competencies, and ignorance to properly evaluate the risk associated with the transition of the roles could mean catastrophe. The required skills and experience for the EoR role or prospective EoR role constrain the available talent pool today due to labour shortage in the mine waste industry. However, the ongoing development of young tailings engineers through educational programs and training combined with formal mentoring support effective succession planning and the transfer of knowledge from one generation to another.

223

EFFECTIVE EOR SUCCESSION PLANNING RECOMMENDATIONS FOR IMPLEMENTATION OF GISTM

Acknowledgments We so appreciate the colleagues and supporters we had throughout the preparation of this research and conference paper. Special thanks to Mrs. Hülya Salihoğlu Ertürk, P.E., for the time she spent reviewing and guiding the research, and to Mr. Christopher N. Hatton, P.E., who actively promotes mentoring and encourages the development of engineers within the industry.

References Geoprofessional Business Association (GBA). 2019. Proposed Best Practices for the Engineer of Record (EoR) for Tailings Dams. Accessed 3/30/2023 at: https://www.geoprofessional.org/news/published-proposed-bestpractices-for-the-engineer-of-record-eor-for-tailings-dams/ Global Industry Standard on Tailings Management (GISTM). 2020. GlobalTailingsReview.org. Accessed 11/15/2022 at: https://globaltailingsreview.org/wp-content/uploads/2020/08/global-industry-standard-ontailings-management.pdf International Council on Mining & Metals (ICMM). 2021. Tailings Management: Good Practice Guide. Accessed 5/25/2021 at: https://www.icmm.com/en-gb/guidance/innovation/2021/tailings-management-good-practice Mining Association of Canada (MAC). 2017. A Guide to the Management of Tailings Facilities. Accessed 12/7/2022 at: https://mining.ca/resources/guides-manuals/a-guide-to-the-management-of-tailings-facilitiesthird-edition/ Morrison, K.F., R.E. Snow, P.W. Ridlen and C.N. Hatton. 2017. What does it mean to be the Engineer of Record (EoR) for a Tailings Storage Facility (TSF)? In Tailings and Mine Waste ’17, Proceedings of the 21st International Conference on Tailings and Mine Waste. Banff, Alberta, Canada. National Institutes of Health (NIH). 2021. Succession Planning: A Step-by-Step Guide. Accessed 12/21/2022 at: hr.nih.gov/sites/default/files/public/documents/2021-03-Succession_Planning_Step_by_Step_Guide.pdf Spencer, D.L., C.A. Bareither, J. Scalia IV, C.N. Hatton and K.J. Ward. 2022. Characterizing tailings professional labor demand. Mining Engineering 74(277):15–25.

224

Chapter Three

Breach and Inundation Estimates

225

226

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Waste Dump Failure Runout Analyses: Applying Improved Empirical Correlation Methods to Waste Dump Datasets Trevor White, BGC Engineering Inc., Canada Andrew Mitchell, BGC Engineering Inc., Canada John Whittall, BGC Engineering Inc., Canada Scott McDougall, The University of British Columbia, Canada

Abstract Mine waste dumps are often constructed as loose, end-dumped slopes on steep terrain. Established practice is well suited for characterizing waste materials and their stability; however, available empirical tools to estimate the potential runout of waste dump failures provide very approximate results. A study in the 1990s established a database of coal mine waste dump failures and developed empirical tools to estimate runout distance and inundation area for mobile failures. The study found limited correlations to the source volume and other attributes, unless events are sorted by foundation conditions and the spreading potential in the travel path. This paper describes a re-evaluation of the earlier waste dump failure database to improve relationships between fall height, deposit volume, and runout length by incorporating additional sitespecific parameters to describe the nature of the runout path. Improved statistical techniques that have previously been applied to rock avalanches and open pit slope failures were used. A good correlation is observed between the runout length equation developed in this paper and the observed deposits when lateral confinement is applied as an indicator variable in a multiple linear regression analysis. This paper provides the methodology to apply the equation in practice and a description of the accuracy practitioners can expect from the relationship. The proposed method is demonstrated by analyzing the 1966 Aberfan failure.

Introduction Open pit mines commonly maintain operational efficiency by constructing waste rock piles or “waste dumps” in a rapid fashion, close to the ore source or pit. This constraint can place waste dumps in adverse terrain and foundation conditions, producing structures that may be susceptible to instability. Waste dumps are typically standalone structures that do not form infrastructure foundations. The consequences of a waste dump slope failure are generally quantified by the failure’s impact footprint,

227

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA specifically its runout length. To minimize risk, critical infrastructure may be excluded from within this estimated failure runout length. Runout lengths can be estimated using simple empirical relationships initially developed for natural landslides and adapted to rapid waste dump failures. The study by Golder Associates for the British Columbia Mine Waste Rock Pile Research Committee (BCMWRPRC, 1995) was a notable effort to establish empirical relationships relevant to coal mine waste dump failures in British Columbia. However, the resulting relationships have relatively weak correlations and therefore only provide very approximate runout estimates (Hungr, 2017). Recent studies of runout lengths for natural rock avalanches and open pit slope failures have produced significant improvements to runout predictability using multiple linear regression analyses (Mitchell et al., 2020; Whittall et al., 2020). This paper applies a similar multiple linear regression approach to the BCMWRPRC (1995) coal mine waste dump failure dataset in an effort to improve the predictability of waste dump runout lengths.

Background Waste dump failures forming long runout paths are generally contractive, brittle, and rapid failures (>3 m/min) of loosely dumped rockfill at its angle of repose (Hunter and Fell, 2003). Practitioners have relied on empirical relationships, which consider basic failure characteristics and dimensions, to predict potential waste dump failure runout lengths. Early empirical correlations developed by Heim (1932) for natural rock avalanches showed that the Fahrböschung angle (or reach angle) is inversely proportional to the landslide volume. The Fahrböschung angle is defined as the inclination of the line connecting the crest of the landslide source with the toe of the deposit (Figure 1).

Figure 1: Sketch of Fahrböschung angle and basic failure dimensions This method is purposely simple, and travel path characteristics such as the downslope angle, lateral runout path confinement, substrate material, or propensity for downslope entrainment are not explicitly incorporated, producing significant scatter. Subsequent researchers have built on Heim’s (1932) work to improve the accuracy of the resulting predictions and adapt the Fahrböschung angle concept to other

228

WASTE DUMP FAILURE RUNOUT ANALYSES: APPLYING IMPROVED EMPIRICAL CORRELATION METHODS TO WASTE DUMP DATASETS contexts (e.g., Scheidegger, 1973; Li, 1983; Corominas, 1996; Hunter and Fell, 2003; Mitchell et al., 2020; Whittall et al., 2020). BCMWRPRC (1995) investigated several empirical relationships specific to waste dumps, including Fahrböschung angle, and considered other failure attributes and qualitative indicators, including downslope topography and runout path material conditions. Two distinct classes of runout behaviour were identified: high mobility and normal mobility. However, the resulting empirical correlations were relatively weak. The Fahrböschung angle versus volume relationship was explored by Hunter and Fell (2003) in a study comprising 350 natural and constructed slope failures with both dilative and contractive initial failure mechanisms. Hunter and Fell (2003) highlighted the importance of considering the failure mechanism, type of slope, slide volume, geometry of the slide, and downslope topography angle and confinement when making correlations. Their results suggested that the runout length is sensitive to the downslope angle and insensitive to the volume over a range of 103 to 107 m3, the magnitude of most waste dump failures. Hungr (2017) provided the most recent and thorough review of runout estimate approaches applied to waste dump failures. Hungr (2017) cautioned that “the use of empirical correlations offers only very approximate means of predicting the runout of waste dump landslides.” Hungr (2017) suggested that the best approach to predict runout lengths may be to form unique relationships based on failures in similar geomorphological settings, rather than using previously generated regression equations.

BCMWRPRC (1995) dataset BCMWRPRC (1995) reviewed and summarized the characteristics of coal mine waste dump failures in western Canada. The study compiled a dataset of 46 waste dump failure cases and developed empirical correlation methods to estimate waste dump runout lengths. Table 1 summarizes the waste dump dataset reported in BCMWRPRC (1995). Of the 46 cases reported, 39 cases with sufficient information for multiple linear regression analysis were considered in the present study. Table 1: Summary of waste dump runout failure dataset as reported in BCMWRPRC (1995) Dump height (m)

Slide area (m2)

Deposit volume1 (m3)

Width (m)

Distance from toe (m)

Distance from crest (m)

Fahr2 (°)

Deposit area (m2)

Spread ratio

Crest to Distal El Difference (m)

Distal grade (°)

Avg.

200

60,300

930,000

200

920

1200

19

182,200

2.3

370

8

Min.

85

14,256

50,000

40

60

415

10

18,000

0.3

170

Runup

Max.

420

254,100

800,0000

600

3,500

3360

32

725,000

7.7

870

32

# of Reported cases

46

35

41

41

46

44

45

22

19

44

43

1. 2.

Deposit volume equals the sum of failed waste rock and entrained path material. Fahrböschung angle (Fahr).

229

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The waste dumps in the BCMWRPRC (1995) dataset were constructed using typical end-dumped waste rock (sedimentary/metasedimentary), progressively filling the steep U-shaped valleys, with some piles up to 400 m high. End dumping causes the rock to segregate by grain size, forming a coarse basal layer and slopes at the angle of repose (~38°), and also produces low relative densities and heterogeneous internal layer structures (Dawson et al., 1998). When founded on steep, loose substrate, rapid undrained failures with long runout lengths may occur (Dawson et al., 1998).

Empirical correlation methods Fahrböschung angle As a basis for regression relationship comparisons, Figure 2 plots the tangent of the Fahrböschung angle (H/L) versus volume (V) for the 39 waste dump failure cases used in the present study, producing the following regression line equation: H/L = -0.078logV + 0.79

(1)

where L is total horizontal runout distance (m), H is vertical fall height (m), and V is deposit volume (m3). The data demonstrates significant scatter about the mean, confirming the findings of BCMWRPRC (1995) and Hunter and Fell (2003). The regression statistics presented in Table 2 suggest that the relationship is statistically significant, but the scatter indicates high uncertainty. The root mean squared error (RMSE) is several hundred metres, supporting Hungr’s (2017) statement regarding prediction accuracy.

Figure 2: Linear regression of H/L vs V

230

WASTE DUMP FAILURE RUNOUT ANALYSES: APPLYING IMPROVED EMPIRICAL CORRELATION METHODS TO WASTE DUMP DATASETS Table 2: Fahrböschung angle regression fit statistics Coefficients

Estimate

Standard error

logV

-0.08

0.03

Intercept

0.79

0.15

Adjusted R2

0.16

p-value

0.006

Root mean squared error (m)

443

Normalized Index (NI) (%)

8.1

Multiple linear regression To improve the empirical correlations, the use of multiple linear regression (MLR) analysis was investigated. This technique was previously applied by Mitchell et al. (2020) for natural rock avalanches in the North American cordillera, using failure volume and fall height as predictors to estimate runout length. Mitchell et al. (2020) also incorporated the influence of downslope conditions on failure mobility using binary indicator variables. Based on the findings of Mitchell et al. (2020), the following multiple linear regression equation, which includes lateral confinement (C) as an indicator variable, was fit to the BCMWRPRC (1995) waste dump failure dataset: logL = β1logV + β2H + β3C + ε

(2)

where L is total horizontal runout distance (m), V is deposit volume (m3), H is vertical fall height (m), C is a binary variable indicating downslope topographic confinement (C = 1 is confined, C = 0 is unconfined), βi are regression coefficients, and ε is a normally distributed error term with zero mean (ε ~N(0,σ2)). Unlike Mitchell et al. (2020), the regression does not take the logarithm of fall height, as all cases in the BCMWRPRC (1995) dataset have fall heights within the same order of magnitude. Figure 3 shows a significant improvement in fit using the MLR approach, producing the following regression equation: logL = 0.1logV + 0.001H + 0.11C + 1.84

(3)

To present the MLR in two dimensions, the horizontal axis in Figure 3 is the product of volume and height. The indictor variable (C) is illustrated by the two regression lines, one for each confinement case. As shown in Table 3, the MLR approach improves the correlation (R2 of 0.73 compared to 0.16) and the prediction accuracy (RMSE 342 m vs. 443 m) compared to the previous Fahrböschung angle versus volume relationship.

231

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 3: Multiple linear regression of L vs. V, H, and C Table 3: Multiple linear regression fit statistics Coefficients

Estimate

Standard error

logV

0.10

0.03

H

0.001

1.9E-04

C

0.11

0.04

Intercept

1.84

0.19

Adjusted R2

0.73

p-value

1.1E-10

Root mean squared error (m)

342

Normalized Index (NI) (%)

2.8

Error analysis Figure 4 illustrates the distributions of regression residuals to assess the assumption of a normal distribution. The histograms for both the Fahrböschung angle relationship and the MLR relationship demonstrate approximate normal distributions. The Q-Q plots support the normal distribution assumption, but indicate more variation about the theoretical normal for the Fahrböschung angle relationship and a tail in the MLR residuals, suggesting there are outliers in the BCMWRPRC (1995) dataset.

232

WASTE DUMP FAILURE RUNOUT ANALYSES: APPLYING IMPROVED EMPIRICAL CORRELATION METHODS TO WASTE DUMP DATASETS

Figure 4: Regression residuals and Q-Q plots for: a) the Fahrböschung angle relationship, and b) the multiple linear regression relationship of L vs V, H, and C

Probabilistic waste dump runout estimates To generate the probable range of runout lengths estimated by the MLR regression equation, a probabilistic approach was adopted using survival functions (Mitchell et al., 2020; Whittall et al., 2020). Survival functions provide exceedance probability estimates; in this case, the likelihood of a runout length exceeding a point of interest for a specific failure fall height and volume. The following survival function equation was formed using coefficients from Equation 3 and the standard deviation of regression residuals:

"#$% – (.*+ , -.-.*"#$/ – -.--(01 – -.((2

P(L≥l| H = h, V = v, C = c) = 1 – Φ!

-.(-3

4, l > 0

(4)

where Φ is a cumulative distribution function of a normal variable, l is total horizontal runout distance (m), h is vertical fall height (m), and v is deposit volume (m3). Figure 5 plots survival functions for six sample cases from the BCMWRPRC (1995) dataset, where the MLR best-fit line equation is the 50th percentile of each function. Failures with confined and unconfined topography are shown in solid and dashed lines, respectively. Figures 5a and 5b show the observed values for three cases each with fall heights of 250 m and 500 m, respectively. Cases 12, 27 and 80 (Figure 5a) and Case 49 (Figure 5b) all plot relatively close to the expected value based on the regression equation. In

233

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Figure 5b, Case 44 is an example of a high mobility event, with an approximate 2% probability of exceedance, while Case 155 is an example of a low mobility event, with an approximate 87% probability of exceedance.

Figure 5: Sample survival functions for six cases from the BCMWRPRC (1995) dataset

Demonstration analysis A demonstration of the application of Equation 4 as a predictive tool was conducted using the 1966 Aberfan Tip 7 flow slide failure in South Wales. The event occurred when approximately 100,000 m3 of coal mine

234

WASTE DUMP FAILURE RUNOUT ANALYSES: APPLYING IMPROVED EMPIRICAL CORRELATION METHODS TO WASTE DUMP DATASETS waste failed, impacting the village downslope (Bentley and Siddle, 1995). Figure 6 shows an outline of the observed impact area from Hutchinson (1986) overlain on 25-foot resolution topography digitized from Davies et al. (1967). Figures 6 uses a modification of Mitchell et al.’s (2020) Probabilistic Runout Estimator (PRE) tool based on Equation 4 to illustrate the estimated probability of exceedance along a user-selected failure path. The topography downslope of the Aberfan tip was unconfined. The estimated probability of exceedance at the distal end of the observed impact zone is approximately 1.3 for undrained conditions in accordance with Swedish Dam Safety Guidelines (SveMin, 2021). For the challenges with stability and safety during construction, the main factors were the induced excess pore water pressures in the silt from the filling and the accompanying bearing capacity problems for the machines. During filling, material placed in 0.5 m lifts (see Figure 6) on top of the silt resulted in excess pore pressures, which thereby reduced the effective stress and in turn reduced the shear strength in the silt. With time, the excess pore water pressures dissipated and the effective stresses were thereby increased, with the result that the silt gained shear strength. In order to avoid any work environment or production-related problems for the contractor, a control program was initiated that helped the contractor to decide when the next lift of fill could be placed. The core of the control program was to monitor the dissipation of excess pore pressure in installed piezometers. The pore pressure dissipation was measured on-line using piezometers installed at two depths (0.3 and 0.7 m) in the silt layer (see Figure 6) in three separate sections of the dam. Pre-determined values for pore water pressures in the control program were used as indicators for when the next lift could take place. In practice the time for the pore water pressure to dissipate was shorter compared to estimations of time; the magnitude of developed pore water pressures was, however, fairly accurate.

Figure 6: Schematic concept of 0.5 m lifts placed on silt foundation and monitoring of pore water pressures at 0.3 and 0.7 m depth Figure 7 shows some photographs of construction on silt. The top left photograph shows the cable from piezometers (yellow shadow) installed in the silt before filling in 2020. The cables are collected on

356

TMF EXTENSION AT THE BJÖRKDAL MINE: DAM FOUNDATION CHALLENGES the upstream (north side) of the dam so as not to disturb further construction works and so as to be used during operation as well. The top right photograph shows placement of the first till layer in 2020 on the silt foundation. The bottom drone photograph is from early 2021 looking west from the left, showing the support fill including filter zones, the till core and the upstream erosion protection. Service roads used during construction can be seen on each side of the dam.

Figure 7: Dam K1, south extension. Top left (2020): piezometers installed in silt foundation. Top right (2020): Placement of first lift of till on silt (view from east to west). Bottom (2021): Overview of south construction site for silt foundation (view from east to west)

North extension, foundation The north extension stretches from the high point at the abutment of the existing dam to the next high point, approximately 300 m further eastwards and then another 150 m to the north, where the abutment at a crest elevation pf +147.0 masl meets the sloping natural ground. The high points are rock outcrops, and the natural ground in between consists of till, silt and, peat that are gradually overlying the bed rock into the low areas. It was decided that all weak and soft soils, such as peat and silt, in the foundation had to be excavated. This meant excavation to great depths (up to 6 m). To conduct these large excavations the works had to

357

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA take place during winter when the peat and silt were frozen. Winter also meant there was less water to manage; however, excavated slopes had to be stabilized to manage snow melt in spring.

Geotechnical conditions (north extension) The soil consisted in general of peat at the surface with underlaying silt, and till above bedrock. The foundation of the extension of the dam was planned to be on bedrock in the area of the two rock outcrops where the depth to bedrock was shallow. In between the rock outcrops, where the thickness of peat and silt layers were larger, excavation to underlying till was planned with a foundation mainly on till, or if necessary on bedrock. A layer of a minimum of 1 m competent till was needed for non-bedrock foundation. The peat and silt layers were approximately up to 1.2 m and 2.8 m respectively, and the till layer was approximately 0.2 to 2.7 m based on geotechnical investigations. At some locations the bedrock was sloping rather steeply (approximately 1V:5H), going in a northern direction at the east rock outcrop. Upstream of the dam extension there was a peat bog at a higher elevation, resulting in challenges with water management during construction.

Construction challenges (north extension) For the north extension of dam K1 the greatest challenge was the foundation. From pre-construction investigations, the estimated levels and excavation depths to bedrock were, in general, surprisingly good. The complexity of the rock surface (highly uneven), the quality of the bedrock (shale, laminated greywacke and graphite), and the amount of foundation on bedrock were not anticipated. The bedrock was extremely uneven within the foundation area, both on a macro and micro scale, which made excavation and especially cleansing of the rock surface difficult and time consuming. To manage the cleansing of the bedrock surface, a combination of water flushing and surface vacuum cleansing was used. This was time consuming and costly, but the result was good. See the photograph in Figure 8 showing the uneven bedrock surface (top left) and vacuum cleansing of the bedrock (top right). To determine whether the bedrock was competent enough, or if measures needed to be taken, a rock geologist mapped the whole bedrock surface. The different measures then used for the bedrock foundation were: • Grouting up to 8.0 m into the bedrock where there were deep cracks and fissures. • Levelling the rock surface with concrete where the surface was too uneven to place material such as till and filters. Figure 8 show reinforcement and levelling of bedrock with concrete (bottom left) and levelling of the bedrock with a special concrete, i.e., concrete with less water during mixing (bottom right). • Sealing the rock surface with concrete where the surface of the rock was fissured.

358

TMF EXTENSION AT THE BJÖRKDAL MINE: DAM FOUNDATION CHALLENGES

Figure 8: Bedrock foundation at dam K1, north extension. Top left: Uneven bedrock. Top right: Bedrock being vacuumed. Bottom left: Reinforcement and levelling of bedrock with concrete. Bottom right: Levelling bedrock with special concrete In addition to the challenges with the bedrock, water management was likewise a challenge as operation and tailings deposition were taking place during construction and the water level within the clarification pond downstream could not be lowered enough. This meant that the level of the foundation at most parts were below nearby water surfaces.

Lessons learned The raise and extension of dam K1 at the Björkdal mine have been challenging. The south and north extension of the existing dam required extensive foundation works. Unforeseen challenges were expected; however, the surprise was the extent of these challenges. In order to prepare for and minimize problems, the Björkdal Mine had signed a joint venture contract with the contractor, Peab, who was engaged at the design stage (however, not from the very start). This was done so that the practical experience of the contractor could be incorporated into the design and the construction documentation. Throughout the construction phase the designer (TCS), the contractor (Peab) and the Björkdal Mine project manager

359

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA continued to work in close cooperation, which enabled the team to find good solutions for design revisions when necessary, in order to secure the integrity of the function of the final dam. Time and budget were, nevertheless, heavily affected. The original time schedule to finalize construction of the dam to the crest elevation of +150.5 masl was set to October 2022. The outcome was a decision in 2021 to raise the dam to +147.0 masl until October 2023, a critical point in time to continue operations at the mine. The final construction cost for crest elevation +147.0 masl was approximately 125% of the originally estimated cost for crest level +150.5 masl. The increase in cost was mainly due to the complicated bedrock foundation, financial aspects (i.e., inflation), and underestimation of how complicated the construction of the dam would actually be. Technically the following lessons were learned: • Constructing new dams, which include large-scale foundation works on bedrock, is unusual in Sweden, where most dams were built in the 1960s and 1970s. This means that parties involved (owner, designer, and contractor) did not have a lot of practical experience, which affected the planning of work, time schedule, and budget. • More extensive pre-investigations would not have yielded better information on the quality of the bedrock. As pre-investigations showed shallow till cover over bedrock it could possibly, in hindsight, have been understood that large parts of the foundation would be on bedrock. That should perhaps have led to a decision on further pre-excavations to examine the rock surface. This would, however, have meant deep and challenging excavations through peat and silt layers in the design stage. However, the information was only available at the construction stage after the bedrock surface was cleared. • Cleansing of bedrock was time consuming and expensive. However, good results were achieved with watering and vacuuming the bedrock surface. Other methods were tested, such as brushing the surface, but they did not work well due to the almost concrete-like till closest to the bedrock surface. • Pore pressure dissipation in the silt at the south extension was faster than theoretically estimated. Due to the control plan and field monitoring the construction time could be shorter than estimated. • Quality assurance and quality control (QA/QC) were continuously adjusted to reflect the actual conditions on site for efficient construction and verification of desired function.

References SveMin. 2021. GruvRIDAS, Gruvbranschens riktlinjer för dammsäkerhet. (Swedish Dam Safety Guidelines). Stockholm September 2021, ABA. (https://www.svemin.se/om-oss/sveminskommitteer/miljokommitten/agda/#GruvRIDAS2021 )

360

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Tailings Stockpile Runout: The Case of Xingu, Brazil Marcos Túlio Fernandes, Vale, Brazil Pedro de Caralho Thá, Fugro, Brasil Willyan Giorgio Debastiani, Vale, Brazil Daniel de Oliveira Dourado, Vale, Brazil Felipe Jorge Teixeira, Vale, Brazil Leonardo Corradi Coelho, Vale, Brazil Zandra Almeira da Cunha, Vale, Brazil Gabriel Henrique Calais, Alvarez and Marsal, Brazil

Abstract Tailings stockpile are structures of great importance in mining operation and management. Their purpose is to store non-commercial materials such as waste rock and mining tailings. Due to their large volumes and heights, a potential failure of a tailings stack can have significant environmental impacts and may result in loss of life. Therefore, it is essential to assess the consequences of a stack failure in order to implement a mitigation and risk management plan. This study presents the runout of a hypothetical failure of the Xingu tailings stockpile located in the state of Minas Gerais, Brazil. Initially, a critical evaluation of available data was conducted, and a plausible failure mode was identified. The failure mode was identified as liquefaction due to the presence of loose saturated tailings, particularly in the base of the stack. The potentially mobilized volume considered the total mobilization of the stack volume. Runout simulations were performed using a turbulent Coulomb rheological model, in accordance with the volumetric concentration of solids in the stack. The mobilized friction angle used in the simulations was estimated using CPTu data. Simulation results indicated that the failure of the stack would generate very rapid flow, and the runout footprint would include mining operation areas, potentially affecting a railway located downstream. A secondary effect of the failure would be an obstruction in the Piracicaba River, located just downstream from the stockpile. The stack failure would result in water accumulation upstream and backwater effects, also affecting areas upstream of the river. A stability analysis was also conducted into the final thickness of the tailings in the landfill area.

361

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Mining is an activity with constant geotechnical risks that can result in losses to the environment and community. The study of hypothetical dam and stack breaches helps identify areas that could potentially be affected in the event of a structural failure. According to Jaboyedoff et al. (2020), considering the volume that can be mobilized and the distance traveled by the resulting flow is crucial to understanding the potential effects on the environment and downstream communities. Runout analysis is not yet well-established in geotechnical engineering, although there are a significant number of studies on slope failures. To conduct a study of the hypothetical failure of a stack, it is crucial to understand both the internal and external conditions that influence the stability of these structures. Regarding internal factors, it is necessary to consider geomechanical characteristics such as material strength and cohesion, as well as the presence of rock discontinuities and the slope geometry. External factors encompass aspects such as the presence of groundwater, which can affect percolation. As described in Ghahramani et al. (2022), depending on the material characteristics and site topography, the resulting tailings flows from the failure can travel at high velocities (exceeding 20 m/s) and reach long distances of flow and inundate widespread areas. Moreover, it is essential to understand the different modes and mechanisms of failure that can occur, analyzing each individual situation and using reliable data to support this analysis. There are various approaches available for modelling and detailed analysis of runout, which can be classified based on different criteria. Currently, common practice relies on integrated dynamic models in 2D, which are a standard approach using the shallow water theory from fluid mechanics and hydraulics. This approach has been extended to consider the flow of particulate material. The analysis is usually expressed in terms of total stress and considers a representative material for the landslide, as well as its properties along the total thickness of the slide. In some models, these properties may vary in space and time. The Xingu stack is part of the Mariana Mining Complex, owned by Vale and a component of the Alegria Mine, located in the municipality of Mariana, Minas Gerais, Brazil. The purpose of this stack is to store iron ore tailings from the mining operations at the Alegria Mine. The objective of the present numerical modelling of the runout is to evaluate the consequences of a potential total failure of the Xingu stack, based on a comprehensive examination of geotechnical data to define a credible runout scenario in terms of the volume and spatial distribution of the volume susceptible to failure, as well as the rheological properties of the tailings during the runout.

362

TAILINGS STOCKPILE RUNOUT: THE CASE OF XINGU, BRAZIL

Methodology Case study The case study was conducted on the tailings stockpile of Xingu, located at the Alegria mine, UTM coordinates 658398 E and 7769389 S. The stack has a total height of 71.74 m and a crest length of 838 m. Figure 1 shows a view of the Xingu stack.

Figure 1: View of Xingu stack

Proposed evaluation flow For the development of the hypothetical failure study of the Xingu stack, an evaluation flow was established, as shown in Figure 2. The adoption of an evaluation flow ensures that the study is concise and reflects the current conditions of the structure and the level of geotechnical knowledge available for the stack.

Figure 2: Flow diagram proposed for mine waste stockpiles runout analysis

363

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Geotechnical parameters used in the modelling Three undisturbed blocks were collected from within the stack to evaluate the strength parameters and characterization of the tailings in the laboratory. The results of these tests were compared with those conducted in other studies of the stack, including the paper by Espósito (2000). In addition to these blocks, field tests were performed using cone penetration tests (CPTu), a standard penetration test (SPT), and boreholes. Figure 3 shows the location where the undisturbed blocks were collected and the tests were conducted.

Figure 3: Location of in-situ and laboratory tests The results of the CPTu tests were used to assess the failure mode of the structure and define the parameters to be applied in the runout analyses. The assessment of liquefaction potential in the stack was carried out by estimating the state parameter, using the methods of Robertson (2010) (Equation 1) and Jefferies and Been (2016). 𝛹 = 0,56 − 0,33 ∗ 𝑙𝑜𝑔!" 𝑄#$,&'

(Eq. 1)

3333 𝑄𝑝 /ln 3 5 𝑘 𝛹=− 7 𝑚

(Eq. 2)

Where: 𝑄#$,() : Normalized corrected tip resistance; 3333 𝑄𝑝 : Corrected tip resistance; 𝑚 7 and 𝑘3: Parameters obtained from the critical state line. The definition of the basal friction angle to be applied in the runout modelling was determined by the liquefied strength ratio of the tailings. The strength ratio was determined using the methods of Olson and

364

TAILINGS STOCKPILE RUNOUT: THE CASE OF XINGU, BRAZIL Stark (2003) and Sadrekarimi (2014). The equations applied to define the peak and liquefied strength ratios can be seen in Table 1. Table 1: Equations applied to define the resistance ratio Method

Peak resistance 𝑠9,:; 0.85), they have the potential for liquefaction. Based on this, the failure mode considered in the study was liquefaction.

Figure 4: State parameters and undrained liquefied strength ratios

Runout study results Comparison of the runout study results obtained with the two friction angles Based on the geotechnical information, two simulations were conducted adopting friction angles of 4.5° (base case) and 2° (worst case), corresponding to the average liquefied strength ratio values and an extreme case. The friction angle considered in the worst-case scenario is lower than the minimum values indicated in the literature. According to Hungr et al. (2002), the back-analysis of runout resulting from the failure of overburden deposits with high mobility resulted in friction angles ranging from 3 to 6°. For these simulations, the liquefaction failure mode was considered, mobilizing the entire stack material and 167,457 m³ of supernatant water. Figure 5 presents the maximum tailings height footprint results obtained for the two scenarios.

367

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

b)

Figure 5 a): Maximum tailings thickness using a Coulomb residual friction angle of 4.5°

Figure 5 B): Maximum tailings thickness using a Coulomb residual friction angle of 2° For both scenarios, the flow of the Piracicaba River was accounted for as a secondary effect. This consideration is due to the different nature of river flow and runout. The runout is so fast that the movement of tailings practically stabilizes after four minutes following the breach. The runout is initiated by the gravitational force down the slope exceeding the basal frictional resistance due to the steep initial surface slope of the Xingu stack, which is greater than 4.5°. From the onset of movement, the tailings discharge is accelerated until the surface gradient becomes smaller than the adopted friction angle. At this point, the frictional forces balance the gravitational forces, but the movement can be sustained due to the inertia gained by the mass during the initial stage of the runout. No external restrictions are applied, except for internal frictional resistance.

368

TAILINGS STOCKPILE RUNOUT: THE CASE OF XINGU, BRAZIL The onset of the runout is quite similar for both scenarios; dynamic differences occur when the flow crosses the bed of the Piracicaba River towards the opposite slope. For the worst-case scenario considering a residual friction angle of 2°, the inertia of the runout mass carries a large volume of tailings towards the opposite slope of the valley beyond the Piracicaba River. After the runout decelerates, the tailings stabilize and adopt a surface gradient that is generally lower than the residual Coulomb friction angle due to inertia effects. Analysis of the terrain profiles along the Piracicaba River basin suggests that, in both scenarios, the runout reaches the riverbed, while the railway/road is less impacted by the runout in the base case. The impact on these two points can be seen in Figure 6.

Figure 6: Terrain profiles along cross-sections through the post-failure tailings volume for both scenarios. The upper right image corresponds to the post-failure tailings thickness of the base case and is used to define the cross-sections. Piracicaba river and road/railway lines are represented by a blue circle and a red triangle respectively on the terrain profiles

Secondary impacts on the Piracicaba River Using the post-failure topography, the potential volumes of water retained upstream of the runout deposits were assessed. The maximum level of the lake that can accumulate in the post-failure topography is defined by the saddle point corresponding to the maximum elevation point along the river direction and minimum elevation in the orthogonal direction, as shown in Figure 6.

369

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA For the base case scenario, the volume of the post-failure retention lake is 1.60 million m³ and increases to 3.67 million m³ for the worst-case scenario. The flow study shows that the flow of the Piracicaba River is 163 m³/s for a two-year return period (RP) and 442 m³/s for a 98-year RP. These flow rates suggest that the filling time is between 1 hour and 2.7 hours for the base case scenario, and more than twice as long for the worst-case scenario. To better understand the possible formation of a breach after tailings deposition and lake formation, a cross-section of the impacted area was evaluated. An analysis of the post-failure topography indicates that the flow direction is towards the railway area, exactly where the lake is controlled and where the deposited tailings thickness is approximately 70 cm. Given this, the most plausible scenario is that the natural flow quickly channels the fine tailings deposited at the saddle point and stabilizes in the more resistant pre-failure topography, creating an outlet channel where the flow would be directed northeastward and eventually return to the riverbed. The gradual lateral erosion of this initial breach would result in a wider channel finding its way towards the original thalweg of the Piracicaba River, but it is unlikely to cause a catastrophic runoff hydrograph comparable to the sudden breach of all deposits.

a)

b) Figure 7: Evaluation of the area affected by the interception of the Piracicaba River: a) Final soil elevation for the base case (cropped below 905 m and above 915 m) illustrating the computation of the saddle point defining the surface lake level. b) Area affected 30 minutes after Xingu failure (base scenario) An additional assessment was carried out for the base case to analyze the extent of the flooded area

created due to the river block, 30 minutes after the Xingu breach (time considered for the definition of the self-saving zone). Considering a constant flow in the river of 442 m³/s, the total volume accumulated in 26 minutes is 344,760 m³. Evaluating the final topography of the base scenario, the water level reached after 26 minutes is 904.09 m. Figure 7b shows the potentially affected area after 30 minutes of the Xingu failure, indicating an increase in the self-saving zone.

370

TAILINGS STOCKPILE RUNOUT: THE CASE OF XINGU, BRAZIL

Evaluation of impacts on railway stability The results indicated that the tailings are retained in the southeast area by the railway embankment. To assess the impact on the railway, a geotechnical evaluation of the effect of the tailings on the embankment stability was conducted. Detailed checks were performed for the base case scenario. However, the effect of the worst-case scenario was also evaluated. To assess the characteristics of the embankment, two Standard Penetration Test (SPT) boreholes and two Cone Penetration Test with pore pressure measurement (CPTu) tests were conducted. In the analyses, the railway surcharge was considered as a uniformly distributed load of 151 kN/m². The maximum velocities encountered in the affected region of the embankment can be seen in Figure 8. The total velocities were decomposed into the normal and tangential directions to the alignment of the embankment. The tangential velocities are greater than the normal velocities. The maximum normal velocity was 4 m/s, which leads to a maximum dynamic pressure of 20 kPa, calculated according to Equation 5. 𝑝56$ = 𝜌

𝑣3 2

(Eq. 5)

Where: 𝑝56$ : dynamic pressure; 𝜌 : tailings density (2500 kg/m³); e 𝑣 : velocity in the normal direction. 8 7

Velocity (m/s)

6 5 4 3 2 1

0 0

20

40

60

80

100

120

140

160

Distance (m) maxTailingsNormalVel [m/s]

a)

maxTailingsTangVel [m/s]

b) Figure 8: Evaluation of dynamic loads on the embankment: a) projection of the affected area; b) maximum velocities encountered The parameters of natural soil and embankment strength were estimated based on field tests. For the

embankment and natural soil material, the SPT results indicate an average value of 𝑁)78 equal to 14 and 26. Using the formulation by Teixeira and Godoy (1996), the friction angle is estimated to be 32° and 38°. Duncan and Wright (2014) provide characteristic values of effective cohesion and friction angle for different types of soils compacted to 100% of the Normal Proctor. According to them, the SM-SC and CLML soil types, which are closer to the embankment and foundation material, have a typical cohesion value of 15 kPa and 23 kPa, and a friction angle of 33° and 32°. As the embankment soil is considered silty sand and sandy silt, a cohesion of 5 kPa and a friction angle of 30° were considered for the embankment.

371

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Obviously, the natural soil is not compacted. However, due to the high value of 𝑁)78 , the parameters used in the stability analysis for the natural soil were a cohesion of 5 kPa and a friction angle of 32°. For the analysis, the tailings were considered as a material without shear strength, acting only as a fluid with a specific weight of 25 kN/m³. The summary of the parameters used in the analysis can be seen in Table 3. Table 3: Properties of the materials used in the analyses Material

Bulk weight (kN/m³)

Cohesion (kPa)

Friction angle (º)

Embankment

19

5

30

Natural ground

19

5

32

The stability of the railway embankment was evaluated for four different loading conditions, including: real condition without additional load; real condition with train overload; maximum dynamic pressure overload due to tailings and train; and maximum pressure overload from tailings and train. The stability analysis was performed using the limit equilibrium method in Slide 2 software by Rocscience. Figure 9 presents the result obtained for the worst loading condition. Table 4 presents the safety factors for all analyzed loading conditions.

Figure 9: Stability analysis for the worst-case scenario

372

TAILINGS STOCKPILE RUNOUT: THE CASE OF XINGU, BRAZIL Table 4: Slope stability results Case Load case Safety factor A No surcharge, downstream slope 1.54 B No surcharge, upstream slope 1.75 C Train load, downstream slope 1.54 D Train load, upstream slope 1.75 E Dynamic pressure + train load, downstream slope 1.54 F Dynamic pressure + train load, upstream slope 2.13 G Static pressure + train load, downstream slope 1.54 H Static pressure + train load, upstream slope 2.07 I Static pressure + train load, downstream slope (Worst-case scenario) 1.54 J Static pressure + train load, upstream slope (Worst-case scenario) 3.56 The results indicate that the stability of the downstream slope is not affected by any of the considered loadings, and as expected, the stability of the upstream slope improves with increased loading. As a sensitivity analysis, two additional cases (I and J) were evaluated, considering the full filling of tailings from the upstream area up to the crest of the embankment. These cases represent a static loading related to the final deposition of the worst-case scenario. Based on the results, the occurrence of stability problems in the embankment due to the hypothetical failure of the Xingu stack is not expected.

Conclusion The present study considered a single layer of tailings characterized by mixed rheology, with Coulomb friction based on a residual friction angle obtained from the liquefaction strength ratio of the tailings, and a variable Manning's roughness for turbulent dissipation. For both evaluated scenarios, the simulations show that the flow extends towards the Piracicaba River. The tailings movement resulting from the runout failure stops after 4 minutes. The result is the blocking of the river, with the potential accumulation of natural flow in an upstream retention area. Based on geometric arguments, the volume of water potentially accumulated in this lake was estimated at 1.60 million m³ for the base case and 3.67 million m³ for the worst case. The filling time of the retention area for the base case is estimated at 2.7 hours for a 2-year return period and 1 hour for a 98-year return period, while it is more than double in the worst case. It is important to note that the reported simulations do not assess the downstream consequences of overflow and possible erosion of the tailing’s deposits from the accumulation of water from the river in the retention area. Preliminary considerations suggest that the risk of sudden failure of the deposits and release of a massive dam-break wave in downstream areas is very low due to favorable topographic conditions at the predicted overflow points. However, these preliminary considerations do not represent a comprehensive assessment of the flood risk downstream of the potential accumulation water release.

373

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Based on the evaluation of the railway embankment stability, it was possible to identify that there is no impact of the stack failure on the railway, as corroborated by the found factors of safety for extreme loading conditions.

References Duncan, J.M., S.G. Wright and T.L. Brandon. 2014. Soil Strength and Slope Stability, 2nd edition. Espósito, T.J. 2000. Metodologia Probabilística e Observacional Aplicada a Barragens de Rejeito Construídas por Aterro Hidráulico. Thesis research, Universidade de Brasília. Faculdade de Tecnologia. Departamento de Engenharia Civil e Ambiental. Ghahramania, N., H.J. Chen, D. Clohan, S. Liu, M. llano Serna, N.M. Rana, S. McDougall, S.G. Evans and W.A. Take. 2022. A benchmarking study of four numerical runout models for the simulation of tailings flows. Science of the Total Environment 827. Hawley, M. and J. Cunning. Guidelines for Mine Waste Dump and Stockpile Design. Routledge, 2017. Hungr, O. 2017. Runout analysis. In Guidelines for Mine Waste Dump and Stockpile Design. Edited by Hawley, M. and Cunning, J. CSIRO Publishing. Hungr, O., R.F. Dawson, A. Kent, D. Campbell and N. R. Morgenstern. 2002. Rapid flow slides of coal-mine waste in British Columbia, Canada. In Catastrophic Landslides: Effects, Occurrence, and Mechanisms. Boulder, Colorado, Geological Society of America Reviews in Engineering Geology, v. XV, pp. 191–-208. Imran, J., G. Parker and J. Locat. 2001. 1D numerical model of muddy subaqueous and subaerial debris flows. Journal of Hydrodynamic Engineering 127(11):959–968. Jaboyedoff, M., D. Carrea, M-H. Derron and T. Oppikofer. 2020. A review of methods used to estimate initial landslide failure surface depths and volumes. Engineering Geology 267: 105478. Jefferies, M. and K. Been. 2016. Soil Liquefaction: A Critical State Approach, Second Edition. CRC Press. Olson, S.M. and T.D. Stark. 2003. Yield strength ratio and liquefaction analysis of slopes and embankments. Journal of Geotechnical and Geoenvironmental Engineering 129(8): 727–737. Robertson, P.K. 2010. Estimating in-situ state parameter and friction angle in sandy soils from CPT. 2nd International Symposium on Cone Penetration Testing, Huntington Beach, CA, USA. Sadrekarimi, A. 2014. Effect of the mode of shear on static liquefaction analysis. Journal of Geotechnical and Geoenvironmental Engineering 140(12). Spinewine, B., M. Sfouni-Grigoriadou and S. Ingarfield. 2015. Modelling subaqueous debris flows – a comparison of two state-of-the-art integrated models. Geophysical Research Extracts 17, EGU2015-12814-1. EGU General Assembly. Teixeira A.H. and N.S. Godoy. 1996. Analysis, project and execution on shallow foundations. In: Hachich et al. (eds) Foundations: Theory and Practice. São Paulo, PINI. 7, 227–264.

374

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Case Study: Decharacterization of Tailings Dike in the Iron Quadrangle, Minas Gerais, Brazil Thatyane Martins Gonçalves, Vale S.A., Brazil Ricardo Cabette Ramos, Vale S.A., Brazil Frank M.S. Pereira, Vale S.A., Brazil Gino Calderon Vizcarra, Vale S.A., Brazil Alessandro Lage de Castro, Vale S.A., Brazil

Abstract Current Brazilian legislation requires mining companies to eliminate all tailings dams built using the upstream raising method, process denominated decharacterization, in order to avoid the occurrence of similar collapses to the Brumadinho (2019) and Fundão (2015) dam failures. The dike located in the Iron Quadrangle region of the state of Minas Gerais was a tailings storage facility that contained 8.8 million m3 of tailings. It was built under the upstream method in four successive risings to reach a height of 21 m and a length of 606 m. In order to comply with Brazilian legislation, this structure was decharacterized in 2022. The geotechnical investigation comprised field tests such as SPT and CPTu soundings, and laboratory tests such as characterization, triaxial compression, and permeability. The engineering analysis consisted of slope stability analysis and numerical modelling to simulate all stages of the excavation works. Regarding the risks imposed by the decommissioning process, all activities were monitoring by automatized equipment, 24 hours per day, in a specific geotechnical monitoring center. The decharacterization consisted of a partial removal of the dike and upstream tailings. Furthermore, an embankment was built downstream to form a regular surface with a slow grade. In this approach, the dike was transformed into a tailings beach with free drainage. Also, a surface drainage system was built, composed of two main channels. In conclusion, the decommissioning process was successful and the learnings could be replicated in other tailings storage facilities. The objective of this paper is to describe the decommissioning process, which included extensive geotechnical investigation, engineering analysis, excavation works and geotechnical monitoring.

375

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Tailings dams are a critical and fundamental component of the mining industry’s development. For many decades, dams were commonly built to contain tailings using different methodologies, such as downstream raising, centerline or upstream. However, following the serious social, environmental, and economic impacts caused by the tragedies of the Mariana dam collapse (2015) and the Brumadinho dam collapse (2019) in Brazil, the construction of upstream raised dams has been prohibited in the Brazilian territory. Additionally, it is legally required to eliminate existing dams that were implemented using this same methodology, process denominated decharacterization. The legislation of the Brazilian National Mining Agency (ANM nº 13/2019) states that tailings containment structures that meet the requirements of the resolution should not permanently receive tailings and/or sediments from their main activity. The structure should no longer possess characteristics or function as a dam, and should go through the stages of decommissioning, hydrological and hydrogeological control, stabilization, and monitoring. Considering the recent dam failures, recent studies have highlighted the importance of considering liquefaction mechanisms in stability analyses of these structures (Turan et al., 2022; Silva et al., 2023). This phenomenon can be triggered by seismic events, vibrations induced by human activity, or changes in soil saturation levels (Zhang et al., 2022). In this context, given the criticality and associated risks of the removal process for structures built on tailings, the design development requires a range of assessments and preliminary studies prior to implementation, as well as careful attention to geotechnical challenges during the implementation activities. This case study refers to the removal of a tailings containment dike. It was constructed on a tailings beach, characterized as an upstream raised structure. This dike is among the first twelve dams to be decharacterized, out of the thirty to be eliminated by the mining company Vale S.A.

Decharacterization of a tailings containment dike The dike is in the Iron Quadrangle region in the state of Minas Gerais, Brazil, and is an upstream raised iron ore tailings containment structure. It consists of a starter dike, founded on the tailings of the main Dam reservoir, this tailings foundation has an approximate thickness of 40 m, and three upstream raises constructed using the upstream method. The structure has a crest length of 606 m and a maximum height of 21 m. The reservoir contains approximately 8.8 million cubic meters. It underwent a reinforcement process through the construction of a buttress to enhance the safety factor prior to the commencement of the decommissioning activities. Figure 1 shows the dike structure and Figure 2 depicts a typical cross-section of the existing dike prior to its removal.

376

CASE STUDY: DECHARACTERIZATION OF TAILINGS DIKE IN THE IRON QUADRANGLE, MINAS GERAIS, BRAZIL

Figure 1: Photograph of the dike embankment prior to stabilization works (Source: Vale)

Figure 2: Typical cross-section of the dike with reinforcement (Source: Vale)

377

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Currently, the structure no longer exhibits the aforementioned embankment characteristics in accordance with Brazilian National Agency (ANM) Resolutions No. 95 and No. 130.

Geotechnical campaign and material characterization Several geological-geotechnical investigation campaigns were carried out on the dike to obtain information about the materials used in the construction of the structure. Among these campaigns, the following tests were conducted on the different materials that constituted the starter dike, 1st and 2nd raises, in order to develop the geotechnical model of the structure: sedimentation grain size tests, specific gravity of grains, permeability under variable load, moisture content tests, compaction tests, Atterberg limits, and consolidated undrained triaxial compression tests (CIU). These investigations aimed to gather crucial data for the geotechnical model of the structure. The tailings present in the foundation and reservoir were characterized through an extensive campaign of Cone Penetration Tests (CPTu) and Seismic Cone Penetration Tests (S-CPTu). Based on the pore pressure generation profiles, it was possible to define the contact between tailings with a predominantly fine behaviour, characterized by significant pore pressure generation, underlying a layer of tailings with low pore pressure generation typical of granular materials. Furthermore, it was observed that there is a soil layer characterized as a deposit of colluvium with typical fine material behaviour. Table 1 presents a summary of geotechnical parameters which were obtained by applying the methodology of Olson and Stark (2003). Tests were also conducted on the underflow tailings present in the planned borrow area to assess the suitability of this material for the buttress construction to ensure an adequate safety factor and the fill material for the decommissioning phase of the dike. After development of a trial embankment and a laboratory testing campaign, it was determined that the material exhibited dilative behaviour at a degree of compaction of 98% regarding the Standard Proctor tests. This fact confirmed the suitability of this material for use in the buttress construction. Table 1: Summary of main geotechnical parameters Material

γ (kN/m³)

c’ (kPa)

Φ’ (°)

su/σ’v0,peak

su/σ’v0,liq

Fine tailings

22

–

27

0,20

0,05

Coarse tailings

22

–

30

0,25

0,08

Embankment

21

29

28

–

–

Rockfill

22

0

40

–

–

Buttress

22

0

37

–

–

Buttress reinforcement

22

0

35

–

–

Colluvium

19

–

–

0,20

–

378

CASE STUDY: DECHARACTERIZATION OF TAILINGS DIKE IN THE IRON QUADRANGLE, MINAS GERAIS, BRAZIL

Material

γ (kN/m³)

c’ (kPa)

Φ’ (°)

su/σ’v0,peak

su/σ’v0,liq

Residual soil

18

0

33

–

–

Design solution The proposed solution for the dike decharacterization design involves the implementation of upstream excavations and the construction of a downstream embankment with the aim of achieving a surface declivity of 3.50%. The development strategy was designed to be implemented during the dry season when rainfall in the region is minimal. The execution process was divided into eight stages (Figs 2-3) to minimize the risk of failure during design implementation. The execution stages considered in the development of the design were: •

Stage 1: Construction and maintenance of access roads and installation of automated monitoring instrumentation;

•

Stage 2: Adjust and preparation of drainage systems for construction

•

Stage 3: Construction of the downstream compacted fill with dry underflow tailings;

•

Stage 4: Installation of upstream sumps and pumping systems, relocation of instrumentation, and begin of excavations;

379

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 3: Stage 1 to 4 of decharacterization design •

Stage 5: Implementation of the transition over the existing berm’s rockfill;

•

Stage 6: Completion of the downstream fill with a slope of 3.5%, implementation of peripheral channels and flow direction berms; demolition of the remaining spillway section and site contouring;

•

Stage 7: Implementation of the operational access for interconnection of the decommissioned structures;

•

Stage 8: Contouring and grass planting in the intervened area.

380

CASE STUDY: DECHARACTERIZATION OF TAILINGS DIKE IN THE IRON QUADRANGLE, MINAS GERAIS, BRAZIL

Figure 4: Stage 5 to 8 of decharacterization design The hydraulic structures were designed for a return period of 500 years, with flow rates obtained by transforming rainfall using the quantiles of rainfall height from the Intense Rainfall Study and Maximum Precipitation (PMP) Calculation. The maximum allowable velocity for flow in unlined channels is 1.50 m/s, ensuring velocities compatible with the terrain and avoiding erosive processes. Based on these premises, the drainage channels were designed with a trapezoidal section, with side slopes of 1v:2.5h, a gradient of 0.5% and a base width ranging from 6m to 25m, fully lined with rockfill.

381

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA In addition to the existing monitoring system in the dike, the design includes the installation of additional monitoring instruments to enhance the geotechnical monitoring during the construction period and, consequently, control the risk of potential instabilities during the activities. Table 2 presents the type and total quantity of instruments during the decommissioning process, and Figure 5 depicts a typical monitoring section. Table 2: Summary of main geotechnical parameters Instrument type

Quantity

Piezometer

68

Water level gauge

1

Geophones

5

Figure 5: typical cross-section of the decommissioned structure

The design development considered the material characterizations obtained from field and laboratory tests, hydrogeological studies of the reservoir area, experimental data, and complementary studies. The final geometry ensures the minimum required safety factors of 1.5 for static analyses with peak parameters, 1.1 for pseudo-static analyses, and 1.2 for analyses considering liquefied parameters. It is important to emphasize that, as an additional safety measure and best practice, all studies and design phases, from conceptual to executive, underwent design review evaluation conducted by an independent technical team prior to their conclusion and implementation.

Challenges and implementation of the solution Throughout the process of development and subsequent implementation of engineering designs, we encountered a series of significant challenges. This section provides a concise list of the challenges faced: • Geotechnical investigation campaign: Conducting the survey campaigns proved to be a challenge due to the restriction on the use of fluids during the drilling process, aiming to ensure the geotechnical safety of the structures built using the upstream method. The drilling alternative

382

CASE STUDY: DECHARACTERIZATION OF TAILINGS DIKE IN THE IRON QUADRANGLE, MINAS GERAIS, BRAZIL methodology, hollow auger, presents difficulties in reaching greater depths, especially in saturated tailings, and it is also commercially scarce in Brazil. The imposed limitations were overcome by compiling existing geological information, conducting geophysical tests, and conducting an extensive CPTU campaign for characterization;

Figure 6: Progress of the excavations • Drainage: A significant modification was implemented to ensure the water quality of reservoir dike during the decharacterization process. Existing flooded areas were utilized for collecting and storing sediments instead of constructing sediment ponds. Surface runoff was conducted to these flooded areas through little trenches and a pumping system. In this way, the risk of new sediment pond failure was eliminated.

383

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Excavation of saturated tailings: As the excavation progressed, the increasing saturation content of the tailings resulted in a significant loss of bearing capacity, which posed serious difficulties for the movement of larger equipment within the reservoir. In order to overcome this limitation and maintain excavation efficiency, load corridors were established using dry material sourced from a borrow area. This measure aimed to ensure operational safety during equipment traffic. Figure 6 depicts an illustration of the load corridors created within the operational scope to facilitate the progress of the excavations. • Criteria for work suspension: The progress of the construction was subject to strict monitoring criteria due to the compressible nature of the foundation tailings contained by Reservoir Dike 03, which are susceptible to liquefaction. The construction activities of embankments, excavations, and vibrations generated by the machinery represent potential triggers for soil liquefaction. To minimize the possibility of liquefaction occurrence, geophones were adopted to control the vibrations acting on the structure, as well as automated vibrating wire piezometers to measure the increases in pore pressure resulting from the loading of the tailings from the reservoir of the dike due to the decharacterization activities. Work suspension occurred when the following conditions were observed: presence of sand boils in the fill, cracks, fractures, or anomalies in the dike or fill, abrupt rise of piezometers above 5.0 kPa for 24 hours, exceeding the limits of the instruments by the piezometers, recording of peak ground velocity (PGV) exceeding 10 mm/s by the geophones, continuous recording of PGV above 5 mm/s, and recording of PGV exceeding 5 mm/s by the geophones accompanied by excess pore pressure in the piezometers.

These challenges were addressed through a combination of technical expertise, collaborative problemsolving, and proactive design management. By overcoming these obstacles, the successful implementation of the solution was achieved, as presented in Figure 7.

384

CASE STUDY: DECHARACTERIZATION OF TAILINGS DIKE IN THE IRON QUADRANGLE, MINAS GERAIS, BRAZIL

Figure 7: Orthophotos (a) before the decharacterization works and (b) after the decharacterization works

After the implementation of the design, a new verification of the stability safety factors was carried out, and it was observed that, according to the objective, the values are higher than those required by the standards and the design, as presented in Table 3.

Table 3: Summary of main safety factor after decharacterization. Condition

Minimum required Safety factor

Safety factor reached

Undrained with peak parameters static

1,5

1,6

Undrained with peak parameters pseudo-static

1,1

1,4

Conclusion In conclusion, the decharacterization design of the dike located in Minas Gerais was successfully executed by implementing a design solution that involved upstream excavations and downstream embankments. The

385

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA design was divided into eight stages to minimize the risk of destabilization during implementation. The hydraulic structures were designed to withstand a return period of 500 years, ensuring the safety of the system. Additionally, a comprehensive monitoring system was installed to enhance geotechnical monitoring throughout the construction process and mitigate potential instabilities. The development of the design considered material characterizations, hydrogeological studies, and experimental data to ensure that the final geometry met the required safety factors. Furthermore, the design underwent independent design review evaluations at each phase, emphasizing the commitment to safety and best practices. Throughout the design, several significant challenges were encountered and successfully overcome. These challenges included conducting geotechnical investigations under restrictions, implementing alternative drainage solutions, excavating saturated tailings, and establishing criteria for work suspension to prevent soil liquefaction. By overcoming these obstacles, the implementation of the design solution was achieved, ensuring the safety and stability of the decharacterization process. The successful execution of the design demonstrates the effectiveness of the proposed solution and sets a valuable precedent for similar engineering endeavours.

References Chang, D., Guo, H., Yin, J. and Xu, W. 2021. Liquefaction analysis of a tailings dam in the Yunnan-Guizhou Plateau, China. Environmental Earth Sciences 80(5): 1–13. Olson, S.M. and T.D. Stark. 2003. Yield strength ratio and liquefaction analysis of slopes and embankments. Journal of Geotechnical and Geoenvironmental Engineering 129(8): 727–737. Silva, M.A.S., Nogueira, F.C. and Nunes, C.B. 2023. Liquefaction analysis of a dam foundation considering local site effects. Journal of Geotechnical and Geoenvironmental Engineering 149(1), 78–89. Turan, N., Kaya, I. and Balcı, C. 2022. Investigation of liquefaction potential of tailings dams using cyclic triaxial tests. Environmental Earth Sciences 81(9): 1–14. Zhang, J., Zhang, J., Tian, H., Guo, W. and Huo, W. 2022. Liquefaction risk assessment of tailings dams considering seismic and dynamic properties of tailings. Journal of Earthquake Engineering 26(1): 64–86.

386

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

From Design to Construction: The Role of the EoR in the Construction of a Filtered TSF Located in a Remote Area Camilo Morales, SRK Consulting, Chile Esteban Barría, SRK Consulting, Chile Francisco Vera, Gold Fields, Chile Johan Boshoff, Gold Fields, Australia

Abstract After the collapse of several mining waste facilities over the last decade, the mining industry was forced to raise its standards to ensure the safety of mine waste facilities. As a result, the Global Industry Standard on Tailings Management (GISTM) was developed. This standard sets out clear guidelines and expectations for companies on managing these facilities safely to prevent tailings facility failures and ensure a high level of safety, governance, and environmental protection. The implementation of the GISTM reinforced the role of the Engineer of Record (EoR), who is responsible for ensuring that the design, operation, construction, and closure of the tailings storage facilities (TSFs) are performed in accordance with the applicable regulations and original design intent. However, since most tailings storage facilities (TSFs) were already operational worldwide, most EoRs had little or no opportunity to correct legacy issues regarding the design or construction stages. This paper presents the experience of SRK Consulting (SRK) with the Salares Norte project owned by Gold Fields. Salares Norte is a greenfield gold-silver project located in the Atacama Region of Chile, a high mountainous environment (over 4,000 m.a.s.l.) with extremely arid conditions and high seismicity. At Salares Norte, the filtered TSF is located on the waste storage facility (WSF) comprising compacted waste rock sourced from the pre-strip phase of the open pit mine. SRK, the appointed EoR, has been involved from the early stages of design to construction. This article summarizes the main design aspects of the project and the challenges that arise during the TSF construction stage related to applying the GISTM requirements to the industry’s current practices. Finally, the knowledge acquired from SRK and Gold Fields related to applying the EoR service during construction is shared.

387

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Since the significant impact of widely published global TSF collapses over the previous decade, the mining industry has improved and re-evaluated the overall dam safety aspects of mining waste storage facilities. This has resulted in significant transformations in tailings management, governance, community and social engagement, emergency response planning, and knowledge management. Moreover, risk identification, analysis, and management have also been improved. The GISTM has also magnified the importance of the role of the Engineer of Record (EoR). The EoR and the associated engineering company should demonstrate sufficient capabilities and resources necessary for the project, and ensure that the terms of the engagement accurately reflect the scope of services and responsibilities undertaken and that risks and liabilities are adequately addressed. Chile, recognized as one of the world’s largest copper producers with an annual output of 6 million tonnes in recent years (Cacciuttolo, 2022), is home to approximately 763 TSFs, the majority of which are inactive or abandoned, while only seven are presently under construction (Sernageomin, 2022). This being the case, only a handful of projects have been administered in alignment with the GISTM philosophy, from the design phase to construction, as is the case of the Salares Norte project. Salares Norte, a new gold and silver mine in the northern section of the Maricunga Belt in Chile’s Atacama Region, is located at an average altitude of roughly 4,500 meters above sea level (m.a.s.l.). The mine is owned by Minera Gold Fields Salares Norte (MGFSN), a subsidiary of Gold Fields Limited. Ore extraction, primarily through an open pit at 4,700 m.a.s.l., is incorporated into its operations. During its operational lifespan, the processing plant is projected to produce approximately 2 million tons of ore annually. The operational plan encompasses a hybrid scheme involving cyanide leaching, followed by Merrill Crowe and carbon-in-pulp (CIP) procedures. SRK Consulting (Chile) has been engaged in all engineering stages of the filtered TSF (FTSF), which comprises dry stacked tailings situated on the intermediate platform of the South Waste Storage Facility (WSF-South). The TSF is equipped with a contact water collection system, which includes a basal drainage system, a collection pond downstream from the TSF, and a storage pond on the southern side of the filter plant. Furthermore, an attenuation facility has been constructed downstream of the TSF to arrest and contain potential minor sloughs. Figure 1 provides a plan view of the TSF’s main components. This paper outlines the key components of the innovative design of the filtered TSF situated atop a Waste Storage Facility (WSF). Additionally, it details the construction process and highlights the primary challenges contractors, supervisors, and the EoR encountered during the design and construction phases. These challenges were exacerbated by the project’s remote location and the adverse weather conditions prevalent on the site.

388

FROM DESIGN TO CONSTRUCTION: THE ROLE OF THE EOR IN THE CONSTRUCTION OF A FILTERED TSF LOCATED IN A REMOTE AREA

Figure 1: Plan view of the FTSF and its main components

FTSF design Site conditions The TSF is a self-supported filtered tailings stack with an innovative design. The filtered tailings will be located on the existing south waste storage facility (WSF) prepared from waste rock sourced from the prestrip phase of the open pit mine. The south WSF fills a ravine where the underlying foundation soil stratum ranges from 4,473 to 4,375 m.a.s.l., maintaining an average elevation of 4,390 m.a.s.l. and an average basal inclination of 3%. The intermediate platform of the South WSF, where the initial phase of filtered tailings deposition will commence, has elevations between 4,434 and 4,430 m.a.s.l. According to the Köppen climate classification, the site’s climate is defined as Tundra (ETH ws), influenced by its high geographical elevation and zero precipitation. This climate categorization aligns with the conditions encountered on the highest peaks of the Andes Mountains, locations characterized by the year-round existence of ice and snow. In addition, the site is characterized by an average temperature of 2 °C, which typically does not surpass a maximum of 10 °C during the summer months, while winter months often have sub-zero temperatures.

389

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The basin lacks permanent water courses, and recent surface runoff is not visibly apparent, rendering it devoid of fluviometric data. This characteristic aligns with regional observations where runoff is typically linked to extreme precipitation, often resulting from the Altiplanic winter. This climatic phenomenon, occurring from November to March in the Andean sector, can also influence lower elevations. Precipitation events predominantly manifest as snowfall in winter (July to September) and, to a lesser degree, rainfall in summer, a climatic pattern that can be attributed to the Altiplanic Winter. Geologically, the site is predominantly characterized by Middle and Upper Miocene volcanic and subvolcanic rocks, subject to hydrothermal alteration. Alluvial deposits in the TSF fill the basin floor, whereas the bedrock primarily features porphyritic andesites, basaltic andesites, dacitic porphyries, and tuffs. Regarding the site’s seismicity, the seismic history is predominantly characterized by earthquakes induced by the subduction of the Nazca plate beneath the South American plate, causing several earthquakes over MW 7.0 in the past 500 years. In addition, a site-specific probabilistic seismic hazard assessment (PSHA) was carried out, obtaining a PGA of 0.33 g and 0.54 g for the design earthquake (return period of 475 years) and maximum credible earthquake (return period of 10,000 years), respectively.

Design criteria As previously explained, the filtered TSF is founded on the south WSF, which is made of run-of-mine (ROM) material from the pre-strip phase of the open pit mine. However, the construction of the upper 5 m of the intermediate platform has been limited to a maximum particle diameter of 18½" (47 cm) disposed of in layers of less than 1.35 m and traffic compacted by the haul trucks at full capacity to limit future settlements due to the TSF construction. Conservatively, the shear strength of the ROM material was estimated by referencing the work done by Indraratna et al. (1993). Finally, a transition layer, 0.3 m thick and comprising material with a maximum particle diameter of 4", was constructed on top of the waste rock platform to prevent possible puncturing of the overlying geomembrane. Figure 2 shows a typical cross-section through the facility (Section A of Figure 1).

390

FROM DESIGN TO CONSTRUCTION: THE ROLE OF THE EOR IN THE CONSTRUCTION OF A FILTERED TSF LOCATED IN A REMOTE AREA

Figure 2: Cross-section A of Figure 1 (not to scale) A liner system covers the filtered TSF floor area and slopes to prevent infiltration from tailings to the WSF. The liner system comprises a 4 mm thick high friction angle (HFA) bituminous geomembrane (BGM) on the downstream area of the base of the TSF, a 3.5 mm thick HFA on the slopes and a standard geomembrane on the upstream area of the TSF base. The BGM was selected based on the outcome of a trade-off between HDPE, LLDPE or BGM-type geomembranes considering the service life, performance in adverse weather, construction process and cost. The geomembrane will be installed in 3 phases (1, 2, and 3) as the TSF is developed and gains height, covering an area of approximately 570,000 m2. During the first construction stage, half of the filtered TSF base and the slopes of the adjacent hills up to an elevation of 4,433 m.a.s.l. will be lined (phase 1). The geomembrane is anchored on the slopes at 4,433 m.a.s.l. on the previously constructed platform, on the downstream side of the TSF and along the interface of phases 1 and 2. During the second lining stage, the geomembrane will be installed on the remaining surface of the TSF base (phase 2) and the slopes up to an elevation of 4,443 m.a.s.l. The third stage will eventually include lining all the slopes to an elevation of 4,470 m.a.s.l. The filtered TSF includes a drainage system installed on top of the liner (overliner system), consisting of a herringbone-type drainage configuration. The drainage system consists of secondary drains that collect seepage water. The secondary drains convey seepage water to the downstream end of the TSF via the main drain, where it discharges into the downstream ditch containing the collector pipe for contact water from surface runoff. The contact water is conveyed under gravity through a pipeline to the TSF collection pond located downstream of the TSF. The secondary drains have a constant slope of between -0.4% and -0.5% depending on the TSF floor geometry, while the main drain has a constant slope of -0.4%. Figure 3 shows the drainage system design.

391

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 3: Herringbone drainage system on the base of the filtered TSF The tailings are subjected to a cyanide detox step to reduce the weak acid dissociable (WAD) cyanide content, thickened in a high-rate thickener, and then pumped to the remote tailings filtration plant near the FTSF, where tailings are filtered with the use of vertical plate filters. The filtered tailings are recovered from the filter plant with a front-end loader and transported to the adjacent FTSF in trucks, where it is conditioned for moisture before being compacted on the TSF in layers of 0.3 m by a roller compactor. The resulting design is unique regarding spatial configuration on top of a WSF. Advantages of a layered and compacted filtered tailings design include improved safety, improved water efficiency, elimination of the need for an embankment, reduced impact in case of failure, reduction of seismic deformations and control of saturation within the facility. In addition, the designs aim to minimize water consumption and dust emissions and limit the TSF footprint to the minimum possible.

Facility classification The filtered tailings storage facility has been classified considering the consequences of a sudden failure of the facility according to international guidelines. For this classification, Chilean regulation (DS No. 248) and international standards, such as CDA and the ICMM, classify the TSF with a low-risk consequence, whereas ANCOLD classifies the TSF with a very low-risk consequence.

392

FROM DESIGN TO CONSTRUCTION: THE ROLE OF THE EOR IN THE CONSTRUCTION OF A FILTERED TSF LOCATED IN A REMOTE AREA

TSF construction and EoR role Earth-moving Throughout the DRF construction process, which began with preparing the South WSF intermediate platform, the EoR team undertook a series of technical inspection visits. Arcadis was permanently on-site as a construction quality assurance inspector (QA/QC). Based on this, the role of SRK included overseeing and documenting the construction process to ensure the design intent was met, quality assurance and the development of detailed construction records. Construction started with the earthworks on phase 1 of the intermediate platform by placing the ROM material and compacting the upper 5 m of the intermediate platform of the south WSF, as shown in Figure 4. The Gold Fields mining team constructed the intermediate platform while SRK assessed the waste rock regarding particle size distribution per the technical specifications.

Figure 4: Construction of the upper 5 m of the intermediate platform of the south WSF During the construction, some snowfall events made it difficult for the work to progress, especially the compaction activities. However, since the lower diversion channel and the north contour channel were already built, the water entering the impoundment was minimized. After the upper 5 m of rockfill was traffic compacted, the final layer of finer material was placed, leaving a smooth surface, as is shown in Figure 5.

393

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 5: Bearing surface ready for the geomembrane Before the BGM deployment in phase 1, geotechnical quality assurance tests of the intermediate platform were carried out to verify that the upper 5 m complied with the minimum requirements established in the design and technical specifications. The tests included in-situ density tests, large-scale particle size distribution tests, triaxial tests, and surface wave-based geophysical surveys, as shown in Figure 6. The results indicated that, in general, the minimum design parameters were met. Construction works could therefore continue uninterrupted and without any modifications. However, during the construction of phase 2, the same tests will be carried out to verify the consistency of the parameters. It is important to mention that third-party organizations and/or consultancy companies performed the material characterization to maintain independence.

Figure 6: (left) In-situ density test and (right) surface waves geophysical tests

394

FROM DESIGN TO CONSTRUCTION: THE ROLE OF THE EOR IN THE CONSTRUCTION OF A FILTERED TSF LOCATED IN A REMOTE AREA

Basal lines and drainage system During the EoR technical and quality assurance visits, it was observed that the deployment of the geomembrane was carried out in compliance with the design intent and technical specifications. However, minor changes were made, recorded in the as-built drawings. As part of the quality assurance campaign, it was verified that all the overlaps were at least 20 cm and that the welding and bevel quality observed were in line with the technical specifications. All this information was recorded in the installation protocols and geomembrane as-built test and installation plans summarizing the installation day, geomembrane thickness and type, dimensions, and the joint/seam location between the different liner panels. Figure 7 shows the BGM welding process.

Figure 7: Bituminous geomembrane welding process The contractor’s internal quality control plan included performing tests on geomembrane seams. The test included a vacuum chamber (ASTM D5641) and destructive tests to assess the tensile shear strength of the BGM (ASTM D7056). In addition, 356 ultrasonic tests (ASTM D7006) were carried out to assess the integrity of the seams. All the tests were recorded in the geomembrane test protocols indicating the seam location, thickness, shear strength, number of identified failures and liner extension, among other parameters. Based on these tests, all joints between panels met the acceptability criteria. Moreover, an external audit included destructive tests and arc testing over the entire surface of the BGM liner system to verify the integrity of the geomembrane. These results were in alignment with the test carried out by the contractor. The third-party tests detected minor perforations (about the size of a pin) across the surface of the geomembrane, allowing Gold Fields to take timely action. Figure 8 shows the arc testing of the BGM. Again, all repairs were documented in the repair and patch log protocols.

395

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 8: External audit by arc testing the geomembrane Non-acid forming (NAF) ROM material was considered to construct the drainage system. The ROM material was selected according to the particle size distribution defined during the design stage, trying to prevent it from being contaminated by fine material transferred by wind action due to the strong gusts experienced on site. To verify compliance of the waste rock with the design particle size distribution (PSD) envelopes, PSD tests of the rock were carried out, which were recorded in the construction protocols. Figure 9 shows the construction of one of the secondary finger drains (without a collection pipe), while Figure 10 shows the herringbone drainage system already completed (for phase 1) and covered with a geotextile.

Figure 9: Drainage system construction

396

FROM DESIGN TO CONSTRUCTION: THE ROLE OF THE EOR IN THE CONSTRUCTION OF A FILTERED TSF LOCATED IN A REMOTE AREA

Figure 10: Phase 1 waterproofed and with the drainage system already completed for phase 1

Engineer of Record role The first task that SRK performed as the EoR of the Salares Norte project was the Design Basis Report (DBR), where the design criteria, assumptions and constraints were summarized in a practical and single document that could be useful for Gold Fields authorities, contractors and independent technical reviewers. This document is part of Requirement 4.8 of the GISTM. In addition, either the EoR or deputy EoR performed a monthly technical visit to assess the construction quality, whether the design intent was being met and whether an accurate construction history was recorded. The EoR also focused on ensuring that proper quality control and assurance measures were in place during construction to support meeting established quality standards and recording design deviations based on modifications made in the field. It is important to verify that any design modifications do not have a material impact on the overall integrity of the facility and that any modifications are risk assessed and mitigated where possible. A Construction versus Design Intent Verification (CDIV) Report documents this process. The CDIV identifies any discrepancies between the field conditions and the design assumptions such that the design can be adjusted to account for the actual field conditions. Any residual risks are captured in a Deviance Accountability Report (DAR) per GISTM requirements. In line with GISTM Requirement 6.3, the EoR prepared a Construction Record Report (CRR). The CRR describes all aspects of the “as-built” product, including all geometrical information, materials, laboratory and field test results, construction activities, schedule, equipment and procedures, Quality Control and Quality Assurance data, CDIV results, changes to design or any aspect of construction, nonconformances and their resolution, construction photographs, construction shift reports, and any other relevant information.

397

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Conclusions and lessons learned SRK Consulting (Chile) has been supporting Gold Fields with the Salares Norte project from the early stages of design and has been performing the role of the EoR since March 2022. As a result, SRK has been involved in the project’s design and construction, which is not common since the implementation of GISTM. Some lessons learned from the EoR services provided for Gold Fields and SRK are summarized below. Construction Planning: It is highly recommended to establish construction milestones and carry out the construction by “packages”, where it is impossible to continue the construction if the quality control tests have not been carried out or approved and the respective protocols received. In this way, it is possible to better handle any pending information and avoid the complete construction with gaps. In addition, the contractor has the pressure to be quick in the protocol delivery. Additionally, Gold Fields included SRK. Construction reports (CDIV and CRR): It is advisable to develop these reports through the construction process (for example, drainage system), accommodating the progress of the works in the field. This provides a better reaction time for contractors and owners to review pending information and act promptly. Contractors: Raise the importance of the EoR to the contractors by giving visibility to the EoR through a defined flow of information in line with the contractual requirements. Moreover, it is very useful to have a shared digital folder (in the cloud) where the contractors upload protocols and as-built drawings for review by the EoR quickly and continuously.

References Australian National Committee on Large Dams. 2012. Guidelines on the consequence categories for dams. Cacciuttolo, C. 2022. Past, present, and future of copper mine tailings governance in Chile (1905-2022): a review in one of the leading mining countries in the world. Journal of Environmental Research and Public Health 19, 13060. Canadian Dam Association. 2014. Bulletin: application of dam safety guidelines to mining dams. Chilean Ministry of Mining. 2011. Supreme Decree N°248. Indraratna, I., Wijewardena, L.S.S. and Balasubramaniam, A.S. 1993. Large-scale triaxial testing of Greywacke rockfill. Geotechnique 43(1): 37–51. International Council on Mining and Metals (ICMM). 2020. Global Industry Standard on Tailings Management. Sernageomin. 2022. Catastro de Depósitos de Relaves en Chile (actualización 19-10-2022). (Data file.) Sernageomin. Accessible at https://www.sernageomin.cl/datos-publicos-deposito-de-relaves/

398

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Rockfill Stockpile Construction Over Loose Tailings Michael Etezad, WSP Canada Inc., Canada Ali El Takch, B2Gold Corp., Canada Ken Bocking, WSP Canada Inc., Canada Monica Ansah-Sam, Vale Base Metals, Canada Renee White, Vale Base Metals, Canada

Abstract Construction of a waste rock stockpile within a tailings management facility is currently ongoing at a mine site in Canada. The stockpile is being advanced over the top of loose subaqueously deposited tailings and accordingly, the potential for a static liquefaction failure of the rockfill structure is a concern. A field investigation program was carried out to determine the thickness and characteristics of tailings and rockfill, to understand the characteristics and liquefaction potential of the tailings, and to assess the impact of potential tailings liquefaction on the stability of the waste rock stockpile. The field investigation consisted of cone penetration tests, seismic cone penetration tests, sonic borehole drilling, electronic vane shear tests, sample collection, and the installation of vibrating wire piezometers. A static liquefaction assessment was carried out based on the data collected during the field investigation. Results of the liquefaction assessment indicated that most of the tailings exhibit contractive behaviour, and consequently they are potentially susceptible to static liquefaction. In this study, the tailings state parameter and the post liquefaction shear strength were determined using different methods available in the literature. Discussion regarding the results obtained using these methods and comparison of the methods is provided. To prevent slope stability failures should liquefaction occur, it was recommended that the leading edge of the advancing rockfill be buttressed by placing tailings against the toe of the advancing edge of the waste rock. Tailings buttressing was included in the design to achieve the minimum required factor of safety.

Introduction A tailings management facility (TMF) at a mine site in Canada is also used for co-disposal of potentially acid generating (PAG) waste rock. PAG has been progressively deposited into the TMF, with the deposition front slowly progressing into the pond. PAG waste rock has been deposited by dumping short and pushing the waste rock into the pond in lift thicknesses of up to about 10 m. The current extent of the PAG placement

399

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA area has a bench, which at the time of the study was at elevation 95 m, and the stockpile crest was at elevation 102 m. Initially, PAG was placed over existing lakebed sediments; however, the PAG placement area later expanded to overlay the subaqueously deposited tailings. Since the PAG waste rock has been placed onto the deposited tailings, and construction of the waste rock stockpile is ongoing, the stability of the stockpile is an important consideration. A geotechnical site investigation and a static liquefaction assessment of tailings was carried out within the co-disposal area. The field investigation program was required to determine the thickness and characteristics of tailings and PAG waste rock stockpile, to understand the liquefaction potential of the tailings, and to assess the impact of potential tailings liquefaction on the stability of the existing and future placed waste rock stockpile. Following the field investigation, laboratory testing, tailings characterization, liquefaction assessment, and stability analysis were carried out. The static liquefaction assessment was based on data collected during the field investigation. This paper provides a summary of the field observations, and geotechnical laboratory results. It also provides a summary of the liquefaction and stability analyses and provides recommendations for future waste rock placement and overall waste rock stockpile stability.

Geotechnical investigation program A site investigation was carried out, which consisted of pushing cone penetration tests (CPTs) and seismic cone penetration tests (SCPTs), drilling sonic boreholes, carrying out electronic vane shear tests (eVSTs), collecting samples, and installing vibrating wire piezometers (VWPs). CPTs were carried out to infer insitu geotechnical properties of the tailings and the underlying native soil. Seismic Cone Penetration Test (SCPTs) were carried out to check the aging of tailings and to provide data that could be used to carry out an assessment of the potential seismic liquefaction of tailings. The purpose of the sonic drilling was to collect samples and to log the subsurface fill and native soil. The CPTs and SCPTs and sonic boreholes were completed using a track-mounted sonic drill rig. A few offshore CPTs and sonic boreholes were carried out from a segmented spud barge. Vibrating wire piezometers (VWPs) were installed to monitor the changes in water pore pressure in contractive tailings. Figure 1 shows a plan view of CPT, SCPT, and borehole locations. A total of 9 CPTs and 2 SCPTs were pushed in tailings material, with 5 CPTs pushed offshore. Drilling through PAG rock fill was required prior to pushing CPT and SCPT at all onshore CPT and SCPT locations. Cone Penetration Tests and SCPTs were terminated at refusal due to hard ground conditions, typically in lake sediments and on till. The groundwater levels inferred from the CPT and SCPT data were estimated based on the results of pore pressure dissipation (PPD) tests carried out at each location.

400

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS

Subaqueously Tailings

Deposited

below

Pond

(Water Level Elev. 99 m)

Figure 1: Plan view of the 2021 CPT/SCPT and borehole investigation locations Electronic Shear Vane Testing (eVST) was carried out in two boreholes next to the CPTs per ASTM D2573 to assess the in-situ peak and remoulded vane shear strength of the tailings. Ten vertical sonic boreholes were drilled using a track-mounted sonic drill rig after the SCPTs and CPTs were completed. The sonic boreholes were drilled approximately 3 m away from selected CPTs and SCPTs. The boreholes were terminated in bedrock, except for five boreholes. SPT tests were carried out in the overburden soil (lake sediments and native soil). Seven VWPs were installed in 2 CPT and 5 borehole locations. Water content, particle size distribution (PSD), Atterberg limits and specific gravity tests were carried out on selected samples to determine the index properties of tailings and overburden soil. The tested tailings samples have an average specific gravity (Gs) of 4.26, which indicates the presence of high pyrite content

401

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA in the tailings. The average unit weight of tailings estimated from the average water content and specific gravity test results is 29 kN/m3. Atterberg test results indicate that the tailings are non-plastic.

Subsurface conditions The stratigraphy encountered in the onshore boreholes and CPTs generally comprised waste rock, underlain by tailings, overburden soil, and then by bedrock. The stratigraphy encountered in the offshore boreholes and CPTs/SCPTs generally comprised tailings underlain by overburden soil and then by bedrock. Overburden soil is mostly composed of sand, gravel, cobbles, and boulders with trace of silt. A layer of low plasticity clayey silt of variable thickness, (from approximately 0.3 m to 2 m thick), was observed under the tailings in the sonic boreholes. The PAG waste rock thicknesses varied from 0.0 m to approximately 16.5 m in the investigated borehole and CPT locations. Tailings thicknesses varied from approximately 8.6 m to 34.0 m.

Liquefaction susceptibility assessment Loose saturated granular soils and tailings tend to contract when subjected to static (monotonic) or cyclic (earthquake) loading. Contraction of these materials causes an increase in excess pore water pressure and decrease in the effective stress. This reduction of effective stress may result in sudden loss of shear strength and stiffness which can lead to flow of tailings.

38B

Methodology

Cone Penetration Test is considered one of the tools in the estimation of liquefaction properties of soil. Table 1 presents a summary of the data analysis methods used to interpret the CPT data. The interpretation of the stratigraphy and the estimation of the engineering parameters from CPT and SCPT data were caried out using CPeT-IT Ver.3.0 software developed by GeoLogismiki, together with an in-house spreadsheet. The tailings unit weight of 29 kN/m3 was used in the CPT interpretation and analyses. In summary, a combination of historical waste rock placement, geotechnical investigation, pore pressure response and modelling were used to provide recommendation on future waste rock placement.

402

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS Table 1: Summary of data analysis methods used for the interpretation of CPTs and SCPTs Interpretation of engineering parameters Stratigraphy Static Liquefaction Assessment

Methodology CPT profiles (Robertson (2016) and adjacent (sister) sonic boreholes

SBTn charts

Robertson (2016)

State parameter, 𝜓"

Method (1): Plewes et al. (1992) Methods (2 & 3): Robertson (2010) and (2022) (a)

Peak Undrained Shear Strength Ratio, Su(peak)/σ′v

Method (1): Olson and Stark (2003) Method (2): Cone and eVST correlations

Liquefied Undrained Shear Strength Ratio, Su(liq.)/σ′v

Methods (1 & 2): Robertson (2010) and (2022) Method (3): Olson and Stark (2003)

Notes: (a) The state parameter estimated from Robertson (2022) was calculated both with and without application of the correction on the soil behaviour type index (Ic) for tailings with Ic > 3.0. In the corrected method, Kc was capped to a maximum value of 8.64 for tailings with Ic > 3 (Kc is the correction factor to account for changing behaviour with increasing fines content that is used to calculate the normalized clean sand tip resistance (Qtn,cs) and hence the state parameter (ψ).)

The in-situ states (contractive or dilative) of soils and tailings were estimated using the CPT-based SBT classification system proposed by Robertson (2016) which describes the soils based on their mechanical behaviour recorded by the cone during CPT testing. A key parameter to screen for contractive behaviour in tailings and soils deposits is the in-situ state parameter (ψ). The state parameter was initially proposed by Been and Jefferies (1985) and represents the difference between the in-situ void ratio (ei) and the void ratio at critical state (ecs) of soil/tailings for the same effective stress. Generally, soils and tailings with ψ < -0.05 are considered to be dilative and they can resist strength loss during static shearing (i.e., they are not susceptible to flow liquefaction). On the other hand, soils/tailings with ψ ≥ -0.05 are considered to be contractive (loose) and susceptible to flow liquefaction and strength loss under rapid loading or cyclic shearing. State parameters were estimated using CPT-based screening methods developed by Plewes at al. (1992) and Robertson (2010 and 2022).

39B

53B

Results of liquefaction susceptibility assessment Tailings behaviour characterization

Visual observations during borehole logging and interpretation of the CPT data indicate the presence of interbedded layers of coarse-grained and fine-grained tailings. This is attributed to the variable deposition

403

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA points associated with movement of the floating tailings pipelines. Coarse-grained tailings are predominantly composed of non-plastic, loose to very loose silty sand materials with a fines content generally less than 50 %. Fine-grained tailings are predominantly composed of non-plastic, loose to very loose sandy silt materials with a fines content generally greater than 50 %. In general, CPT-based SBT classification charts obtained from Robertson (2016) show a good agreement with the visual observations. The Robertson (2016) SBT charts indicate that coarse-grained and fine-grained tailings are mostly contractive. The calculated shear modulus K(G) of tailings based on two SCPTs carried out shows values less than 330 which is an indication of minimal to no cementation and/ or aging of the tailings.

54B

In-situ state parameter

Profiles of the state parameter for three CPTs, estimated using various methods shown in Table 1, are presented in Figure 2. The results from state parameter calculation show that tailings at all CPT and SCPT locations have mostly a y greater than -0.05, indicating that the tailings are mainly in contractive state. Hence, the deposited tailings are susceptible to static liquefaction.

Figure 2: In-situ state parameters for CPT21-03, CPT21-06, and CPT21-09

404

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS

VWP pore pressure response field test program A field test was carried out to measure the pore water pressure response in the contractive tailings. A CAT D9T dozer and loaded CAT 777 haul trucks were driven back and forth over the waste rock bench for a duration of 15 minutes to 30 minutes at the location where VWPs had been installed and the pore water pressure response in the VWPs was measured. As shown in Figure 3, the data from the VWPs indicates no change or increase in pore water pressure due to the movement of equipment.

Figure 3: Field test program – VWP data at BH21-05 and SCPT21-05

Liquefaction triggering mechanisms As presented, the tailings are saturated and mainly contractive; therefore, they could be susceptible to static liquefaction. Contractive tailings could potentially liquefy if they are subject to one or more triggering mechanisms. Some guidelines and research papers (e.g., ICOLD, 2022; Robertson, 2022) caution against relying solely on the triggering mechanisms to provide a safe design due to the complexity in addressing all the triggering mechanisms in the short and long term, and recommend that the post-liquefaction analysis be carried out. Hence, the post-liquefaction stability assessment was carried out to check the stability of the waste rock stockpile considering that the tailings were to liquefy.

Stability assessment Analysis method and stability criteria Stability analyses were completed for the three selected sections using the limit equilibrium slope stability program Slope/W Ver. 2021 developed by Seequent. The Morgenstern-Price method of slices was employed

405

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA to find potential failure surfaces and their corresponding factor of safety (FoS). Both circular and noncircular slips were considered. The slip surfaces were optimized. Slopes were analyzed for three critical sections and for two loading conditions: • Peak undrained condition: Using peak undrained shear strength for tailings which represents the condition of a rapid rate of rise in the waste rock stockpile • Post-liquefaction condition: Using a liquefied undrained shear strength ratio for tailings which represents the condition during a full liquefaction event of tailings. The minimum required factor of safety (FoS) for peak and post-liquefaction conditions for the rockfill are 1.3 and 1.0, respectively. The recommended FoS for post-liquefaction analysis remedial design is generally around 1.1.

41B

Material shear strength parameters

The calculated peak and liquefied undrained shear strength ratio profiles obtained using the various methods listed in Table 1 are presented for three CPTs in Figures 4 and 5, respectively. As it can be seen, coarsegrained tailings have higher peak and liquefied undrained shear strength ratios when compared to finegrained tailings. Hence, the undrained shear strength ratios between coarse-grained and fine-grained tailings are differentiated in the stability analysis.

Figure 4: Peak undrained shear strength for CPT21-03, CPT21-06, and CPT21-09

406

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS

Figure 5: Post-liquefaction shear strength profiles for CPT21-03, CPT21-06, and CPT21-09 The shear strength parameters used for the stability analyses for peak and post-liquefaction conditions are summarized in Table 2. To account for spatial variability, the 33rd percentile of the peak undrained shear strength ratio (Su(peak)/σ′v) of all CPTs and SCPTs was estimated using the Olson and Stark (2003) method for both fine- and coarse-grained tailings. Similarly, the 33rd percentile of the liquefied undrained strength ratio (Su(liq)/σ′v) of all CPTs and SCPTs was estimated using the Robertson (2022) method for both fineand coarse-grained tailings. 19B

Table 2: PAG and foundation material shear strength parameters used for the stability analysis Material

Unit weight (kN/m3)

Su(peak)/σ′v = 0.25 Minimum strength = 15 kPa Su(peak)/σ′v = 0.21 Minimum Strength = 15 kPa

Post-liquefaction shear strength Su(liq)/σ′v = 0.09 Minimum strength = 1(1) kPa Su(liq)/σ′v = 0.04 Minimum strength = 1(1) kPa

Coarse-grained tailings

29

Fine-grained tailings

29

PAG waste rock

22

c′ = 0 kPa φ′ = 38°

c′ = 0 kPa φ′ = 38°

Overburden soil (lake sediments and native soil): contractive silt and clay

20

Su(peak)/σ′v = 0.21 Minimum strength = 15 kPa

Su(liq)/σ′v = 0.04 Minimum strength = 1(a) kPa

Overburden soil (lake sediments and native soil)

20

c′ = 0 kPa φ′ = 32°

c′ = 0 kPa φ′ = 32°

Peak shear strength

Note: (a) Based on Robertson (2022)

407

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA 42B

Stability analysis results The stability analysis results indicate that the existing waste rock stockpile benches meet the minimum required FoS for the peak undrained condition. However, the toe of the waste rock stockpile may experience some sloughing. The stability results of a critical section are shown in Figure 6. The post-liquefaction assessment carried out on existing waste stock benches for the three analyzed cross sections do not meet the minimum recommended FoS. A sensitivity analysis was carried out to check the effect of relatively thin cohesive lake sediments on the waste rock stability. The lower bound post-liquefied strength was used. A 3 m layer of contractive lake sediments (silt and clay) was modelled, which results indicate that the presence of a basal layer of soft lake sediments does not influence on the global stability of the waste rock. 20B1B

Figure 6: Static peak undrained condition stability analysis results

Recommendations for current and future PAG waste rock placement Additional stability analyses indicated that the adequate post-liquefaction FoS values can be obtained by following the two strategies below: • by buttressing the toe of El. 95 m waste rock stockpile bench (Fig. 1) with tailings to increase the effective overburden pressure and thus increase the resisting force against failure, and • by limiting the waste rock stockpile elevation to less than 100 m to decrease the driving mass along the critical slip during the post-liquefaction condition (Fig. 7b).

408

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS In order to model the future waste rock stockpile advancement towards the TMF pond, where no CPT and borehole data are currently available, it was conservatively assumed that the future tailings forming the foundation of advancing waste rock stockpile will consist of fine-grained tailings. Sensitivity slope stability analyses were carried out for the current foundation condition, for future waste rock stockpile placement, different crest height elevations, and different pond water elevations. The Factor of Safety was checked to ensure the minimum required FoS was achieved or met. A minimum required tailings toe buttress elevation was recommended for different stockpile crest heights (e.g., a tailings buttress toe elevation of 92.5 m for the stockpile crest elevation of 98 m as shown in Figure 7a). Under these conditions the tailings act as a buttress and the minimum FoS is achieved.

(a)

(b)

Figure 7: Remedial design stability results – Waste rock crest elevation at (a) 98 m, existing condition (b) 100 m, future condition Based on the review of the bathymetry survey, tailings deposited against the south side of the waste rock stockpile toe provide a buttress support which allows the post-liquefaction condition to be satisfied. However, some tailings need to be placed at the toe of the north side of waste rock stockpile (Fig. 1) so that the minimum post-liquefaction FoS can be achieved.

409

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Conclusion The liquefaction potential of tailings below a waste rock stockpile was assessed based on the results of the geotechnical investigation, which included field and laboratory testing. The liquefaction susceptibility analyses were carried out using CPT-based screening methods. Limit equilibrium analyses were carried out to assess the stability of the PAG waste rock benches. Results of this study are summarized below: •

Visual observations during borehole logging and interpretation of the CPT and SCPT data indicated interbedded layers of predominantly coarse-grained tailings and fine-grained tailings. The tailings are non-plastic.

•

The results of the liquefaction assessment indicated that most of the tailings tested at CPT and SCPT locations have a state parameter greater than -0.05 and were therefore considered to be in a contractive state (potentially susceptible to static liquefaction).

•

Field testing was carried out to measure the pore water pressure response in the VWPs installed in the underlying tailings in response to construction equipment on the surface of the overlying waste rock. The data from the VWPs showed no change or increase in the pore water pressure due to the movement of equipment along the PAG waste rock surface.

•

The post-liquefaction assessment carried out on the existing waste rock stockpile benches indicates that the waste rock stockpile would not meet the minimum required factor of safety if liquefaction occurred in the tailings underlying the waste rock in certain areas.

•

Under the existing and future operating conditions, adequate post-liquefaction FoS values can be maintained by buttressing the first stockpile bench toe with deposited tailings to an appropriate elevation. As waste rock stockpile placement into the TMF is continued, regular bathymetry checks were

recommended to confirm the elevation of tailings adjacent to the toe of the advancing waste rock stockpile. Placement of tailings at the toe of waste rock stockpile, as required, would be carried out to obtain the minimum post-liquefaction FoS.

References ASTM D2573 / D2573M-18. 2018. Standard Test Method for Field Vane Shear Test in Saturated Fine-Grained Soils. ASTM International, West Conshohocken, PA. Been, K. and M.G. Jefferies. 1985. A state parameter for sands. Géotechnique 35(2): 99–112. Hawley, P. Mark and J. Cunning. 2017. Guidelines for Mine Waste Dump and Stockpile Design. CRC Press. International Commission on Large Dams (ICOLD). 2022. Tailings Dam Safety (Draft).

410

ROCKFILL STOCKPILE CONSTRUCTION OVER LOOSE TAILINGS Jefferies, M.G. and K. Been. 2016. Soil Liquefaction – A Critical State Approach. Second edition. Taylor & Francis Group, LLC. Olson, S.M. and T.D. Stark. 2003. Yield strength ratio and liquefaction analysis of slopes and embankments. Journal of Geotechnical and Geoenvironmental Engineering 129(8): 727–737. Plewes, H.D., M.P. Davies and M.G. Jefferies. 1992. CPT based screening procedure for evaluating liquefaction Susceptibility. 45th Canadian Geotechnical Conference, Innovation conservation and renovation, Toronto, Canada. Robertson, P.K. 2010. Evaluation of flow liquefaction and liquefied strength using the cone penetration test. Journal of Geotechnical and Geoenvironmental Engineering 136(6): 842–853. Robertson, P.K. 2022. Evaluation of flow liquefaction and liquefied strength using the cone penetration test: an update. Canadian Geotechnical Journal 59(4): 620–624. Robertson, P.K. 2016. Cone penetration test (CPT)-based soil behaviour type (SBT) classification system – an update. Canadian Geotechnical Journal 53(12): 1910–1927.

411

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

412

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Transition to Filtered Tailings at LaRonde Gold Mine Edouard Masengo, Agnico Eagle Mines Ltd, Canada Edouardine-Pascale Ingabire, ArcelorMittal Mining Canada G.P., Canada Sylvain Boily, Canada Yanick Létourneau, Agnico Eagle Mines Ltd, Canada Patrick Laporte, Agnico Eagle Mines Ltd, Canada Francis Guay, Agnico Eagle Mines Ltd, Canada Marielle Limoges Shaigetz, Agnico Eagle Mines Ltd, Canada Jessica Huza, Agnico Eagle Mines Ltd, Canada Michel R. Julien, Agnico Eagle Mines Ltd, Canada

Abstract Since 1988, LaRonde Gold Mine (LaRonde) has exploited a polymetallic orebody in Quebec, Canada. Slurry tailings are deposited in two tailings storage facilities (TSF) named “Main TSF” and “Extension A4.” At the end of 2022, Main TSF and Extension A4 TSF reached their full storage capacity with slurry tailings deposition. In order to reduce the risks related to the slurry tailings management, LaRonde decided to transition to filtered tailings deposition in October 2022. This is an emerging technology that has been gaining in popularity over the past few years. A new filtered tailings stack with 12 million tons capacity is to be constructed over the existing Extension A4 TSF. The transition to filtered tailings required the commissioning of a filtration plant and the construction of a new water pond. To enable the construction of the filtered tailings stack, it was necessary to place a waste rock platform (bridgelift) over the slurry tailings stored within Extension A4. This paper describes the geotechnical investigations of the foundation soils and the slurry tailings carried out at Extension A4 TSF to support the design of the filtered tailings stack, and the construction, monitoring, and surveillance program implemented as of March 1st, 2023. The paper also discusses the results of large-scale field loading tests completed to verify and inform the design of the filtered tailings stack.

Introduction LaRonde Gold Mine (LaRonde) is owned and operated by Agnico Eagle Mines Limited (AEM) and is located approximately 60 km west of Val d’Or in the Abitibi region of north-western Québec, Canada. Since 1988, LaRonde has exploited a polymetallic orebody of gold, silver, zinc, and copper. LaRonde currently

413

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA operates 3.2 km below the surface, making it the deepest mine in the Americas.

Figure 1: Satellite view of Main TSF and Extension A4 TSF Over the years, ore was extracted from various pits and underground mines. As of 2022, LaRonde was operating LaRonde Zone 5 (LZ5) and LaRonde (LAR) underground mines. Two mills are in operation producing two different streams of tailings: LZ5 tailings and LAR tailings. Combined, the two mills have the capacity to process approximately 9,000 t/d. LaRonde Main TSF was constructed in 1988 and was expanded into the East Extension in 1997. Main TSF is confined by Dike 1, to the West, and Dike 7, to the East. Dike 2, a central internal embankment, divides Main TSF. The starter dikes were constructed using zoned earth fill and were built to an elevation of 337 m. Between 2000 and 2002, the perimeter embankments were raised, using the central method, to reach an elevation of 343 m. Between 2000 and 2019, Dikes 1 and 7 were incrementally raised by the upstream method at an average rate of about 1.0 m/year to their final elevation of 358 m. Main TSF covers an area of approximately 90 ha. The maximum height of the starter dikes is approximately 18 m at Dike 1 west. Extension A4 TSF was constructed in 2010 to elevation 332 m as part of the expansion of the underground mine. Extension A4 TSF is confined by Dike 10, which is a zoned earth fill embankment with a central low permeability till core. Extension A4 TSF covers an area of approximately 68 ha. The maximum height is approximately 22 m at Dike 10 south. Main TSF and Extension A4 TSF reached their full storage capacity with slurry tailings deposition at the end of 2022. Figure 1 shows a satellite image of Main TSF and Extension A4 TSF. LaRonde completed the transition to filtered tailings in October 2022. This paper focuses on the construction of the filtered tailings stack over slurry tailings in Extension A4 TSF.

414

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE

Tailings deposition Initially, tailings were produced ranging from 25% to 35% solid by weight. Approximately 35% to 40% of the tailings is used for underground cemented paste backfill produced in two paste plants for LZ5 and LAR mines. Therefore, the tailings deposition rate can vary daily from about 2,000 t/d, when both paste backfill plants are in operation, to 9,000 t/d when both paste backfill plants are not in operation. On an annual basis, the average tailings deposition rate is approximately 6,000 t/d. Between 2010 and 2015, Extension A4 TSF was used exclusively for water management. Subaqueous deposition occurred during the initial years of filling. A sub-aerial beach of increasing length was developed during the last years of slurry tailings deposition.

Geotechnical investigation Between 2019 to 2021, geotechnical investigations were carried out to characterize slurry tailings and cohesive soils in Extension A4 TSF to support the design of the filtered stack. The investigation program included 19 cone penetration tests (CPTs), 11 boreholes with SPT and Shelby tube sampling, and 12 test pits. At several locations, the CPTs were pushed close to boreholes for calibration purposes. The soundings were advanced to depths of between 2.7 m and 29.6 m. Shear wave velocity (𝑉! ) testing, seismic compression wave velocity (𝑉" ) testing, and pore pressure dissipation tests were completed at every CPT location. Additional field testing included electronic field vane shear testing (eVST) in the slurry tailings and Nilcon field vane testing (FVST) in the cohesive deposits. Inclinometers (INCs) and vibrating wire piezometers (VWPs) were installed at critical cross-sections at the downstream toe of Dike 10. Disturbed tailings samples and high-quality cohesive soil samples were collected for laboratory testing. Laboratory tests included index testing, for soil identification, and advanced testing to assess consolidation behaviour, hydraulic parameters, critical state lines, as well as peak and liquefied shear strength parameters. In general, the Extension A4’s foundation is characterized by glacial till overlying the bedrock. Discontinuous pockets of cohesive soils were identified above the glacial till. The thickest clayey layer is located in the northern section of Dike 10 east. The inferred thickness of cohesive soils is illustrated in Figure 2a, based on information collected since 2009. The cohesive soil thickness varies between 5 m to 6 m below the footprint of the northern section of Dike 10 east. The inferred thickness of slurry tailings prior to construction of the filtered stack is illustrated in Figure 2b. A maximum inferred slurry tailings thickness of 12 m to 14 m is located immediately upstream of Dike 10 south.

415

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 2: Stratigraphy within Extension A4 TSF footprint (a) cohesive soil thickness (b) tailings soil thickness

Material characterisation Slurry tailings within Extension A4 TSF can be described as a non-plastic silty sand (SM) to silt (ML). Fines content (𝐹𝐶) varies between 35% and 100% for LAR tailings. LZ5 tailings are generally finer, with 𝐹𝐶 ≥ 95%. Tailings segregate with depth and with distance from Dike 10 due to the temporal variations in the deposition environment. The varved cohesive deposit varies from a low-plasticity silt (ML) to clay (CH). In the upper 2 m to 3 m, an over-consolidated crust is noted, below which the clays are generally normally consolidated with a water content higher than the liquid limit and a liquidity index higher than 1.0. Results from FVSTs indicate a very soft to firm consistency in the normally consolidated clays.

Figure 3: Cone penetration tests along a cross-section of Dike 10 south

416

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE

Cone penetration testing Figure 3 shows the normalized cone resistance (𝑄) and pore pressure ratio (𝐵# ) profiles obtained from CPTs along a cross-section of Dike 10 south. Tailings segregation is apparent by analyzing the CPT signatures. Three types of LAR tailings were inferred from particle size distribution and CPT interpretation: coarse (35%< 𝐹𝐶 -0.05).

Filtered tailings 2022 project In order to reduce the risks associated with the slurry tailings management, LaRonde transitioned to filtered tailings deposition in October 2022. This change in the tailings management strategy was motivated by: • The reduction of the environmental footprint by minimizing the area impacted by mining activities. • A better management of the physical, environmental, and social risks. • The integration of progressive reclamation during the filtered stack operation.

417

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The transition to filtered tailings required the construction of a filtration plant over a three-year period and a new water management infrastructure, Cell 5, in the vicinity of Extension A4 TSF. Figure 5 shows an aerial view of the filtration plant and Cell 5. Cell 5 was designed to manage the watersheds of Extension A4 TSF and part of Main TSF, as well as the filtration plant process water. Cell 5 has a water storage capacity of 2.5 Mm3 under the design project flood. Cell 5 is confined by zoned earth fill embankments and is lined with an LLDPE geomembrane.

Figure 5: Aerial view of the filtration plant, Extension A4 TSF, and Cell 5 (September, 2022) The filtration plant was designed to accommodate the fluctuating tailings production tonnage and the large range of LAR:LZ5 tailings ratio. On an annual basis, it is expected the tailings production will be close to a ratio of 45LAR:55LZ5. From the two mills, the tailings flow through a 36-meter diameter thickener to increase the solids content from approximately 32% to approximately 60%. At the output of the filter presses, the targeted water content is 19% for a 90-second drying time. The water content varies within ±3% for the same drying time to account for the different LAR:LZ5 tailings ratio. These design parameters were selected to be able to produce the filtered tailings as close as possible to their optimum water content to achieve a minimum of 95% standard maximum dry density (SMDD). The filtered tailings are transported by a conveyor to a temporary stockpile located south-west of Extension A4 TSF. The filtered tailings are then transported by trucks to Extension A4 TSF, where a filtered stack is built over the slurry tailings. For trafficability purposes, the placement of the filtered tailings stack required the

418

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE construction of a 1.5 m to 2 m thick bridgelift composed of waste rock over the slurry tailings. The filtered stack is planned to be constructed to an elevation of 353 m with 5H:1V side slopes, in addition to a stability toe berm. The final configuration of the filtered stack is illustrated in Figure 6.

Figure 6: Final configuration of the filtered tailings stack in Extension A4 TSF

Figure 7: Development plan of the filtered tailings stack The filtered stack will be constructed to its final height, sequenced in 3 different zones, as illustrated in Figure 7: Z1 at the south, Z2 at the east, and Z3 at the north-west. This development strategy allows the engineers to: • Construct the bridgelift progressively. • Construct the tailings stack at a raising rate which satisfies geochemical and geotechnical stability. • Limit the tailings exposed area to reduce dust generation. • Update the design and operation of subsequent zones based on the observed previous performance of the filtered stack, slurry tailings, and foundation materials. • Start progressive reclamation of the completed zones of the tailings stack.

419

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Filtered stack operation Construction The construction of the bridgelift started in Spring of 2022 in Z1 with the objective of capping 175,000 m2 of Extension A4 TSF by the commissioning of the filtration plant. Over 800,000 tons of waste rock were placed with a bulldozer and were nominally compacted by the movement of the earthwork equipment (bulldozers and trucks). A series of control measures were put in place to reduce the risks related to the construction of the bridgelift over wet and saturated tailings. First, the water level was progressively lowered in order to create a large tailings beach before the bridgelift construction began. Also, the contractor was instructed to work on a wide front to allow the dissipation of the excess porewater pressures of the slurry tailings. If local instability of the bridgelift was observed during the movement of the earthwork equipment, construction activities were stopped in the affected area to allow the dissipation of the excess porewater pressures. It was generally observed that a 24-hour resting period was sufficient and that the construction activities could resume safely the following day. The filtered tailings were placed in a 300 mm lift and were compacted to 95% SMDD using a bulldozer and a vibrating compactor. Z1 working area was divided in 8 blocks to allow deposition and resting periods. This strategy allowed the dissipation of excess porewater pressures in the slurry tailings and in the cohesive soils. Figure 8 and Figure 9 present photographs of the construction of the bridgelift and the filtered stack respectively.

Figure 8: Bridgelift construction over Extension A4 TSF (September, 2022)

420

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE

Figure 9 Filtered tailings stack construction (May, 2023)

Monitoring and surveillance LaRonde site personnel were responsible for the monitoring and the surveillance activities. The management and the governance applied at the site allows the development of the in-house capabilities and favours the ownership of the team while reducing the risks related to the construction of the filtered tailings stack. The monitoring activities included: visual inspections documented in daily and weekly reports; test pits to confirm the bridgelift thickness; surveys to monitor the filtered tailings lift thicknesses; quality control of the placed filtered tailings; installation of geotechnical instruments as the construction of the bridgelift was progressing; reading and interpretation of the instruments documented in weekly reports; and regular site visits by the Designer and the Engineer of Record. Visual inspections were regularly carried out by LaRonde to ensure compliance to the compaction instructions given to the contractor. These inspections were meant to obtain a visual assessment of: the lift thickness; the number of compactor passes; the surface quality (removal of snow, ice, and remoulded filtered tailings layers); the slurry tailings behaviour during construction; and any signs of instability. Quality control of the filtered tailings included nucleo-densimeter testing and laboratory testing. During the beginning of the filtered tailings stack construction, samples were taken weekly or whenever a

421

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA change to the LAR:LZ5 tailings ratio was notified by the filtration plant or mill operation teams. This frequent testing was important to confirm the properties of the filtered tailings. The testing included: particle size distribution, specific gravity (𝐺! ), water content, and standard Proctor. Forty-six filtered tailings samples were taken between October and December 2022. Table 1 summarizes the results of 𝐹𝐶, 𝐺! , and standard Proctor. The observations are consistent with the design hypotheses and are generally as follows: • LZ5 tailings are usually finer than LAR tailings, with 87% < 𝐹𝐶 < 90%. • Specific gravity is generally lower for LZ5 tailings, with values ranging from 2.91 and 2.96. • The water content operational range of the filtration plant (19% ±3%) allows producing tailings that are close to the optimum water content to reach the targeted 95% of the SMDD. • LAR tailings have an SMDD ranging between 1,749 kg/m3 and 1,812 kg/m3, while LZ5 tailings have a lower SMDD, ranging between 1,619 and 1,655 kg/m3. Given the variability of the Proctor test results, an in-situ density based on the LAR tailings SMDD is targeted. When the targeted density is not achieved, a protocol is followed to explain the reasons why and additional samples are collected for testing and verification purposes. Table: Summary of laboratory testing on filtered tailings produced between October and December 2022 FC (%)

Specific gravity (-)

SMDD (kg/m3)

Optimum water content (%)

Number of tests

43

24

24

24

Min

78

2.91

1,619

17.5

Max

90

3.08

1,815

22.5

Average

84

3.02

1,723

19.6

Geotechnical instrumentation Geotechnical instrumentation within Extension A4 TSF included VWPs, push-in VWPs, INCs, and settlement plates. Monitoring data are collected to inform the deposition sequence strategy in the different blocks of the Z1 working area. To verify and inform design parameter selection, on March 1st, 2023, a large-scale loading test was carried out upstream of Dike 10 south at the location of INC-22-02. The following geotechnical monitoring instruments were installed at this location: 2 VWPs in the tailings, 1 VWP in the cohesive soils underneath the slurry tailings, and 1 VWP in the till. A 2.6 m high waste rock load was placed in a relatively short time (instantaneous load) to attempt to simulate undrained conditions and to monitor the generation and dissipation of the excess porewater pressures in the slurry tailings and the cohesive soils in the foundation. At the time of the loading test, 3.8 m of filtered tailings were already placed.

422

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE

Figure 10: Schematic of the large-scale field loading test at the location of INC-22-02 Figure 10 illustrates a schematic of the loading test with the VWP depths measured from the top of the bridgelift. Figure 11 illustrates the porewater pressure (PWP) variation caused by the loading test.

Figure 11: Porewater pressure variation caused by the large-scale loading test at the location of INC-22-02 PWP data in the slurry tailings, at the location of VW149655, indicate a sudden increase of 11.94 kPa followed by excess PWP dissipation. The time to reach 50% and 90% consolidation was 8 hours and 63 hours respectively. In the cohesive soils, at the location of VW149920, the time to reach 50% and 90%

423

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA consolidation was 7 hours and 90 hours respectively. The VWP in the till showed no change in porewater pressures due to the sudden loading increase. Table 2 shows the PWP coefficient (𝐵,) and PWP ratio (𝑅% ) calculated during the loading test. 𝐵, is defined as , = 𝐵

∆𝑢 ∆𝜎

where ∆𝑢 is the PWP increase due to the applied load and ∆𝜎 is the stress increase at each VWP depth. ∆𝜎 was estimated using Boussinesq equation. 𝑅% is defined as 𝑢 𝑅% = 𝜎& 3 ) and porewater pressure Table 1: Porewater pressure coefficient (𝑩 ratio (𝑹𝒖 ) during the large-scale loading test at the location of INC-22-02 VWP ID

VWP depth (m)

Soil

3 (-) 𝑩

𝑹𝒖 (-)

VW149665

7

Slurry tailings

0.45

0.24

VW149699

12

Slurry tailings

0.58

0.31

VW149920

15.5

Cohesive soils

0.66

0.31

VW149687

18.5

Till

N/A

N/A

where 𝑢 is the PWP reading following the loading test and 𝜎& is the total vertical stress at each VWP depth. 𝐵, in the cohesive soils is higher than in the slurry tailings. 𝐵, values are between 0.5 and 0.7, which was found to be encouraging as these values suggest partially drained conditions. 𝑅% values in the slurry tailings and the cohesive soils are close to 0.3. An 𝑅% threshold value of 0.7 is generally considered as an onset value for liquefaction (Sadrekarimi, 2014) in the tailings and the results are well below this value.

Closure LaRonde mining complex transitioned to filtered tailings deposition in October 2022 with a filtered tailings stack to be constructed over an existing slurry tailings TSF. This change in the tailings management strategy was mainly motivated by the reduction of the environmental footprint and the societal risks, and by the possibility to integrate progressive rehabilitation during the construction of the filtered stack. The transition to filtered tailings required the construction of a filtration plant and a new water management infrastructure. The placement of the filtered tailings stack required the construction of a bridgelift over the slurry tailings within Extension A4 TSF. Supported by the Designer and the Engineer of Record, LaRonde developed a monitoring and surveillance program for the construction of the bridgelift and the filtered tailings stack. Results from

424

TRANSITION TO FILTERED TAILINGS AT LARONDE GOLD MINE laboratory testing and readings from geotechnical instruments are used to inform and update the operations and the design. It is very difficult to convince the management to move forward with projects such as transition from slurry to filtered tailings because of the associated high capital costs and operation expenses. Therefore, to be able to convince the management of the feasibility of filtered tailings projects, it is important to include the closure and rehabilitation costs and demonstrate that they will be significantly reduced. During construction and operation, challenges were faced regarding the targeted density, the control of the tailing’s humidity at the plant, the generation of dust, the development of a procedure to place and compact the tailings, and the snow management. However, the transition is considered very positive as no upset conditions causing operation interruptions were encountered, very little material was out of specification, and the instrumentation readings indicate good geotechnical performance of the foundations, including the slurry tailings on which the bridgelift and filtered stockpile were constructed.

Acknowledgements The filtered tailings 2022 project was made possible with the valuable support of Agnico Eagle corporate management. Many LaRonde employees have contributed to the success of this project. The authors also acknowledge the efforts of Golder Montréal, WSP-LaRonde team, Galarneau Entrepreneur Général, Dubé, and Services Miniers Nord-Ouest.

References Plewes H.D., Davies, M.P. and Jefferies, M.G. 1992. CPT based screening procedure for evaluating liquefaction susceptibility. In Proceedings of the 45th Canadian Geotechnical Conference, Toronto. Robertson, P.K. 2016. Cone penetration test (CPT)-based soil behavior type (SBT) classification system – an update. Canadian Geotechnical Journal 53(12): 1910–1927. Sadrekarimi, A. 2014. Effect of the mode of shear on static liquefaction analysis. Journal of Geotechnical and Geoenvironmental Engineering 140(12): 4014069.

425

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

426

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

The Development of LKAB’s Tailings Facilities, from the Beginning and into the Future Sara Töyrä, LKAB, Sweden Dan Lundell, Tailings Consultant Scandinavia, Sweden Roger Knutsson, Tailings Consultant Scandinavia, Sweden Thomas Bohlin, Tailings Consultant Scandinavia, Sweden

Abstract LKAB is a mining company in northern Sweden with a history of mining since 1890. This case study describes the development of LKAB’s tailings management strategy at three mine sites. All three facilities were initially designed with influence from the hydro power sector, and acted as artificial lakes with retaining embankments. Through different methods, the three facilities have been transformed with the goal of significantly reducing the amount of free water in the facilities. The surrounding embankments, originally designed as low permeable structures, have now been altered into structures with higher permeability. To reduce credible failure modes to a minimum, there are detailed plans on how to further change the water management and how to integrate waste rock placement with embankment construction. These plans include thickening to reduce incoming water, as well as altering the location of the free water in the ponds to obtain longer distances between free water and embankments. The integrated waste rock placement includes transforming older embankments, some constructed using upstream raises, through downstream placement of large amounts of draining support fill. These changes are carried out at a time when there is a focus on safety, environmental impact, and efficient operations, and the tailings management strategy at each site takes these aspects into count. This case study will also give insight into how changes in this area have been acknowledged during the 50 years of operation of the tailings facilities and during the development of the strategies. In relation to this case study, the concept dam is also discussed: how to alter the conditions of a dam to convert credible failure modes into non-credible failure modes? And what is the definition of a dam – when is a dam no longer a dam?

Introduction This case study aims to show the development path of transforming tailings deposition in an artificial lake, built with conventional impermeable embankments, into facilities that no longer fit the traditional idea of a dam.

427

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA LKAB was formed in 1890 and has been in operation at the Kiruna and Malmberget sites since then. The use of tailings facilities started in the middle of the 20th century, and the current tailings facilities have been in operation since 1962 (Svappavaara), 1977 (Kiruna), and 1982 (Malmberget). The starting point is three facilities, which have a partially common history with a legacy from the hydro power sector (Svenska Kraftnät, 2023), but have since diverged. The evolution of each facility and the planned future actions, including how the strategy is to be adjusted to meet current expectations, will be examined. Additionally, the potential of further measures that could transform these facilities into non-dam structures will be explored. The practical implications of the term “dam,” including the requirements and expectations associated with it, as well as the consequences of redefining these facilities, will also be discussed.

Figure 1: LKAB’s three mines in Kiruna, Svappavaara, and Malmberget in northern Sweden and the two harbours in Narvik, Norway and Luleå, Sweden

The LKAB operations sites LKAB is Europe’s leading mining and mineral company that specializes in producing refined iron ore products for steel production. The company is mainly based in northern Sweden, but is also present in 12 other countries around the world. LKAB employs approximately 4,300 people, with around 2,000 based in Kiruna, in northern Sweden. LKABs operations include iron ore mines, processing plants, and ports located in northern Sweden and Norway. The company's iron ore is extracted from both underground mines in Kiruna and Malmberget, as well as from open-pit mines in Svappavaara. After extraction, the ore is enriched and refined into pellets at processing plants in Kiruna, Svappavaara, and Malmberget. The finished products are then transported by rail to the ports located in Luleå and Narvik before being shipped out to customers around the world. An overview of LKABs tailings facilities are shown in Figure 2.

428

THE DEVELOPMENT OF LKAB’S TAILINGS FACILITIES, FROM THE BEGINNING AND INTO THE FUTURE

Figure 2: LKAB’s tailings facilities. From left to right, Kiruna, Svappavaara and Malmberget. All of them consist of a tailings pond and a clarification pond

The Kiruna site LKAB mines iron ore from the mountain of Kiirunavaara (the name means ptarmigan mountain). The mine started as an open pit in the 1890s. During the 1960s it was transformed into an underground mine. Today, the mine is a large and modern underground mine for iron ore extraction. The ore body in Kiirunavaara is coherent with a dip of 60 degrees and around 4 km long. The known depth of the deposit is at least 2,000 m and on average about 80 m wide. The deposit consists of magnetite with an average iron content of 63%. The deposit is mined through large-scale sub-level caving. For tailings management, there is a facility consisting primarily of a tailings pond, a clarification pond, and equipment for tailings disposal. Tailings are deposited in the northern parts of the tailings pond from slurry pipes with numerous spigots to build up a beach, which causes the tailings surface to slope from north to south. As a result, the embankments in the north are raised more frequently than in the south. The clarification pond is located south of the tailings pond, as illustrated in Figure 2. Difficulties with slurry pumping and water circulations during winter have posed major challenges (Töyrä et al., 2017).

The Svappavaara site In Svappavaara, LKAB currently operates two open-pit mines, Leveäniemi and Gruvberget. Leveäniemi was operational from 1965 to 1983 and from 2015 to the present. Gruvberget was operational from 2010 to 2018. LKAB also operates facilities for enrichment and pelletizing. Ore processing in Svappavaara dates to the mid-1960s and involves processing ore from various sources within the company. The processing plant was in operation and the tailings facility was active during the time when the mining operations at the site were on hold.

429

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The first embankment in the tailings management system was constructed in the mid-1960s to create a clarification pond. In 1974, a barrier embankment was constructed to create a separate tailings pond and the clarification pond. In 2012, a transition towards thickened deposition was initiated, with the result that the solids content in the tailings has been increased from about 5 to 10% to about 70%. The transition encountered several difficulties with adapting an existing tailings storage facility to become suitable for deposition with thickened tailings, and took several years to complete (Töyrä et al., 2018). That resulted in a smaller amount of water in the facility, and the tailings can be deposited in the form of a cone. The facility is illustrated in Figure 2.

The Malmberget site The Malmberget mine comprises around twenty ore bodies that consist mostly of magnetite ore, with some hematite ore also present. The ore processing plants are located above ground in Vitåfors, where hoisting, crushing, enrichment, and pelletizing take place. The tailings are deposited in a tailings pond located close to the plant. Deposition of tailings is carried out through spigotting along the crests of the embankments or via a single-point discharge. Spigotting creates a beach close to the embankments, which is beneficial for future embankment raises. Water is led to the clarification pond and returned to the process. Excess water is discharged to the Lina River, a tributary of the Kalix River, around 90 km downstream. The facility is illustrated in Figure 2.

Common closure requirements The closure requirements for all three sites are quite similar and straightforward. Most of the tailings pond area will be covered with a simple cover and then vegetated. (In areas near single-point discharge where the accumulation of coarse fraction with higher sulfide levels may occur, a more complex cover may be required.) The embankment slopes will be flattened and covered with a simple cover. The clarification ponds will be emptied, and the embankments will be partially or completely removed, when their function is no longer needed.

Tailings management – historic challenges The operation of all three facilities began with the construction of artificial lakes with impermeable embankments and deposition from a single discharge point. The inspiration was drawn from the hydropower industry (Svenska Kraftnät, 2023), which is a large sector in Sweden, while there are few tailings facilities within the mining sector. The approach resulted in impoundments with a large free water volume and thus, a requirement for impermeable embankments. The embankments were often constructed

430

THE DEVELOPMENT OF LKAB’S TAILINGS FACILITIES, FROM THE BEGINNING AND INTO THE FUTURE as conventional hydropower dams, with a standing impermeable core combined with downstream filters, and a support fill of waste rock, and were raised with a centerline or downstream approach. This concept led to increasing costs over time as the embankments became higher with a broader base, coupled with an increase in the rate of depositing – that is, the raises were more expensive and more of them were required. Additionally, the requirements for filter material gradually got stricter, and the filters that were constructed initially, based on then available crushed products, needed to be replaced with specially designed filters. Consequently, the construction costs of the embankments increased significantly over time.

Different solutions at the three sites To achieve cost-effective operation as tailings volumes increased, different concepts were introduced at the three different sites.

Figure 3: 2001: Single point discharge with impermeable water retaining embankments. 2010: Additional discharge point to distribute tailings flow more evenly. 2014: Spigotting from the north embankment to create a sloping tailings surface from the north. 2018: Additional spigotting from the north, only the south embankments are kept as impermeable structures At the Kiruna site deposition was concentrated to the northern side of the facility, utilizing the slope of the tailings to accumulate water in the southern part. This created a situation where the embankments in the north needed to be raised more frequently but could be constructed as permeable. Meanwhile, only the embankments in the south need to be impermeable (the impermeable structures are more expensive to raise)

431

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA and these embankments need to be raised to a lesser extent. An example of the evolution of the Kiruna tailings facility is shown in Figure 3. At the Svappavaara site the elevated hill east of the tailings pond inspired the creation of a tailings cone with a high point. Thickened deposition was implemented to achieve deposition from the high point, with a steep slope to achieve large volumes with a limited need for embankment raises. The embankments were originally built as impermeable structures, but then changed into permeable structures. At the Malmberget site upstream raises were implemented (on top of the impermeable embankments); to limit the cost of construction tailings were used as construction material for the raises. This involved constructing upstream raises on potentially contractive tailings, which led to a need for large buttresses. The properties of the tailings have been evaluated in both Kiruna and Malmberget (Engström et al., 2020) and a critical state line has been determined for Malmberget. The buttresses are designed to withstand fully liquefied tailings. The evolution of each site is illustrated in Table 1. Table 1: Evolution of the tailings facilities at each site Kiruna – Left: old strategy. Right: current strategy

Kiruna historic development. Left: Old strategy with a single point discharge with a low solids content and impermeable embankments (gray area) resulting in a flat surface and a large water surface. Right: Present strategy with spigots for tailings deposition, combined with inflow of water from a single discharge point. The spigotting has allowed the buildup of a sloping tailings surface, resulting in a significantly reduced water surface. Parts of the embankments have been transformed into permeable structures (white lines).

Svappavaara – Left: old strategy. Right: current strategy

Svappavaara historic development. Left: Old strategy with a single point discharge with a low solids content and impermeable embankments (gray area) resulting in a flat surface and a large water surface. Right: Present strategy with a thickened tailings deposition, with overflow water directed to the clarification pond, combined with parts of the embankments being permeable (white lines), resulting in a significantly reduced water surface.

432

THE DEVELOPMENT OF LKAB’S TAILINGS FACILITIES, FROM THE BEGINNING AND INTO THE FUTURE Malmberget – Left: old strategy. Right: current strategy

Malmberget historic development. Left: Old strategy with a single point discharge with a low solids content and impermeable embankments (gray area) resulting in a flat surface and a large water surface. Right: Present strategy with spigotting from the embankments, allowing formation of beaches, resulting in a significantly reduced water surface, with parts of the embankments being permeable (white lines), resulting in a significantly reduced water surface. Most of the embankments have been transformed into permeable structures (white lines).

Similarities in the development Although the development of the three sites has been different, several similarities exist among them. Firstly, all three have implemented measures to reduce the cost of embankment raises, albeit in different forms. Secondly, they have reduced the amount of free water in the tailings facilities. Thirdly, each facility has faced the challenge of meeting Swedish guidelines for an instantaneous discharge capacity of emergency floods at the retention level. Furthermore, at all three sites, great advantages of coordinating waste rock placement with embankment construction have been identified. Tailings and waste rock management are still planned separately, leading to separate waste rock dumps and unnecessary, costly embankment construction. Consequently, the embankments were not initially prepared for closure, and steep slopes were utilized during operation to minimize costs, with slope flattening instead being part of the closure plan. Emphasizing these common trends into a concept moving forward would include: •

Continue the separation of water management, with water mainly located in the clarification ponds to minimize water in the tailings ponds.

•

Minimize the amount of free water present near the embankments.

•

Create sufficient freeboard to be able to challenge the requirement for specific discharge capacity, which tends to lead to complex spillway structures requiring deep water close to the perimeter of the facility.

•

Start coordinating construction of embankments with waste rock deposition, to be able to construct high embankments with flat slopes in a cost-effective manner.

433

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Future strategies Based on the introduced concepts, derived from earlier trends, strategies have been developed for the three sites.

The Kiruna site The amount of incoming water will be further reduced by transitioning from the current deposition by spigotting, to thickened tailings disposal (see Figure 4). This will be combined with downstream raises using waste rock for all embankments. For the previously impermeable embankments, two options will be evaluated. The first is to maintain the embankments in the south as impermeable. However, a more attractive alternative, given the intended principles, would be to allow the southern side to consist of permeable embankments and lead the seepage to the downstream clarification pond. This latter option would result in minimal amounts of water in the tailings pond. Extreme floods will be managed through an emergency spillway leading directly to the clarification pond. The future strategy for Kiruna is illustrated in Figure 4.

Figure 4: Left: Current deposition with spigotting. Right: Transformation to a strategy where thickened tailings disposal is introduced with the intent that the overflow water is directed to the clarification pond. All embankments at the tailings pond are raised downstream using permeable (white line) waste rock, resulting in a tailings pond without free water surrounded by embankments supported by significant amounts of drained waste rock

The Svappavaara site The concept of thickened tailings disposal has already been implemented. This will be combined with downstream raises using waste rock for all embankments, and where suitable, the embankments will be transformed into filling the function of a waste rock dump, as illustrated on the left of Figure 5. For the impermeable embankments at the tailings pond in Svappavaara, two options will be evaluated, similarly to the Kiruna site. These two options are: maintain the embankments in the south as impermeable, or allow the southern side to consist of permeable embankments and lead the seepage to the downstream clarification pond. This latter option would result in minimal amounts of water in the tailings pond.

434

THE DEVELOPMENT OF LKAB’S TAILINGS FACILITIES, FROM THE BEGINNING AND INTO THE FUTURE Similarly, to the Kiruna site, extreme floods will be managed through an emergency spillway leading directly to the clarification pond. The future strategy for Svappavaara is illustrated on the right of Figure 5.

Figure 5: Left: Illustration of the opportunity to add waste rock to an embankment, giving the function of both support and deposit. Right: Svappavaara future development, where all embankments at the tailings pond will be permeable (white line), raised downstream using waste rock, resulting in a pond without free water surface surrounded by embankments and supported by significant amounts of permeable waste rock

The Malmberget site The future concept for the Malmberget site involves abandoning the upstream method for raising the embankments and instead using the downstream method, as illustrated on the left of Figure 6. This means moving away from future embankment raises on potentially contractive tailings, which eventually will eliminate concerns regarding the impact of liquefiable tailings on embankment stability. The transition to downstream raises will be combined with moving the free water surface to the side of the facility, which is limited by a substantial waste rock dump. This approach allows for the construction of permeable embankments, using significant amounts of drained waste rock as support, while keeping the water level at a distance, resulting in low phreatic levels near the embankments. Extreme floods will be managed through a large freeboard, which is possible to obtain given the distance from free water surface and the embankments, combined with pumping to restore the water level to the normal operating level after floods. The future strategy for Malmberget is illustrated on the right of Figure 6.

435

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 6: Left: Illustration of embankment cross-section and different stages of transforming upstream raises to downstream raises. Right: Transformation to a future strategy with all embankments at the tailings pond being permeable (white line), raised downstream using waste rock, resulting in a pond with no free water close to the embankments, and where the embankments are supported by significant amounts waste rock. (The illustration includes a planned expansion of the clarification pond.)

Evaluation of common failure modes Evaluating these suggested concepts in relation to common failure modes for tailings facilities leads to the following conclusions. Overtopping: The minimal amount of water within the facilities reduces the risk of overtopping. Additionally, cost-effective embankments raised by the deposition of waste rock enable the construction of embankments with significant freeboard, which greatly limits the possibility of overtopping. Internal erosion: The minimal amount of free water in the tailings pond, combined with large distances from free water to embankments, greatly reduces the risk of internal erosion for most of the embankment sections. The exceptions that require special consideration are the areas where significant water depth may occur during high flows. Stability: Moving away from upstream and centerline raises, and instead utilizing downstream raises, leads to embankments with significant amounts of drained support fill, which is generally favourable for stability. This also means moving away from raises on potentially contractive tailings. This will eliminate the risk that liquefaction of tailings alone could result in an embankment failure. However, if an embankment failure should occur, parts of the tailings mass may still potentially liquefy and possibly contribute to a significant outflow of tailings.

Are these embankments also “dams”? Throughout the text the compounding structures of the tailings pond have been referred to as embankments and not “dams” – is there a difference? Possibly not, but in Sweden the term “dam” has a legal status. Similarly, for all countries in Europe, a definition to categorize these structures will be necessary, in relation

436

THE DEVELOPMENT OF LKAB’S TAILINGS FACILITIES, FROM THE BEGINNING AND INTO THE FUTURE to ongoing work with including dams in the European design code (Eurocode). When IEG (the Swedish national implementation commission for Eurocode 7) put together a summary of definitions, it became obvious that: •

The primary objective of a “dam” is usually to prevent the flow of free water. It has been illustrated that while solutions to historic challenges may have been different, they still

share similarities. Through these similarities, a concept for continued development has been suggested and developed into site specific strategies. The suggested strategies will reduce risks in relation to common failure modes and transform the facilities even further away from the conventional concept of a dam. •

The current embankments are rarely, if ever, in contact with free water and are usually only exposed to a low groundwater table. This case is true not only for LKAB’s facilities but also a desired trend for many tailings facilities.

However, even though liquefaction of tailings may not itself result in an embankment failure for the downstream raised embankments, a failure of an embankment could in some cases trigger liquefaction in parts of the tailings, resulting in a major tailings outflow.

Conclusion A lesson has been learned from the historic choice, namely that it was not optimal to follow the hydropower industry and construct impermeable embankments for tailings storage. Bearing in mind that all engineers working with tailings have tried to follow what was known to be best practice at the time. However, this choice has led to: •

Unnecessary costs, mainly connected to strict construction requirements on impermeable zones and filter materials.

•

A limitation in expanding the storage facilities, due to limitations in drainage of the tailings and the possibility of raising this type of embankment. When put to the test, as production volumes increased, changes have been introduced to these

facilities. All LKAB sites, although following different paths, have converged to similar new strategies: •

Minimize water in the tailings ponds and minimize free water near the embankments.

•

Avoid complex spillway structures requiring deep water close to the perimeter of the facility.

•

Coordinate construction of embankments with waste rock deposition, to be able to construct high embankments, raised downstream with flat slopes, cost-effectively. In short, to move away from compounding free water. This also means moving away from what is

often seen as the main function of a “dam.” So, are these structures still “dams,” or are they better considered

437

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA as something else – potentially an integrated waste landform? Is it the primary function of the structure that will determine if it is a “dam,” or is it instead to be determined by the potential consequences of a failure (for example whether there is a risk for a significant flood of liquefied tailings)? The exact answer to these questions follows from the definition of “dams.” In a Swedish context the definitions are still changing and leave room for discussion. However, moving forward with developing new Swedish guidelines and extending Eurocode to include dams, one important conclusion could be drawn from this case study: if in the first place, these structures had not fallen into the category of “dams,” the benefits of separating tailings from water might have been acknowledged earlier, in turn reducing both risks and costs. Going forward, developing new guidelines together with the hydropower industry – both on a Swedish and a European level – how should we avoid repeating the same mistake? Are these structures “dams” – and if they necessarily are, what requirements may be accepted, so as not to risk ending up with a definition that limits the development of these facilities?

References Engström, K., Töyrä, S., Danilov, S. and Knutsson, R. 2020. Variability in sand characteristics — a case study at LKAB Mine Tailings Facility in Sweden. In Proceedings of Tailings and Mine Waste 2020. Svenska Kraftnät. 2023. Book on Dams. The Swedish Experience. Developed by Svenska Kraftnät (with support of SwedCOLD, prepared for ICOLD annual meeting 2023). Töyrä, S., D. Lundell and A. Bjelkevik, 2017. Tailing management at the Kirunavaara iron ore mine. In Proceedings of the Twenty-first International Conference on Tailings and Mine Waste Proceedings 2017. Töyrä, S., P. Marthin, K. Jokinen and D. Lundell. 2018. Adjustments to tailings deposition with thickened tailings in Svappavaara. In Proceedings of the 22nd International Conference on Tailings and Mine Waste 2018.

438

Proceedings of Tailings and Mine Waste 2023 November 5-8, 2023, Vancouver, Canada

Evolution of Tailings Storage at the Campbell Mine over the Past 40 Years Desiree Wilkins, Evolution Mining – Red Lake Operation, Canada Tara Rothrock, WSP E&I Canada, Canada

Abstract The Campbell mine is located in Northwestern Ontario, Canada, and has been continuously producing gold since the late 1940s. In 1983, a conceptual plan and initial engineering design was prepared for a new tailings facility to service the operation for the expected 20-year remaining life of mine. The facility originally consisted of four dams with five planned staged raises to be constructed between 1983 and 1999. It was designed to accepted engineering standards for the time, which evolved as the facility was raised. The facility is located adjacent to the community of Balmertown and Ontario Highway 125 and is currently referred to as the Campbell Main Tailings Pond (MTP). Initial construction of the MTP starter dykes began in 1983. During construction, instability issues required the main containment dam to be re-aligned. The facility is located in an area with challenging foundation conditions, with a relatively deep deposit of varved glaciolacustrine clay, and zones of potentially liquefiable tailings underlying much of the MTP. These two units controlled the stability of the structures in previous designs. Almost 40 years later, the MTP continues to be used for storage of tailings produced from the Campbell Mill. A slow rate of rise, stabilizing toe berms, and wide upstream beaches were used to manage stability concerns. Over the years, as staged construction of the facility progressed, the design was updated based on: evolving geotechnical standards and environmental regulations, changing mine plans, additional geotechnical investigation information, surveillance, and monitoring results. Construction of the final raise, Stage 7, was completed in 2020. The MTP is now contained by 5 dams, with a maximum dam height of 14 m. The facility is in the final infilling stage, with closure planning advancing. Based on current mine plans, approximately 2 years of solids storage space remains in the MTP. Numerous operational changes were faced over the 40-year life of the facility, including: ownership changes, revised tailings deposition and water management practices, expanded and enhanced monitoring and surveillance, and evolving governance. Challenging geotechnical conditions and 40 years of mine changes required diligent design and operation on the part of the owners and designers. On-going monitoring is a key component of performance

439

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA assessment. Surveillance includes a remotely monitored piezometer network, slope inclinometers, InSAR, annual ground and bathymetry surveys, and a visual inspection program. A long-term collaborative and cooperative relationship was developed between the owner and the geotechnical Designer of Record (DOR)/Engineer of Record (EOR). The same DOR/EOR has provided technical direction for dam safety and design of the facilities over the last 20+ years. An independent technical review board (ITRB) was established in 2019 and meets annually, completing independent review of all designs. Critical controls for the facility have been developed and robust verification processes established. Overall, the MTP has performed well historically, and Evolution Mining is committed to ensuring this continues throughout the life cycle of the facility.

Introduction The Campbell Mine is an underground gold mine, located in Balmertown, Ontario, Canada, about 150 km northeast of the city of Kenora and 6 km east of the town of Red Lake. Since 2020 the mine has been owned by Evolution Mining Limited and is a key part of the Red Lake Operation (RLO). The Campbell Mine commenced production in 1949 and has undergone various modifications and upgrades since. Mill throughput was initially 330 tonnes/day with gold recovery achieved through cyanide leaching and amalgamation of the free gold. A roasting circuit was added in 1951 to assist with gold recovery from high sulphide content refractory ore. Several processing improvements were made and the milling rate was increased many times through to 1990, when it reached 1,200 tonnes/day. In 1991 an autoclave was commissioned to replace the roaster, significantly changing the chemistry of the produced tailings and effluents. Daily mill throughput gradually increased through to 1999, when it peaked at 1,700 tonnes/day. Lower tonnages were milled from then to 2020 to match mine production. Today the mill is permitted to process an average of 2,200 tonnes/day, and the current mine life has been extended to 2042. Tailings deposition planning, environmental regulations, and dam safety guidelines have evolved since the start of the mine, requiring changes in how tailings and water were managed. The tailings and water management systems were upgraded over the years in order to satisfy successively stricter environmental regulations, particularly with regard to effluent discharge quality. Historically, mill tailings were discharged to various locations, initially with no dam structures in place. From 1968 to 1983, tailings were deposited in the Balmer Tailings Area, a basin on the southern shoreline of Balmer Lake. From 1983 through the present, Campbell Mill tailings (minus tailings used for backfill for the underground mine) have been deposited in what is now known as the Main Tailings Pond (MTP). The MTP has an approximate area of 1 km2, bound by dams with an overall length of about 4,300 m, and currently stores approximately 8 M tonnes of tailings that have been classified as non-acid generating. Aside from the MTP, there are several facility components to manage contact and non-contact water

440

EVOLUTION OF TAILINGS STORAGE AT THE CAMPBELL MINE OVER THE PAST 40 YEARS from the site. Diversion ditches are in place along the north sides of the various facilities to route noncontact water to Balmer Lake. The Primary Clearwater Pond (PCWP), and South Dam Collection Pond (SDCP) collect contact surface water and seepage around the perimeter of the MTP. The Polishing Pond and constructed Wetlands, formed from the old Secondary Clearwater Pond (SWCP), are used for final polishing of treated effluent and contact water. Final discharge is released to Balmer Lake from the outlet of the Wetlands (WETOUT), and represents the point of compliance for the Campbell Mine.

Figure 1: Campbell mine site plan

Overview of design/foundation The MTP is formed almost entirely by dam structures (North, East, South, West, and Northwest Dams). High ground exists at the southeast end of the facility where the East and North Dams tie into natural

441

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA ground. The surficial geology of the MTP facility site was well established from subsurface investigations conducted in the past three decades with over 100 boreholes and 100 cone penetration tests completed. The MTP is underlain by a sequence of Holocene, Glaciolacustrine and Glacial deposits founded on preCambrian bedrock. The foundation soils generally consist of Glaciolacustrine silty varved clay overlying glacial till and bedrock. In the lower lying areas, peat overlies the clay deposit and the clay deposit is underlain by discontinuous glaciofluvial silt and sand. Historical tailings overlay peat under portions of the South and North Dams. The peat is fibrous and compressed under the weight of the tailings and dam fills, creating a firm to compact layer. The glaciolacustrine and glaciofluvial deposits thin out and disappear toward higher elevations. The two main stability design concerns for the MTP containment dams have been the foundations that are dominated by glaciolacustrine clay deposits, and the tailings deposit that has potentially liquefiable zones. Stability concerns associated with the design have been mitigated by staged raising of the dam crest in small increments, construction of stabilizing toe berms, where needed, construction of wide upstream tailings beaches, and diligent performance monitoring. Where situated on normally consolidated clay and/or potentially liquefiable tailings, overall downstream slopes of the dam range from about 16H:1V to 20H:1V. For dam sections situated on weathered/desiccated clay, downstream slopes range from about 7.5H:1V to 10H:1V. The overall maximum height of the dams ranges from about 3 m to 14 m.

Figure 2: Cross section of critical section of the North Dam

Glaciolacustrine clays Foundations for the MTP structures are dominated by glaciolacustrine clay deposits. The glaciolacustrine clay is medium to high plastic and low to medium sensitivity. In its natural state, the upper portion of the clay unit is weathered and desiccated, with over consolidation ratios (OCR) in range of about 1.5 to 4.5. The bottom portion of the layer is normally consolidated and soft to very soft with depth. The depth and

442

EVOLUTION OF TAILINGS STORAGE AT THE CAMPBELL MINE OVER THE PAST 40 YEARS extent of the glaciolacustrine clay in each of the dam footprints were delineated in past site investigations. The critical realigned section of the North Dam is underlain by a 5 to 7 m thick layer of normally consolidated clay. At the West, East and Northwest Dams, there is no normally consolidated clay and the desiccated/weathered zone is relatively thin (1 to 3 m). In the main section of the South Dam, there is a thick weathered clay crust (about 2 to 4 m) with a relatively thin deposit (0.5 to 1 m) of normally consolidated clay. In past raise designs, considerable effort was made to characterize the glaciolacustrine foundation clays. Cone penetration testing (CPT) is an essential tool to delineate the weathered/desiccated and normally consolidated zones within the unit. Hydraulic piston samplers are used to obtain “undisturbed” samples of the clay for advanced lab testing. Strength relationships were established considering CPT, in-situ vane testing, laboratory strength tests, and empirical correlations (i.e., index testing). The site-specific strength relationships were compared with well-established relationships in literature. A geotechnical database was developed with the results from the various site investigations completed at the site, and regularly updated as additional investigations become available, which has proven invaluable for subsequent raise designs for other facilities on site. Over the years, the design approach transitioned from using only total stresses to considering undrained strength ratios for strengths of the foundation clays. Utilizing undrained strength ratios allows for estimates of strength gains in response to increases in the effective stresses over time as the clay consolidates and the excess pore pressures imposed by the dam fill, dissipate. In this respect, the undrained shear strength of the clay at a given time during and after construction of each raise can be determined based on measured pore pressures from an appropriate suite of piezometers. The collected data indicate a peak undrained strength ratio (ratio of undrained strength, su, to vertical effective stress, σv’) of 0.22 is appropriate for the normally consolidated clay zones. In areas where previous failures occurred (discussed later), the clay was deemed to have been highly sheared to its remolded state, and an undrained shear strength ratio of 0.05 was considered appropriate, based on strength values determined from historical lab testing and CPT. Peak and post-peak undrained strength ratios for the weathered/desiccated clays range from about 0.3 to 0.5 and 0.2 to 0.45, respectively (for the stress range under consideration).

Foundation tailings Prior to construction of the MTP Dams, the mine discharged tailings from a location near the current South Dam. Consequently, tailings underlie the main sections of the South and North Dams. Due to high ground on the west side of the impoundment, the tailings delta did not extend to the west or east and tailings are not present in the foundations of the Northwest, West, or East Dams. The tailings consist of mainly siltsized particles with some sand-sized and a trace of clay-sized particles (i.e., rock flour).

443

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Zones within the foundation tailings and upstream tailings deposits at the MTP are potentially liquefiable. There were discrete interbedded dilatant layers of tailings within the contractant layers observed. This is common within tailings deposits where moving slurry discharge points create layered deposits with varying sizes of material. Static liquefaction can be triggered by additional loading from increasing dam heights, increases in phreatic surface under constant shear stress, or by stress unloading resulting from foundation movements or excavations at the dam toe. Seismic liquefaction is triggered by the cyclic shear stresses induced by ground motion. In historical designs for the MTP, the risk of seismic or static liquefaction were considered low given the low seismicity of the site and operational conditions at the facility (low rate of rise of the dams:1 to 1.5 m every three to four years, controlled construction, long tailings beaches, and buttresses to manage stability concerns in the foundation clays). From design review in 2016, it was no longer considered “best practice” by the designer or the owner to consider the factor of safety against triggering as the primary risk management tool given the uncertainties around quantification of the in-situ response of loose tailings to static and dynamic loading, and given that a dam failure due to liquefaction can occur rapidly without advance warning. For subsequent design work, liquefiable tailings were assumed to have liquefied and assigned liquefied shear strengths in the stability analyses. For the majority of the tailings deposit, a reasonable lower bound liquefied undrained shear strength ratio (Su(LIQ)/sv’) of 0.05 was considered reasonable, based on standard CPT screening assessments.

History of the MTP Construction of the MTP began in 1983. The initial design included the original North Dam (now known as the Primary Clearwater Dam), South Dam, West Dam, and Secondary Clearwater Dam. The South Dam was designed as the primary tailings solids storage structure for the facility, with an initial rockfill starter dam and raised with tailings in the upstream direction. The original North Dam, Secondary Clearwater Dam, and West Dam were designed as water-retention dams (with a low permeability core and downstream filter zone to maintain low downstream pore pressures). The dams were meant to be constructed in several stages over 15 years to El. 367.7 m to support the (then) mine life of about 15 to 20 years. It was recognized in the initial design that the North Dam was to be constructed over difficult foundation soils. The first stages of the original North Dam and the Secondary Clearwater Dam were constructed in 1983 to Elevation 360.6 m. The South Dam was constructed to elevation 363.3 m. Two water decant towers were also constructed to transfer water from the MTP to the SCWP and from the SCWP to Balmer Lake. In 1984, the West Dam starter dam was constructed and the original North Dam was raised to 361.5 m. Movement in the foundation of the original North Dam was observed 8 days into construction just as the final crest elevation was achieved. At the time the movement was attributed to a construction layout error that resulted in oversteepening of the slope. Repairs were made consisting of a waste rock stabilizing

444

EVOLUTION OF TAILINGS STORAGE AT THE CAMPBELL MINE OVER THE PAST 40 YEARS berm constructed on top of filter fabric at the downstream toe. In 1986 and 1987, attempts were made to raise the original North Dam, but each time, the dam slumped after completion. The clay foundation soils were recognized as being very weak. In 1987, after the third slump, options were assessed and decision made to abandon the original North Dam alignment and construct a new North Dam further upstream to the south as it was hoped that foundation conditions would be improved. The design included excavation of a wide shear key through the tailings foundation present at the east end, to mitigate the potential for a failure due to liquefaction of the tailings foundation soils. A downstream waste rock stability berm was also included in the design. A third decant structure was added, near the middle of the North Dam to transfer water from the MTP to the PCWP. In 1989 the South and West Dams were raised to elevation 365.8 m. Construction of the realigned North Dam first stage started in 1988 and was completed in 1990 to elevation 362.8. To take advantage of more favourable foundation conditions to the west, the North Dam alignment was shifted during construction to form its current “S” shape. Tailings were discharged from the West or South Dams forming long beaches in front of these structures and pushing the water pond against the North Dam. Water drained through the decant at the North Dam into the PCWP, followed by discharge through Decant 2 into the SCWP, and Decant 3 into Balmer Lake. In 1991, cracking and heaving were observed along the downstream stability berm of the realigned North Dam. The Geotechnical design consultant was changed, the tailings management plan was updated, and existing stability of structures reviewed. Due to operational requirements, the North Dam was raised to 365.5 in 1992. The raise was designed as a zoned earthfill embankment with the intent to retain water and tailings. Considerable work was completed in 1992 and 1993 to further characterize the glaciolacustrine clays in the dam foundations. Based on the site investigations and engineering evaluations, it was decided to develop wide spigotted tailings beaches upstream of the dams, which pushed the free water pond further away from the dams, resulting in improved stability by lowering pore pressures within the downstream shell of the dam. The beaches would allow for future centerline and upstream raises. The raise designs did not consider liquefied strengths in the tailings, as it was determined that the risk of liquefaction was low, given the low seismicity of the site, low height of the structures, and proposed operational conditions at the facility (slow construction, long tailings beaches and buttresses). The 1994 raise (subsequently referred to as Stage 1) consisted of a crest raise to 367.3 m for all three of the MTP structures. The North Dam construction included a downstream stabilizing berm. In 1997 the geotechnical consultant developed a conceptual design of the facility to end of mine life, which at that time was planned to be 2016, and involved 4 additional raises to maximum elevation of 372.8 m. In 1997 the North, South, and West Dams were raised to 368.2 m (Stage 2). The North Dam was raised using the downstream method, the South Dam was raised using the upstream method and the West Dam

445

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA was raised using centerline construction. Ground improvement measures were undertaken in 1999 for the normally consolidated clays at the North Dam, to allow the dam to be raised on the schedule required by the mine plan. Wick drains were installed in 1999 over a portion of the North Dam footprint with the thickest soft clay layer. A large downstream waste rock buttress was constructed between 1999 and 2002. In 2000 the Stage 3 lift was completed, with the North, South, and West Dams raised to 369.4 m. A centerline raise was completed for the North Dam as the tailings beach was sufficiently developed for this to be done. This allowed for less material to be placed in the downstream area which from a stability perspective, was deemed better for the soft soils. Upstream raises were completed for the South and West Dams. In 2002 the geotechnical consulting firm was changed to AMEC Earth and Environmental (now known as WSP E&I Canada), as the design engineers working on the facility had moved to this company. The current Designer of Record (DOR)/Engineer of Record (EOR) for the facility was a part of the original AMEC team and has worked continuously on the facility since 2002. From 2003 to 2005 the majority of tailings solids were being placed underground as Paste Backfill. As a result, there were insufficient tailings beach widths upstream of the West Dam. In 2003, RLO developed a method of controlled beaching to accelerate tailings beach formation next to the dams. Waste rock “training berms” were constructed to create cells, which the tailings slurry was then deposited into. The waste rock was sufficiently permeable to allow water to pass through to the centre pond area, while still retaining the tailings solids. This allowed the beach to build up near the dams faster, allowed the beach to drain more quickly, and helped to mitigate dust generation from the beach. To further retard dust generation, a thin layer of waste rock was pushed over the completed cells during the winter months. RLO still uses this method of beaching. Subsequent raises to the facility were completed in 2005 (Stage 4 to El. 370.6), 2009 (Stage 5 to El. 371.6), and in 2012 (Stage 6 to El. 373.3 m). In 2012, the South Dam was extended towards the high ground near the mill facility. This extension is now referred to as the East Dam. Because the MTP had expanded almost to the Diversion ditch, the Northwest Dam was constructed for containment along the northwest side of the facility (between the North and West Dams). The Northwest Dam was constructed with a low permeability core and downstream filters to act as a water retaining structure for the water pond as the mine moved towards closure. In 2016 an evaluation of stability of the MTP structures following updated CDA guidelines and industry practice was completed, and a design report for upgrades subsequently issued in 2017. As discussed in the section above, the updated design methodology was to assume triggering of tailings can occur, assign undrained liquefied strength consistent with current practice, and to then determine the resulting overall factor of safety of the dam. To meet updated stability requirements, the overall downstream

446

EVOLUTION OF TAILINGS STORAGE AT THE CAMPBELL MINE OVER THE PAST 40 YEARS slopes were flattened by constructing large downstream waste rock buttresses at both the South and West Dams in 2018 to 2020. The most recent, and final lift (Stage 7) to elevation 375 m was constructed in 2020 to 2021. The raise was completed in the upstream direction, with a step-in onto the existing tailings beach, in order to flatten the overall slope and keep the toe buttresses required on mine owned property. The raise was constructed with waste rock, along existing interior berms on the tailings beach, previously constructed as training dykes for controlled beaching activities. A low permeability compacted tailings blanket was included on the upstream slope to limit seepage onto the existing downstream Stage 6 beach during active tailings discharge.

Operational practices All mine impacted effluents and Campbell mill tailings (minus solids not required for paste backfill) are discharged to the MTP year-round. Tailings slurry is typically discharged upstream of the North, East, South, and West Dams into cells formed by waste rock berms, as discussed in the previous section. Once the cells are full, tailings are spigotted off the inner most internal structures into the main basin area. This method of controlled beaching enables long tailings beaches to be developed next to the dams, to meet stability requirements. The Northwest Dam does not have an upstream beach as it is designed as a water retaining structure. Mill processes and explosives use in the mine result in the water in the MTP containing significantly elevated levels of ammonia and dissolved metals, namely copper, nickel, lead and zinc. A barge mounted pump located on the East side of the facility is used to remove excess water off the facility, typically from May – November, at an average rate of about 350 m3/hr. The water is pumped to a metal hydroxide precipitation circuit located within the Campbell Mill, known as the ET circuit. Treated water is then pumped to the Polishing Pond system (1 Mm3 capacity, commissioned in 1995) for further clarification and natural degradation of ammonia. Polishing Pond water is transferred to a 15 ha constructed wetland (initially built in 2006 and expanded over the years), typically from May to October, for treatment of ammonia and final polishing. The discharge from the wetland, WETOUT, forms the final compliance point for the site. The wetland system works extremely well to remove ammonia, however peak treatment is limited to the summer months, with negligible ammonia removal realized from November – April. Ammonia treatment is the bottleneck for the entire Campbell tailings and water management system and requires sufficient available water storage capacity to store all water inputs into the overall tailings system from late Fall to late Spring, termed “winter water”. Currently, much of this winter water is stored in the MTP.

447

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA In order to prevent dust migration off property, testing of multiple solutions over the years was completed and informed development of current practices. RLO determined the most effective dust control strategy was to deposit tailings into relatively small paddock style cells described previously (i.e. controlled beaching), filling one cell before moving on to the next. In the fall and early winter, when sufficient frost has developed in the ground, a bulldozer is used to track pack accumulated snow to keep it in place, when conditions allow. When the frost depth increases, typically by January, a thin layer of mine waste rock is spread over the tailings surface. Over the past few years RLO has hired a contractor to supply and spray a dust suppression product on exposed tailings beaches in early October to bridge the time period before the snow cover is established. In 2023, it is planned to trial the use of drones to apply the dust suppressant. During the winter months, spigotting, rather than end dumping, is typically used to maintain more wet tailings beach area to help prevent dust generation off the freshly placed beach. Performance monitoring at the Campbell TSF consists of a comprehensive surveillance program as well as geotechnical instrumentation. The current surveillance and geotechnical instrumentation monitoring program was developed for the MTP in 1997. It provides guidelines for conducting visual inspections of the containment structures and for collecting, interpreting, and responding to data from geotechnical instrumentation within the various structures at the facility. In 2020 all MTP vibrating wire piezometers were connected to data loggers that transmit data daily to three hub locations. The hubs then transmit the data to a cloud where it can be remotely retrieved. One of the South Dam inclinometers has also been connected to the system using an in-place inclinometer. An online platform visually displays the data and provides emailed alerts. The data is available to site personnel as well as the EOR. A large portion of the instrumentation is focused on monitoring the pore pressures in the foundation clays. Piezometers in the glaciolacustrine clays have historically responded to construction loading as well as to tailings deposition/beaching activities. Data collected indicated that pore pressure response in the weathered/desiccated layer tends to be more muted than the normally consolidated clays, where the instantaneous B-bar response (for limit equilibrium stability calculations, the B-bar is defined as ratio of the change in measured pore pressure to the change in total vertical stress due to loading) to construction loading are typically about 1.0. In the thicker normally consolidated clay deposits, close to 100 percent dissipation of excess pore pressures takes several years. After the toe buttress at the North Dam was completed in 2002, excess pore pressures were not fully dissipated until about 2013. Inclinometers are installed along the MTP dams to monitor for potential movements in the foundations. Settlement is typically observed in the inclinometers, indicated by kinking in a sinusoidal pattern. Movement rates slowed to less than 5 mm per year in recent years (strain rates less than 0.05% per year). InSar monitoring during the snow free periods was implemented in 2021 across the entire operation. Monthly reports are provided as well as an online viewing platform for additional data analysis. Drone

448

EVOLUTION OF TAILINGS STORAGE AT THE CAMPBELL MINE OVER THE PAST 40 YEARS Lidar survey and pond bathymetry are collected annually to check for settlement and to develop an updated storage elevation curve for the MTP. Inputs and outputs to the facility are measured and used to update the water balance and to compare against the tailings management plan. Water levels in facility ponds are also measured often and are an input to the water balance. After 40 years of operation the MTP is entering the final infilling stage and detailed closure planning has commenced. To optimize the capacity of the facility during infilling, water management practices are being changed and include installation of a new pumping station on the Northwest side of the MTP equipped with a shallow water intake. This will lower the minimum pumping level and allow water to be moved off the facility year-round. The Polishing Pond, rather than the MTP, will be used for winter water storage, freeing up more space for solids storage (i.e. infilling) in the MTP. An emergency closure spillway will be constructed on the Northwest Dam in 2023, which will both reduce the risk of the facility and reduce the required flood water storage capacity. An infilling plan has been developed that pushes the water pond towards the northwest where the spillway will be located. The plan establishes the general grading of the facility required for surface water drainage on the final landform, while maximizing the amount of solids that the facility can store. An engineered cover design is currently underway with the goals of supporting development of a long-term stable landform while limiting contaminants to surface and groundwater pathways. The site as a whole is working to minimize fresh water consumption and increase paste backfill use, to further extend the life of the MTP. Once the MTP is full, Campbell Mill tailings will be deposited into RLO’s other active tailings facility, that will first be expanded and upgraded, design and permitting of which is currently underway.

Governance and risk reduction The same geotechnical DOR/EOR has provided technical direction for dam safety and design of the facilities over the last 20+ years. A long-term collaborative and cooperative relationship was developed between the owner and DOR/EOR. RLO’s Tailings Facility Engineer and the DOR/EOR typically communicate on a weekly basis. Any potential concerns or abnormalities at the site are immediately communicated to the DOR/EOR with advice given promptly, and DOR/EOR recommendations are addressed by the site in an appropriate timeframe. A high level of trust has been developed between the two parties, and this diligence has greatly contributed to the long-term safe operation of the facility. The first official Operations, Maintenance, and Surveillance Manual and Emergency Preparedness Plan was developed for the site in 1997 and is updated regularly. Trigger Action Response Plans (TARPs) are in place for all potential tailings facility issues. The site has its own Emergency Response Team and also works collaboratively with local emergency services.

449

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA An independent technical review board (ITRB) was established for the site in 2019 with annual meetings occurring since its inception. The first independent design review on the Campbell tailings facility was completed in 1997, with the first formal Dam Safety Review (DSR) conducted in 2014, with a subsequent DSR done in 2019, and the next scheduled for 2024. Formal dam safety inspections commenced in 1997 and have since been completed annually by the DOR/EOR. The site has adopted a priority ranking system for all tailings facility actions, which helps to prioritize completion of the actions. All actions are assigned owners and timelines for completion, and are tracked through the site’s risk management software. In 2019 the site developed critical controls and verifications for the facility focused around preventing loss of integrity of the facility. The critical controls have been updated and formalized under Evolution ownership, with critical control plans developed for each and verifications completed and tracked. Current Critical Controls include: Detailed Design and verification, Construction in accordance with design, Monitoring of TSF and associated response, and confirmation of operation in accordance with design. Evolution has developed a set of Sustainability Standards, including a Tailings Storage Facility Standard and a Water Management Standard. The company has also established a Tailings Governance Committee that meets quarterly and includes the Accountable Executive. Outcomes of the quarterly meetings are presented to the company board of directors. Over the years, the DOR/EOR and the site have worked together to minimize the overall risk of the facility. In addition to the above, the following risk mitigation measures have been implemented: developing a robust understanding of the foundation conditions and strengths of the materials present and using appropriately conservative parameters in the design, accounting for liquefied strengths being mobilized in liquefiable soils in recent designs, moving the water pond away from higher consequence structures creating even longer tailings beaches, overall reduction in the amount of water stored on the facility, and transitioning the facility towards closure rather than raising the facility further.

Conclusion With nearly 80 years of mining at RLO, tailings management has evolved to meet changing owner expectations, environmental regulatory requirements, dam safety guidelines, and best practices. Despite these changes and challenging geotechnical conditions at the site, an overall well-functioning tailings storage facility has been established and safely maintained. With mine operation expected to continue for another 15 to 20 years, it is expected that there will be additional changes and challenges to come. Continued diligent operation and monitoring of performance of the tailings facility, and an integrated approach with designers, mine planners, operations, community stakeholders, and regulators will be key to meet those challenges. Evolution Mining is committed to ensuring the socially responsible operation of the MTP throughout the life cycle of the facility.

450

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Modelling of Static Liquefaction of Cadia Failure with Material Point Method Pepe Aynaya, Anddes Asociados SAC, Peru Fabricio Fernández, Universidad Católica del Norte, Chile Raquel Quadros Velloso, Pontifical Catholic University of Rio de Janeiro, Brazil

Abstract Currently, the mining industry performs stress-strain analyses based on continuum models to represent the fundamental mechanics of the problem using classical numerical platforms based on the Finite Element Method (FEM) and Finite Difference Method (FDM). However, these numerical methods satisfactorily model only the onset of the failure at limited strains and displacement levels. On the other hand, more advanced numerical methods, such as the Material Point Method (MPM), are suitable to simulate the failure initiation, the runout processes, and the final deposition of the filled mass. Motivated by recent tailings dam failures worldwide and advances in numerical analysis techniques, this paper seeks to assess and understand the runout characteristics of dam failures caused by the static liquefaction process. We aim to study failure initiation, flow behaviour during the runout process, and deposition patterns of the dam failure. In this work, we model the Cadia dam failure using the Material Point Method (MPM) to understand problems characterized by large displacements and deformations. The geotechnical parameters of the materials used in the numerical model were obtained from the Report on NTSF (Northern Tailing Storage Facility) Embankment Failure of the Cadia (Morgenstern et al., 2019). Additionally, we reviewed the results of the limit equilibrium analyses prepared for the NTSF (Morgenstern et al., 2019) to determine the failure mechanics imposed into the MPM model. This study used an elastoplastic constitutive model with exponential softening to represent the behaviour of the materials. The numerical simulations were performed using the MPM-PUCRio software (Fernández, 2021), an in-house developed C++ code, where the motion equations are solved under an explicit integration scheme.

Introduction Tailing storage facilities (TSF) are the largest geotechnical structures in the mining industry and are used to store residues from the mineral extraction process. These structures are intended to ensure that the tailings

451

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA are safely stored. When a TSF fails and the stored mass spills into the surrounding environment it leads to economic, human, and environmental losses. Usually, these structures are built using the coarse fraction of the tailings or using the rock waste materials from operations; typically, there are three types of raises: upstream construction, downstream construction, and centerline construction. The mining industry has experienced a lot of TSF failures in recent years. The WISE uranium Project published the historical data of failures (http://www.wise-uranium.org/), and Table 1 presents failures from December 2001 to January 2023 around the world. Table 1: Chronology of major tailings dam failures Date

Location

Parent company

Ore type

Type of incident

2023, Jan. 31

Alberta, Canada

Imperial Oil

Bitumen

Process water drainage pond overflow

2022, Nov. 7

Williamson, Tanzania

Petra Diamonds

Diamond

Tailing dam failure

2022, Sep.11

Jagersfontein, South Africa

Jagersfontein Developments

Diamond

Tailing dam failure

2022, Jul. 23

Agua dulce, Bolivia

Mineras de Potosí

Silver, Zinc

Tailing dam failure

2022, Mar. 27

Shanxi Prov. China

Shanxi Daoer Aluminum Co

Bauxite

Tailing dam failure

2022, Jan. 20

Banjhiberana, India

JSW Bhushan Steel Limited

Iron

Breach of tailing pond

2022, Jan. 8

Mina Gerais, Brazil

Vallourec S.A.

Iron

Overflow

2021, Dec. 8

Ulundi, South Africa

Menar Group

Coal

Slurry dam failure

The consequences of TSF failures are catastrophic, so it is crucial to evaluate the stability of those structures using numerical methods and advanced constitutive models to simulate their mechanical behaviour and identify potential catastrophic failures. This paper presents a case study of a TSF in Australia (Cadia) founded on a complex soft clay foundation. After the failure, the Independent Technical Review Board (ITRB) started geotechnical investigations to determine the failure mechanism and the possible causes of the event (Morgenstern et al., 2019). This study reviewed information from the ITRB report regarding the geotechnical properties of the materials and the geometrical definition of the critical section of the analysis. The MPM-PUCRio software (Fernández, 2021) was used to perform a back analysis, using different liquefied zones extension to adjust the numerical model to the final topography.

452

MODELLING OF STATIC LIQUEFACTION OF CADIA FAILURE WITH MATERIAL POINT METHOD

Case study: Cadia failure Failure of Cadia The failure mechanism of Cadia was characterized by 2 phases: Phase 1 was characterized by vertical and horizontal displacements that were not noticed by the mining company’s engineers. After failure, the InSAR images have showed that the displacements had been happening since 2017 and that they increased in 2018. The ITRB associates this information with the load increase during the construction process, especially during the construction of buttress 1 where the displacements increased. The buttresses were recommended to improve the overall FOS of the structure under static and dynamic loading. The increase in load during the construction of buttress 1 caused an accumulation of deformations in the dam foundation. With recent field investigations it was demonstrated that there was a very compressible clay layer with softening behaviour. After the discovery the unit was divided into two subunits: FRV UnitA and FRV Unit-B. Previously it was considered as one geological unit, FRV. The FRV Unit-A layer controls the stability of the dam due to its characteristics, resulting in a progressive failure mechanism. The increase of deformations produced in Unit-A and the small excavation that was made at the foot of the slope accelerated the displacements of the slope, causing the tailings lateral support. This action suddenly causes loss of strength of the tailings, which ultimately leads to an increase in load for the foundation, which had already suffered deformations. The resulting imbalance of forces pushes the dam outwards. Phase 2 was induced by phase 1: the tailings in the affected region passed from the drained condition to the undrained condition due to the reduction of the confining stress and the increase of the deviator stress in the region closest to the slope. The effective stress dropped immediately and finally caused the liquefaction of the most susceptible region. Before phase 1, the tailing was saturated, and very contractive to shear, so it already had the characteristics for liquefaction. Figure 1 shows a cross-section view of the topography of the two phases after failure.

Figure 1: Cross-section view of the Cadia TSF after failure

453

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

MPM model of Cadia TSF The MPM model of the Cadia failure consists of 13,170 particles uniformly distributed in an Eulerian mesh (cells with a 5 m side). The geometry and materials considered for the MPM model have been taken from Morgenstern et al. (2019) and Pierce (2021). Stages 1 to 3 of the Cadia TSF have a slope of 1.5H:1V, and stages 4 to 9 have a slope of 2H:1V. Figure 2 shows the typical section of analysis considered after the revision of the Independent Technical Review Board (Morgenstern et al., 2019). Figure 2 shows the Eulerian grid and Figure 3 shows the material points distribution with the definitions of the materials used in the model. 1. Material 1: Rock fill 2. Material 2: Saturated silty tailing 3. Material 3: Foundation material Unit-A saturated 4. Material 4: Foundation material Unit-B saturated 5. Material 5: Excavation 6. Material 6: Rock fill saturated 7. Material 7: Clay core

Figure 2: Cross-section view of the MPM model (Eulerian grid)

Figure 3: Cross-section view of the MPM model (Material Points)

Geotechnical properties The TSF of Cadia consists of rockfill and clay core, mining tailings, and foundation material. The geotechnical properties of the materials were obtained from the ITRB report (Morgenstern et al., 2019).

454

MODELLING OF STATIC LIQUEFACTION OF CADIA FAILURE WITH MATERIAL POINT METHOD The parameters were obtained from laboratory tests developed before, during, and after construction. After the failure event, laboratory tests were focused on the foundation materials and tailings.

Tailings The tailings of the dam are classified as silty-sandy (SM) according to the USCS (Unified Soil Classification System) classification. The shear strength parameters for the Mohr-Coulomb model were obtained from the undrained triaxial (CU) tests carried out by ATC Williams in 2017, the same ones that were used during the construction of the dam. The material has a cohesion of 0 kPa and a 40° friction angle. The summary of parameters for the material is presented in Table 1.

Rockfill The rock fill is classified as gravel-coarse (GC) according to the USCS classification. The material was modelled with the Mohr-Coulomb model; the shear strength parameters that were considered are shown in Table 1 for the different analysis conditions. Shear strength values were assumed according to the literature due to the difficulties of carrying out a large-scale triaxial test. Initially a strength-stress function based on Leps (1970) was assigned; however, the ITRB updated to a strength function (Morgenstern et al., 2019).

Clay core The core material and the transition material are classified as CL and CH according to the USCS classification. Both were modelled considering the Mohr-Coulomb model; the shear strength parameters considered are shown in Table 1.

Foundation materials Recent laboratory investigations have shown that the foundation can be subdivided into two subunits: FRV (Forest Reef Volcanics) Unit-A and FRV Unit-B. FRV Unit-A is a highly weathered intermediate plasticity clay with low density and high voids index. The FRV Unit B is a highly weathered layer thicker than Unit A, with lower plasticity than Unit A, denser material, and close to rock. The DSS (Direct Simple Shear) and CD (Consolidated Drained) laboratory tests show that Unit-A has a softening behaviour. Tests with shear stresses of up to 1,200 kPa have been developed. On the other hand, Unit-B was characterized using CU (Consolidated Undrained) triaxial tests, which do not show softening behaviour, but for deviation stresses greater than 400 kPa, show softening and lower friction angle. The strength parameters for the Mohr-Coulomb model obtained at the peak of the stress-strain curves are shown in Table 1. However, to model the softening of Unit-A it was necessary to calibrate the exponential stain-softening incorporated in the Mohr-Coulomb model. Figure 4 shows the calibration of a triaxial test for the FRV-Unit A.

455

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 4: Calibration of the Mohr-Coulomb model with exponential softening Table 1: Geotechnical properties of the materials Young Modulus E (kPa)

Poisson Ratio (v)

Rockfill

78,000

Silty tailings

Unit Weight 𝒌𝒈

Cohesion (kPa)

Friction Angle (𝝋′)

0.3

2,100

0

35°

81,000

0.2

1,663

0

32°

FRV Unit A

39,000

0.3

1,390

10

16°

FRV Unit B

39,000

0.3

1,550

40

22°

Excavation

39,000

0.3

1,390

10

16°

Rockfill

78,000

0.3

2,100

0

35°

Clay

37,000

0.2

1,738

10

26°

Material

!𝒎𝟑%

Numerical analysis Methodology Figure 5 shows how we simulated the rupture of Cadia TSF. First, we calculated the initial elastic stress distribution. Next, the Mohr-Coulomb model was used to simulate the conditions before the failure with parameters at the peak of the stress-strain curve. Finally, the excavation, softening, and undrained conditions are activated to capture the failure of the dam.

Figure 5: Flow diagram for the analyses to simulate TSF of Cadia failure

456

MODELLING OF STATIC LIQUEFACTION OF CADIA FAILURE WITH MATERIAL POINT METHOD The linear-elastic analyses were performed to obtain the initial elastic stress distribution. In this stage, we created graphics of the distribution of pore pressures, stresses, and displacements and then compared them with analytical solutions to verify the model. Second, coupled elastic-plastic analysis was performed with the peak parameters of the stress-strain curve. Third, we considered the excavation, softening of UnitA, and the undrained behaviour of the tailing. The failure sequence obtained with the numerical model is shown in Figure 7. Several figures were generated to try to explain the mechanism of the rupture. In order to get the final model we varied different liquefied zones extensions to adjust the numerical model to the final topography taken by ITRB. Figure 6 shows the final topography for each case.

Figure 6: Comparison of topographies for different cases

Figure 7: Simulation of the Cadia failure at different times

457

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 8: Mechanism of Cadia failure In general, the results have shown that MPM can simulate the rupture of the Cadia dam. Figure 8 shows how the rupture of the Cadia dam occurred. Prior to excavation the FRV Unit-A had experienced strains due to the construction of buttress 1; after that, the Cadia failure started with the excavation, represented by point 1. Immediately there was an increase of strains in the FRV Unit-A of the foundation, until it reached the residual shear strength and caused lateral movement of the main slope (see point 2); after that, the tailing changed to undrained condition, and it collapsed, which is represented by point 3 and finally the stress state changes until the model gets the equilibrium, defined by point 4. Figure 9 shows the displacement that has occurred at different stages during the simulation. At t=0 s, it can be seen that the displacements are close to 0; after 2 seconds, there is a displacement of 4.9 m. For t=10 s, the displacements reached 65 m. Finally, for t=50 s, the maximum displacement was approximately 240 m.

458

MODELLING OF STATIC LIQUEFACTION OF CADIA FAILURE WITH MATERIAL POINT METHOD

Figure 9: Displacements during Cadia failure

Conclusions The paper demonstrates the use of MPM to simulate the trigger, the runout process, and the final deposition of the failed tailings in the Cadia dam failure. •

The methodology presented to simulate the failure was consistent with the failure mechanism presented by the ITRB.

•

The final topography of the simulation presents good agreement with the final topography taken from the ITRB.

•

It was shown that the excavation was an important factor in triggering the failure of the TSF.

•

The results showed that final deposition depends on the area of tailings that can change to undrained conditions.

•

The simulation was completed using a grid cell size of 5 m per side and 4 particles in each cell. It is recommended to apply this numerical method to TSFs to evaluate the potential of failure and

possible consequences after the event, varying parameters and identifying the influence of the material on the runout.

459

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Acknowledgements The authors thank the Pontifical Catholic University of Rio de Janeiro-Brazil for their financial support of this study.

References Azam, S. and Li, Q. 2010. Tailings dam failures: a review of the last one hundred years. Geotechnical news 28(4): 50–54. Diehl, Peter. 2023. WISE Uranium Project. Accessed July 8, 2023. http://www.wise-uranium.org/. Fernández, F. 2021. Modelagem numérica de Problemas Geotécnicos de Grandes deformações mediante método do ponto material. PhD thesis, PUC-Rio. Leps, T. 1970. Review of shearing strength of rockfill. Journal of the Soil Mechanics and Foundations Division 96(4): 1159–1170. Morgenstern, N.R., Jefferies, M., Van Zyl, D. and Wates, J. 2019. Report on NTSF Embankment failure. Cadia Valley Operations for Ashurt Australia by Independent Technical Review Board. Pierce, I. 2021. Applying the material point method to identify key factors controlling runout of the Cadia tailings dam failure 2018. Thesis, Virginia Polytechnic Institute and State university.

460

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Case Study: Geotechnical Studies to Improve Understanding of Material Parameters and Address Changes in Stability Requirements, Granny Smith Gold Mine, Western Australia Louise McNab, Gold Fields Ltd, Australia Johan Boshoff, Gold Fields Ltd, Australia Calvin Wang, Gold Fields Australia, Australia

Abstract Upstream construction relies on the integrity of the tailings for stability. As such, while this construction method has been successfully used for decades, these designs require greater ongoing scrutiny. The Tailings Storage Facility (TSF) complex at the Granny Smith mine comprises three cells and was designed preANCOLD guidelines with minimum allowable factors of safety for static conditions of 1.3. Previous investigations on Cell 1 identified horizontal layering resulting from deposition of primary and oxide ore tailings. The oxide layers were located beneath the top elevation of and adjacent to the original starter embankment, covering approximately 50% of the TSF floor area. The Engineer of Record assessed that the oxide tailings would likely exhibit brittle behaviour and very low post-peak undrained shear strength ratios. As a result, preliminary stability assessments yielded factors of safety below 1.5 under undrained static loading conditions. Deposition into Cell 1 was therefore suspended and deposition diverted to Cell 3. This paper presents the approach followed to confirm the stability of the cells and to optimize the geometry of the toe buttress (including geotechnical investigations, laboratory testing, and deformation modelling). The dynamic deformation modelling completed to predict the deformation of the TSF under the seismic design load and to check the propensity of the low predicted post-seismic strengths to trigger under the seismic design event is discussed. The paper highlights the importance of material characterization, particularly for upstream constructed dams, and that brittle materials are a special case that requires conservative design.

Background The Granny Smith Gold Mine, situated in the Goldfields-Esperance region of Western Australia, is home to a significant Tailings Storage Facility (TSF) complex consisting of four cells: Cells 1, 2, 3, and 4. The

461

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA design of these cells predates the implementation of the Australian National Committee on Large Dams (ANCOLD) guidelines, with minimum allowable factors of safety set at 1.3 for static conditions. While upstream construction has proven effective over several decades, it necessitates heightened scrutiny as design practices evolve. The stability of the TSF holds paramount importance, as the integrity of the tailings serves as the foundation for upstream construction. Thus, it becomes imperative to conduct geotechnical studies and enhance our understanding of material parameters to ensure the continued stability of these structures. Soil and Rock Engineering Pty Ltd undertook the original design of Cell 1 and Cell 2 for Placer Dome Pacific, with the first stage of construction completed in 1989, utilizing low permeability, well-graded, overburden waste materials. Over the years, the embankments of Cell 1 and Cell 2 were raised ten times in an upstream direction using compacted tailings, reaching their final permitted elevations in 2016 and 2012, respectively. In 2001, Cell 3 was designed by Knight Piésold (KP), with the starter embankment constructed between 2001 and 2002 at RL 420.0 m. Subsequent raises of Cell 3 were carried out, with the most recent raise (Stage 3F) designed by Golder and constructed from September 2021 to February 2022, bringing the crest elevation to RL 433.7 m. Another raise (Stage 3G) to the final approved crest elevation of RL 437 m is proposed for 2024. Since March 2022, Cell 3 has been the active deposition cell, accommodating ongoing tailings deposition. An aerial photograph of the facility is presented in Figure 1.

Figure 1: Aerial Image of the Granny Smith TSF

Overview of the geotechnical studies conducted Study objectives In 2019, preliminary limit equilibrium models indicated that a buttress may be required along a portion of

462

CASE STUDY: GEOTECHNICAL STUDIES TO IMPROVE UNDERSTANDING OF MATERIAL PARAMETERS AND ADDRESS CHANGES IN STABILITY REQUIREMENTS, GRANNY SMITH GOLD MINE, WESTERN AUSTRALIA the Cell 1 TSF embankment. This outcome was derived from assumed information and limit equilibrium models that had not been independently reviewed. Following an independent review, fundamental flaws were found in the modelling approach. Given the gaps in the geotechnical characterization information, in April 2020, Gold Fields commissioned a comprehensive geotechnical investigation at the Granny Smith Gold Mine Tailings Storage Facility (TSF). The primary objective of the field investigation was to gather crucial data to confirm stability analyses and, if required, develop deformation numerical models, aimed at understanding the behaviours of the materials within the TSF. The deformation models would also be used to refine the geometry of any buttresses required for portions of the TSF embankment. To achieve the study outcome and geotechnical field campaign, samples of foundation materials, starter embankment materials, and tailings were retrieved for subsequent laboratory testing.

Description of the methodology employed in the investigation and laboratory testing campaign The geotechnical investigation involved a methodology to ensure accurate data collection and analysis. Field sampling was conducted, targeting core samples from various locations within the TSF complex. Additionally, a cone penetration testing (CPT) campaign and borehole drilling were carried out, providing access to piston tube samples for laboratory testing. These samples encompassed the foundation, starter embankment, and tailings materials, allowing for a comprehensive assessment of their physical and mechanical properties. The scope of work for the drilling campaign included sonic drilling to determine suitable locations for cone penetration testing (CPT) probing and in-situ sampling. The objective was to establish precise locations for further data collection and analysis. Regarding the laboratory testing campaign, the scope of work encompassed the following: • Conduct classification and advanced laboratory testing on selected foundation, starter embankment, and tailings materials. This involved comprehensive analysis to determine the physical and mechanical properties of these materials. • Interpret the geotechnical behaviours of the materials, focusing on both monotonic and cyclic resistance. The laboratory test results were analyzed to gain insights into the response of the materials under different loading conditions, including static and cyclic loading. • Estimate material parameters for slope stability and deformation analyses. This involved utilizing the limit equilibrium method (LEM) and performance-based numerical deformation analysis. By determining the material parameters, such as shear strength, stiffness, and deformability, it was possible to perform detailed stability and deformation analyses to assess the performance of the TSF under various loading scenarios.

463

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Identifying horizontal layering and oxide tailings in Cell 1 A specific focus of the geotechnical investigations was the identification and characterization of horizontal layering and oxide tailings within Cell 1 of the TSF. This involved careful examination of the retrieved samples and laboratory testing to determine the distinctive properties of these materials. Understanding the presence and behaviours of oxide tailings, which were found beneath the top elevation and adjacent to the original starter embankment, was crucial for subsequent stability analyses. By pinpointing the horizontal layering and oxide tailings, the geotechnical investigations aimed to provide valuable insights into the unique challenges posed by these materials and their influence on the stability of Cell 1.

In-situ sampling in the field Push tubes, core, and bulk samples were obtained through borehole drilling and cone penetration testing (CPT) at locations around the TSF. Material representative of starter embankment material was also sampled from local waste rock dumps to complement the lack of material required to carry out the laboratory testing program on the coarser material type. The material's representativeness was confirmed through comparison of particle size distribution testing (PSD). All samples were collected by Golder and sent to Golder's Perth laboratory for testing.

Optimization of toe buttress geometry Rationale for optimizing the geometry of the toe buttress The initial limit equilibrium analyses performed for the Granny Smith TSF indicated that a buttress was necessary to achieve satisfactory performance in accordance with international and local regulations. This preliminary limit equilibrium assessment had fundamental flaws found by an independent reviewer and was repeated using measured site and laboratory test information. The revised limit equilibrium models also highlighted that a buttress would be required. As such, advanced deformation models were developed. The deformation analyses undertaken also indicated that buttress construction was prudent. However, the initial buttress dimensions assumed via limit equilibrium modelling could be reduced while keeping the same level of performance. The optimization of the toe buttress geometry was undertaken to enhance the stability and performance of the TSF at the Granny Smith Gold Mine. The toe buttress plays a critical role in providing additional support and stability to the embankment, particularly in areas where the tailings are deposited upstream. By optimizing the geometry, the aim was to ensure that the toe buttress effectively resisted potential deformations and increased the overall stability of the TSF. This optimization process was driven by the need to mitigate potential risks and ensure the long-term integrity of the facility.

464

CASE STUDY: GEOTECHNICAL STUDIES TO IMPROVE UNDERSTANDING OF MATERIAL PARAMETERS AND ADDRESS CHANGES IN STABILITY REQUIREMENTS, GRANNY SMITH GOLD MINE, WESTERN AUSTRALIA Through the combination of geotechnical investigations, laboratory testing, and deformation modelling techniques, the optimization of the toe buttress geometry aimed to improve the stability and performance of the TSF at the Granny Smith Gold Mine. By considering the rationale for optimization, the details of the investigations and testing, and the use of deformation modelling, a comprehensive approach was employed to enhance the overall stability of the facility. Overall, the paper highlights the importance of conducting geotechnical studies to ensure the continued stability of the Tailings Storage Facility at the Granny Smith Gold Mine. The investigation and laboratory testing provided crucial data for stability analyses, deformation modelling, and optimization of the toe buttress geometry. By understanding the material parameters and behaviours, including the identification of horizontal layering and oxide tailings, the study aimed to improve the overall stability and performance of the facility. The findings of the geotechnical investigations, laboratory testing, and deformation modelling techniques contribute to the ongoing efforts to ensure the long-term integrity of the TSF complex at the Granny Smith Gold Mine in Western Australia.

Applying advanced geotechnical monitoring to the buttress Gold Fields collaborated with the Australian Minerals Research Institute to gain access to the Granny Smith TSF as part of their TSF monitoring project. The objective was to establish a test site for trialling a range of advanced geotechnical monitoring equipment. This equipment was intended to enable Gold Fields to monitor and assess the condition of the facility before and after the construction of the buttress. The technology installed included: •

radar monitoring,

• real-time accelerometers, • seismic geophones, • fiber optic geophones, •

InSAR (Interferometric Synthetic Aperture Radar),

• continuous monitoring stations, and • vibrating wire piezometers. These monitoring tools provided valuable data and insights into the behaviour and performance of the TSF, allowing for effective monitoring and ensuring the ongoing stability and safety of the facility.

Lessons learned and practical implications The case study at the Granny Smith Gold Mine provided valuable insights and lessons learned in addressing stability concerns. Discovering a low factor of safety can be alarming, and addressing such concerns can

465

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA be complex and time-consuming, particularly due to the intricacies of geotechnical modelling. The preliminary limit equilibrium model, upon undergoing an independent review, revealed fundamental flaws in the assumed material properties. This highlights the importance of independent reviews in verifying and correcting modelling assumptions. While the field investigation took time to complete, the outcomes instilled confidence in the proposed buttress design and validated the need for further measures to ensure stability. Although deformation modelling was a costly and time-consuming exercise, it played a pivotal role in optimizing the design and reducing costs. By refining the required buttress geometry through deformation modelling, the project achieved significant cost savings amounting to millions of Australian dollars. This underscores the importance of investing resources in accurate and comprehensive deformation modelling, as it can result in substantial cost reductions in the long term while maintaining the necessary level of stability and performance. The lessons learned from the case study at Granny Smith Gold Mine emphasize the criticality of conducting thorough independent reviews, investing in comprehensive field investigations, and utilizing deformation modelling techniques to refine design parameters. These practices can lead to improved understanding, more efficient designs, and cost-effective solutions for ensuring the stability and long-term viability of tailings storage facilities.

Conclusion In conclusion, the case study conducted at the Granny Smith Gold Mine has provided valuable insights and practical implications for the stability and performance of the TSF. The study emphasized the importance of conducting geotechnical investigations and enhancing our understanding of material parameters to ensure the continued stability of these structures. Through comprehensive field investigations, laboratory testing, and deformation modelling techniques, key findings were obtained. The lessons learned from the Granny Smith Gold Mine case study underscore the importance of thorough independent reviews, comprehensive field investigations, and accurate deformation modelling techniques. By implementing these practices, engineers and stakeholders can improve their understanding of TSF behaviour, refine design parameters, and ultimately enhance the long-term viability of tailings storage facilities. Continued scrutiny and ongoing efforts to enhance material characterization and stability analysis are vital in ensuring the integrity and safety of these structures in the mining industry.

References Australian National Congress on Large Dams. 2012. Guidelines on Tailings Dams – Planning, Design, Construction, Operation and Closure, Australian National Congress on Large Dams, Hobart.

466

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Back Analysis of Runout Mechanism of a Static Liquefaction Induced Failure Using the Material Point Method Masood H. Kafash, AECOM, USA Alfonso Cerna-Diaz, AECOM, USA Lisa Yenne, AECOM, USA Richard Davidson, RRD LLC, USA

Abstract Estimating the extent of runout is a key factor for risk assessment and risk mitigation of tailings storage facilities (TSFs). Although the fluid-type runout analysis using dam breach tools has been well established, the evaluation of slump-type runout is still developing and requires validation of estimation tools with case histories. In this paper, a well-documented slump-type runout from a static liquefaction case history of an upstream tailings storage facility will be presented, and the potential use of the Material Point Method (MPM) to estimate runout mechanism will be examined. MPM was developed initially by Dr. Deborah Sulsky and her colleagues at the University of New Mexico in 1994. In the MPM approach, the continuum body is discretised by a predetermined number of Lagrangian particles or Material Points (MP). The use of MPM in geotechnical engineering and estimation of runout has been expanded and advanced over the past few years. Results of the computed case history back analysis will be compared with the observed runout distance and runout travel time to evaluate the capabilities of MPM to effectively capture the runout mechanism. The results highlight the advantages of using a MPM model for simulating slump-type runout in geotechnical analyses.

Introduction This paper presents a slump-type runout analysis of an upstream tailing impoundment that failed at approximately 4:15 p.m. on February 3, 1998, due to static liquefaction, and evaluates the potential use of MPM method to capture the runout mechanism. In this paper, tailings impoundment is referred to as “TSF.” The following provides an overview of events prior to failure, the failure mechanism, events following runout, and MPM analysis of the runout.

467

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA During the construction of a step-back berm along the northeast corner of the TSF, failure of the berm and underlying tailing slope occurred, which initiated a slump-type runout of tailings material for 300 to 500 feet. The reported failure durations vary, but the best estimate was 10 to 15 seconds. The failure, shown in Photograph 1, initiated a major investigation and reconstruction program, which revealed several important findings, as discussed below.

Photograph 1: Failed slope ln September 1997, approximately six months prior to the failure, the designer identified that pore pressures at the TSF northeast corner were higher than projected in initial design. The higher pore pressures dropped the static undrained factor of safety below the required minimum value of 1.3. As a result, the designer recommended: 1. complete the next raise with a 100-foot step-back of the embankment crest; 2. install additional wick drains; and 3. construct a toe berm along the north slope to increase the factor of safety to 1.3. By February 3rd, 1998, the installation of the first stage of additional wick drains and the construction of the first northeast corner step-back had been completed, and the company was in the process of

468

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD constructing the toe berm at the northeast corner. The berm construction methodology proposed included a mix of mechanical placement of dry whole tailings taken from previously deposited material on the southeast corner of the impoundment and hydraulic tailings deposition. The general berm construction sequence consisted of site preparation, dry whole tailings import, and hydraulic tailings deposition. The contractor began site preparation work on October 23, 1997. This consisted of developing an access road along the east end of the berm area, constructing containment dikes to form two cells, and installing the whole tailings feed pipeline and decant system. Containment dikes were constructed both from dry whole tailings hauled from the southeast corner and from limited local on-site excavation from the existing dike roads along the cells. Dry whole tailings from the southeast corner were hauled to the containment dikes and to the berm fill area beginning on October 30, 1997, and continued through February 3, 1998 (the day of the failure). An estimated 136,418 cubic yards of whole tailings was imported to the berm. The hydraulically placed tailings materials were provided by installing a 20-inch diameter HDPE pipeline and valve assemblies from the perimeter tailings distribution system to the east end of the proposed berm area. One decant structure was installed in the west end of each of the two berm cells to allow for discharge of clarified tailings water through a 16-inch diameter HDPE pipeline, emptying to the north of the existing impoundment. The decant structures were constructed to allow extension of the structures as the berm was raised. Whole tailings slurry deposition into the two cells began on November 24, 1997. The original plan called for deposition to occur on a four-day cycle for 24 hours at a time. However, the actual records indicated deposition took place only six times, being mainly accomplished during the day shift. Difficulties encountered by the contractor in operating the bulldozers in freshly deposited whole tailings required extending the cycle time to allow more time for the tailings to dry between deposition cycles. An estimated 64,000 cubic yards of whole tailings were deposited hydraulically. ln December 1997, it was decided to replace the hydraulically-deposited whole tailings with underflow sand. Hauling of dry whole tailings import was continued to complete berm construction. Hydraulic placement of the tailings underflow was handled in a similar manner as whole tailings deposition. Underflow deposition was configured to allow single point discharge into the lower cell and header-andspigot discharge into the upper cell. The underflow slurry was transported to the berm via a 12-inch diameter pipeline, and decant water was removed through the same decant structures installed previously. Hydraulic placement of underflow began on January 3, 1998. Underflow was not placed every day because, as before, the contractor also had difficulty tracking through the underflow without allowing time for drying. On or about January 8, 1998, the methodology of depositing into two cells was stopped. Because of limited width in the lower cell, the underflow was not drying sufficiently to be workable by bulldozers, and several bulldozers became stuck in the underflow deposited. It was decided to begin placing underflow

469

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA only in the upper cell, which allowed the material to flow into thinner lifts across a larger area. The thinner lifts were more manageable with equipment, and the water in the underflow slurry decanted more readily. The material deposited in the upper cell was then bulldozed to the lower berm area as it dried. The conversion from a two cells hydraulic deposition to an upper cell only was completed on January 12, 1998. According to the contractor records, between that time and the failure, approximately four feet of tailings was bulldozer-pushed from the upper to the lower cell at the east end of the berm over an unknown lateral extent. Several problems were observed during periodic inspections of the deposition cells in the weeks preceding the failure including: • Containment dikes in the upper and lower cells were often breached. • Significant channelized erosion was observed during spigotting in the upper cell. • Seepage was observed at the toe of the lower cell starter dike slope. • The cells drained much slower than expected. • Pore pressures were raising in the lower portions of the impoundment slope. • Two sets of cracks were observed within the upper cell tailings, believed at the time to be settlement-induced. Some of the cracks were longitudinal parallel to the containment dike, and some were transverse across the cell.

Eye-witness observations on the day of failure The failure was witnessed by six employees, including two bulldozer operators who were carried by the slide some 300 to 500 feet without injury or damage to the equipment. The eyewitness locations are depicted in Figure 1.

Figure 1: Location of upper and lower cells, eye-witnesses and dikes

470

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD Reportedly the lower dike had been raised 4 feet on the day of the failure. A portion of that dike had previously eroded 2 feet, which was apparently a common problem in the weeks before the failure. Tailings were being spigotted above the upper cell during the failure. The failure probably initiated on the lower starter dike near the center of the failure zone and progressed rapidly up slope and laterally. The reported failure durations vary, but the best estimate was 10 to 15 seconds. Intact blocks of “dry” material apparently flowed downstream on a mass of liquefied tailings. The bulldozer operators described it as though “an enormous piece of plastic was being pulled out from under the tailings embankment.”

General description of material types The following section provides a general description of the soils and materials encountered at the northeast corner and establishes the nomenclature to be used in the analysis discussion.

Embankment tailings ln general, the tailings in the northeast corner embankment can be divided into two categories: whole tailings, which are highly interbedded and relatively coarse-grained in nature, and typically classified as a silty sand interbedded with silts and silty clays; and decant pond clay, which is fine-grained and typically classified as a low-to-medium plasticity silty clay. There are two units of whole tailings at the northeast corner: an upper layer of sandy beach deposits and a deep layer underlying the decant pond clay in some areas. The sandy layers of whole tailings have a maximum particle size passing a No. 40 sieve and typically contain between 40% and 70% passing through the No. 200 sieve. The decant pond clay is a medium to high plasticity clay with a liquid limit ranging from approximately 40 to 75, with an average of about 64, and a plasticity index ranging from 23 to 46 with an average of about 34. The tailings along the failure surface were interbedded clays and sands, as shown in Photograph 2.

Photograph 2: Sample through northeast corner failure surface (1998) Cone penetration test (CPT) soundings were advanced through the failed tailings and the failure surface to characterize the tailings below the slide and to develop a profile of the failed section prior to the failure, as shown in Figure 2. Plots of data for CP98-953 are shown in Figure 3. The CPT data also indicates there are interbedded clays and sands in this area, and that the tailings are contractive and have high pore pressures.

471

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 2: Cross-section of failed section and location of CPTs

Figure 3: CPT signature and SBTn charts – CP98-953

472

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD Back analysis of the failure based on the estimated pre-slide configuration and the length of the runout for the failure indicated that the undrained shear strength ratio along the failure surface was approximately 0.12. Although there were sufficiently high pore pressures and there were low permeability cohesive tailings in the failure area, there is no indication that void redistribution occurred.

Dikes Three dike units are found at the northeast corner: the Starter Dike, the 50 Dike, and the 52 Dike. The Starter Dike consists of 3 to 4 dike raises and was reportedly constructed of dumped mine waste ranging in size from gravel to boulder-sized particles. The competency of the starter dike materials varies; in places with substantial quantities of boulders and cobbles, relatively competent Quartzite rock is found. However, in other areas the starter dike material has weathered considerably to silt and sand size matrix material, and is missing the boulders found in other areas. The 50 Dike and 52 Dike were at least partially constructed of mine waste; however, in some areas alluvial sands and gravels are exposed on the slide scarps surrounding the failure area.

Foundation The foundation of the TSF consists of lake clays interbedded with lenses of sands from various deltaic cycles. The top of the upper clay unit, which comprises the upper approximately 10 to 15 feet of the foundation, is marked by the Gilbert Red Beds, so-called because of oxidation stains that give it a reddish appearance. The clays in this unit are occasionally interrupted by sand beds, typically less than 2 feet thick. Beneath the upper clay unit are various interbedded clays and sands, referred to as the interbedded sediments. Within the free field the foundation clays are over-consolidated, with overconsolidation ratios (OCR) ranging from 4 to 15. They are of low to medium plasticity, with liquid limits ranging from 35 to 52, with an average of about 40, and a plasticity index ranging from 12 to 30, with an average of about 20. The sands are generally fine- to medium-grained, poorly graded and dense.

Pore pressure conditions Previous analyses at the TSF have found the pore pressure regime at the northeast corner to be extremely complicated and ill-suited for the use of a piezometric line to calculate in-situ effective stresses. As a general rule, the pore pressures in the whole tailings exhibit a downward gradient, typically between 60 percent and 70 percent of hydrostatic. However, in the decant pond clay at the northeast corner, pore pressures significantly higher than hydrostatic have been observed since about 1993. A series of wick drain campaigns have been completed since 1993 just below the embankment crest to control excess pressures in the slope.

473

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Therefore, pressure contours have been developed to characterize these complex conditions. In-situ pore pressures in the embankment have been measured using either pneumatic or vibrating-wire sensors installed in isolated-tip piezometers.

Runout analysis approach Runout analyses were performed to estimate the runout distance from the failed section using the MPM method, following total stress and effective stress approaches. In the total stress analysis approach, a predefined failure surface, based on the failure survey, were modelled, and total stress properties back calculated from the failure were applied to the model. In the effective stress analysis approach, a fully coupled PM4sand/PM4silt model (Boulanger and Ziotopoulou, 2019) were used to initiate the triggering of liquefaction, no pre-defined failure, and then results were imported into the MPM model for runout analysis, as further discussed below.

Material Point Method (MPM) background The MPM was developed by Dr. Deborah Sulsky in 1994. In the MPM, the continuum body is discretized by a predetermined number of Lagrangian particles or Material Points (MPs). In addition to carrying external loads, these MPs carry all the information of the continuum such as mass, volume, density, gradient, displacement, velocity, acceleration, stress, strain, material parameters, strength parameters, etc. In geotechnical engineering problems, the MPs represent a small portion of the entire continuum and not the individual soil grains. Each material point moves attached with the solid body and hence constitutes the Lagrangian representation of the medium. An Eulerian mesh is used, which covers the entire domain in which the body is supposed to move. All displacements and deformations are numerically calculated in this computational mesh. The system of equilibrium equations is solved in the mesh grid, but usually it does not deform with the body, unlike the Finite Element Method (FEM). Mapping functions are used to transfer the variables from the MPs to the nodes. The conservation of mass is satisfied because each material point always contains a fixed amount of mass (Sulsky et al., 1994; Soga et al., 2016).

Total stress analysis approach Previous back analysis performed on the 1998 northeast failure case history was used to extract the constitutive behaviour of the tailings and foundation materials from small to large strain stress-strain range. The following MPM analysis builds on the northeast failure case history, based on the following assumptions and inputs from the failure:

474

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD • Knowledge of the failure surface (see Figure 4) after an extensive site investigation performed in 1998. • Knowledge of the post-failure geometry (see Figure 4) and range of observed runout distance that occurred in the failure.

Figure 4: Pre-failure and post-failure geometry of the northeast failure The MPM analysis was first based on a total stress analysis using a Mohr Coulomb material model that allows for evaluations of the mobilized frictional resistance of the contractive whole tailings materials and the mobilized friction between the foundation materials and the tailings. Simulations were completed using Anura3D software. The actual observed failure surface was used to divide the moving mass with respect to the fixed base. Figure 5 shows the assigned material and the failure surface in total stress analysis. The instability was triggered using gravity and the strengths of materials, assuming that all materials above the failure surfaces reached residual conditions instantaneously at t=0. The following items summarize the main modelling aspects: • The Mohr coulomb constitutive model was used for soils/tailings (moving material). The base of failure plane was assumed to be linear elastic with high density and elasticity. • The size of each element for meshing was 2 m. • The thickness of the model in z-direction was 2 m or 6.6 feet (1 element thickness). • Tetrahedral elements were used for meshing. Four MPs per element were used for all the moving material and one MP per element for the basal shear plane.

475

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 5: Material definition

MPM runout back analysis results – total stress analysis The results of the runout back analysis are presented in Figures 6 and 7 at the end of runout.

Figure 6: MPM analysis of northeast corner initial geometry, final time histories locations, and runout time histories

Figure 7: Observed versus MPM post-failure geometry

476

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD Figure 6 provides the final location of key points within the runout mass. In addition, time histories of key points are provided in this figure. As shown in this figure, the toe of the slide moved approximately 400 feet in 25 seconds. This measurement compares well with the observed time, 10 to15 seconds, and the distance reported for the runout, which was 300 to 500 feet. Figure 7 compares the deformed shape of the runout and the post-failure geometry of the failed slope. As shown in this figure, the deformed shape also reasonably matches the observed field condition.

Effective stress analysis approach Effective stress analyses were completed with the help of FLAC (Fast Lagrangian Analysis of Continua, version 8.0) (Itasca, 2019) along with PM4Silt and the PM4Sand (Boulanger and Ziotopoulou, 2019) constitutive model to initiate the triggering mechanism and identify the liquefied area for the modelling assignment. In effective stress analysis the full geometry of the slope was modelled, instead of a pre-defined failure surface as used in the total stress approach. Fully coupled effective stress analysis was performed by building the model in stages following the construction history sequences. The material properties were calibrated to the observed behaviour, including the failure mass and the pore pressure generation prior to the failure, as measured by a piezometer nearby. Figure 8 provides the shape of failure simulated in FLAC2D, which compares well with the observed failure. Figure 8 also provides a comparison between the pore pressure generation during the construction prior to the failure and a FLAC2D simulation. This figure provides an example of sensitivity analyses that were performed to match the observed measurements by changing the rate of strain softening of the tailings material. Further details on triggering analysis will be presented in a forthcoming paper.

Figure 8: Simulated failure in FLAC2D and comparison of pore pressure generation and simulations MPM analyses were then completed using the Mohr-Coulomb model within the Anura3D, following the steps discussed above and summarized in Figure 9. As shown in this figure, the liquefied area identified from FLAC2D was modelled using the liquefied shear strength parameters.

477

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 9: MPM properties assignment and modelling approach

A comparison between MPM runout analyses from effective stress analysis and observed movements is provided in Figure 10. As shown in this figure, the MPM model provides an acceptable estimation of runout distance, with the toe of the slide moved approximately 270 to 400, feet as compared to the 300 to 500 feet observed. The MPM model also showed a time within 25 to 30, seconds as compared to 10 to 15 seconds. This measurement compares well with observed time and distance reported for the runout.

478

BACK ANALYSIS OF RUNOUT MECHANISM OF A STATIC LIQUEFACTION INDUCED FAILURE USING THE MATERIAL POINT METHOD

Figure 10: MPM analysis of northeast corner runout distance comparison with observed values

Summary and conclusion Estimating the extent of runout is a key factor for risk assessment and risk mitigation of TSFs. While the fluid-type runout analysis using dam breach tools has been well established, the evaluation of slump-type runout is still developing and requires validation of tools with case histories. In this paper, a welldocumented slump-type runout from a static liquefaction case history of an upstream tailings storage facility was presented, and potential use of MPM on estimation of runout mechanism was examined. The analyses were performed using total stress and effective stress approaches. Results of case history back analysis were presented and compared with the observed runout distance and runout travel time. The results show reasonable agreement between simulations and observed runout distance and runout time and highlight the advantages of MPM models for simulating slump-type of runout in geotechnical analyses.

References Boulanger, R.W. and K. Ziotopoulou. 2017. PM4Sand: A Sand Plasticity Model for Earthquake Engineering Applications, Version 3.1. Report No. UCD/CGM-17/01, Center for Geotechnical Modeling, Department of Civil and Environmental Engineering, University of California, Davis. Boulanger, R.W. and K. Ziotopoulou. 2019. A constitutive model for clays and plastic silts in plane-strain earthquake engineering applications. Soil Dynamics and Earthquake Engineering 127: 105832.

479

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Itasca Consulting Group, Inc. (Itasca). 2019. FLAC — Fast Lagrangian Analysis of Continua. Version 8.0. Minneapolis, MN: Itasca. Soga, K., Alonso, E., Yerro, A., Kumar, K. and Bandara, S. 2016. Trends in large-deformation analysis of landslide mass movements with particular emphasis on the material point method. Géotechnique 66(3): 248–273. Sulsky, D., Chen, Z. and Schreyer, H.L. 1994. A particle method for history-dependent materials. Computer Methods in Applied Mechanics and Engineering 118(1–2): 179–196.

480

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Evaluation of Flow Liquefaction Susceptibility of Uranium Tailings from CPTu Stefhany Melendez, University of Applied Sciences, Peru Brahian Roman, University of Applied Sciences, Peru

Abstract Large-scale mining operations generate vast quantities of tailings that are placed in hydraulic-fill tailing dams. The stability of these tailings dams against flow failure triggered by liquefaction is of paramount importance to protect lives, infrastructure, and the surrounding environment. In 2019, a massive catastrophic flow failure of an iron ore tailings dam took place in Brazil, raising concerns on the safety of these type of facilities around the world. The phenomena of flow failure triggered by either static or dynamic liquefaction of uranium tailings dams have not been studied extensively. Unlike conventional tailings from metal mining, the different mineralogy and radioactive properties of uranium tailings pose a challenge on its geotechnical characterization. Furthermore, effects of ageing and cementation are not clearly understood. This paper aims to study the flow liquefaction susceptibility of uranium tailings in a case study located in a country with low seismic hazard, and therefore where static liquefaction is prevalent. The framework of critical state soil mechanics was applied using data from a comprehensive site investigation including several Cone Penetration Tests (CPTu). The purpose is to enhance the understanding of the behaviour of uranium tailings at large deformations and to verify the applicability of state-of-practice screening methods such as Plewes (1992) and Robertson (2010). Results show that these screening methods for defining whether a uranium tailings layer is either contractive or dilative are not in agreement most of the time. Thus, a word of caution is recommended when applying these methods to uranium tailings, and it is necessary to conduct further analysis.

Introduction Many studies have been carried out on tailings dams with the aim of evaluating their susceptibility to flow liquefaction using empirical methods and the framework of critical state soil mechanics. This trend started after the catastrophic failures due to static liquefaction that took place in the last decade such as Fundão in 2015 (Brazil), Luoyang in 2017 (China), Cadia in 2018 (Australia) and Córrego de Feijão in 2019 (Brazil).

481

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The application of in-situ testing offers an appropriate assessment of susceptibility to flow liquefaction in mine tailings through empirical and semiempirical methods. The Piezocone test with pore pressure measurement (CPTu) provides three continuous records of the soil being tested, estimating cone tip resistance (qt), sleeve friction (fs), and dynamic pore pressure (u2). Mayne and Sharp (2021), Castillo et al. (2022), Faria et al. (2023), Dos Santos Junior et al. (2022) and Smith et al. (2021) presented case studies using the CPTu test to assess the susceptibility of silty soils including mine tailings to flow liquefaction. Several methodologies (screening tools) were used, including empirical correlations to obtain the state parameter (Plewes et al., 1992), the boundary of the normalized clean-sand tip resistance (Qtn,cs) proposed by Robertson (2016), the Yield Strength Ratio (Mayne et al., 2019) and the contours stated by Shuttle and Cunning (2008). Unlike other types of mine tailings (copper, gold, bauxite), uranium tailings lack specific published studies on their undrained behaviour at large deformations. This knowledge gap needs to be filled through case studies and applied research. Therefore, this paper aims to show the evaluation of the susceptibility to flow liquefaction of a uranium tailings dam, where two of the aforementioned screening tools were applied. In this research, the information was obtained from CPTu testing carried out in a uranium tailings dam located in an area where seismic hazard is low and therefore static liquefaction is prevalent. In this research, two screening tools were used: empirical evaluation of the state parameter by Plewes et al. (1992), and the boundary for contractive and dilative soils using the clean-sand normalized tip resistance established by Robertson (2016).

Literature review Inferring state parameter from CPTu To identify whether a soil behaves in a contractive or dilative manner at large deformations, Shuttle and Cunning (2008) defined a threshold for the state parameter ''ψ'' equal to -0.05, where ψ-0.05 indicates contractive behaviour. The following expressions were used for processing the CPTu data: 𝑄" = 𝑘 exp(−𝑚𝜓) =

𝑞. − 𝜎01 𝜎 2 01

Where: 𝑄" : Normalized tip resistance by the vertical effective stress

𝜓: state parameter

Then, the following equation is obtained: 𝑄 ln < " = 𝑘 𝜓=− 𝑚

482

EVALUATION OF FLOW LIQUEFACTION SUSCEPTIBILITY OF URANIUM TAILINGS FROM CPTU

And: 𝑘 =8+

0.55 𝜆DE − 0.01

𝑚 = 8.1 − 2.3 log 𝜆DE

Where: 𝑚, 𝑘: state parameter constants

Being λ10 a value for each type of soil found by the CPTu, which can change significantly according to the gradation and stress conditions.

Method based on normalized tip resistance (Robertson, 2016) Robertson (2016) proposed a boundary between contractive and dilative soils using clean-sand normalized tip resistance (Qtn-cs). The threshold for Qtn-cs is equal to 70. If Qtn-cs 1.55

4 ha - 15 ha

Figure 2: Operating parameters critical to liquor management and factor of safety BHPOD applies the observational approach (ANCOLD, 2012) with TSFs constructed using the upstream method with a buttress. Throughout the life of operation, improvements have been implemented to improve drainage and pore pressure management in the TSFs. A comparison between old and new TSFs is shown that demonstrates the pore pressure improvement and subsequent impact this has on buttressing activities when applying the limit equilibrium approach for a post-peak design case. In addition, lessons learned in managing the liquor balance are shared, and is a key enabler in identifying when additional EP storage is required to ensure TSF ponds are suitably managed.

Liquor balance model A steady-state condition of evaporative area is defined to maintain TSF pond sizes within safe operating

717

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA limits. The site liquor balance is a function of both the evaporative area and storage volume. Storage volume is due to seasonal changes in rainfall and evaporation, as well as the complicating fact that solids, or salt precipitates, accumulate in the base of the EPs. TSF5 was commissioned in 2011, with the design developing a liquor balance model to forecast future evaporation pond requirements. Key components of the liquor balance and the GoldSim model developed are included in Figure 3. Complicating factors of the model included determining seepage from the impoundment, wet beach evaporation, and entrainment of liquor across the entire profile of the tailings beach. It was recognised that without routine maintenance of the model through actual measurements (such as wet beach evaporation), calibration and revalidation of the model was not sustainable.

Figure 3: Key water balance components

A new GoldSim model was developed in 2018 using a simplified approach. The balance within the TSF is presented in the equation below, where the supernatant runoff is the slurry liquor reporting to the decant pond, i.e., after accounting for entrained liquor, beach seepage, and beach evaporation. 𝑆𝑢𝑝𝑒𝑟𝑛𝑎𝑡𝑎𝑛𝑡 𝑅𝑢𝑛𝑜𝑓𝑓 + 𝑅𝑎𝑖𝑛𝑓𝑎𝑙𝑙 𝑅𝑢𝑛𝑜𝑓𝑓 = 𝐷𝑒𝑐𝑎𝑛𝑡 𝑃𝑜𝑛𝑑 𝐸𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑖𝑜𝑛 + 𝐷𝑒𝑐𝑎𝑛𝑡 𝑃𝑜𝑛𝑑 𝑆𝑒𝑒𝑝𝑎𝑔𝑒 + 𝐷𝑒𝑐𝑎𝑛𝑡𝑒𝑑 𝑂𝑢𝑡𝑓𝑙𝑜𝑤 𝑡𝑜 𝐸𝑃𝑠 + 𝐶ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝐷𝑒𝑐𝑎𝑛𝑡 𝑃𝑜𝑛𝑑 𝑉𝑜𝑙𝑢𝑚𝑒

718

OPERATIONAL LIQUOR MANAGEMENT, DRAINAGE IMPROVEMENTS AND INFLUENCE ON BUTTRESS DESIGN Due to the difficulty in estimating the processes involved between liquor entering the TSF as part of the slurry and reaching the decant pond, the supernatant runoff is estimated as a proportion of the slurry liquor inflow. Observed data was used to balance the above equation to estimate the proportion of the slurry liquor inflow that results in supernatant runoff. The key variables that affect the TSF balance and that have been used as variables in the calibration exercise are the tailings rainfall runoff coefficient and the pan evaporation factor. An iterative process was used to calibrate these key variables while considering the TSF balance and historic data. Calibration of a water balance model requires pond measurements at sufficient frequency. BHPOD measures pond sizes monthly, utilising drone imagery. Satellite remote sensing products may be used as the alternative data set in water balance model calibration when in-situ monitoring data is not available (Liu and Ludlow, 2023). Model calibration results using a monthly pond size measurement step (drone surveillance) and utilising Sentinnel-2 satellite imagery (11-day frequency) are shown in Figures 4 and 5, respectively. Figure 5 demonstrates model performance accurately simulating significant rain events that occurred in January 2022 (70 mm) and October 2022 (140 mm). Another key lesson learned is the importance of continually maintaining a GoldSim model. Calibration and revalidation of the model are carried out annually.

Figure 4: GoldSim model calibration results – total volume (EP and TSF)

719

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 5: GoldSim model calibration results with sentinel imagery – total volume (EP and TSF)

Drainage improvements Understanding the pore water pressure distribution(s) within the deposited tailings is critical for determining embankment stability. The pore pressures within the tailings behind the perimeter embankment not only influence the design geometry of the embankment, but also have a major impact on its stability. Pore water pressure within the tailings is dependent upon several factors including: • tailings permeability; • foundation drainage conditions; • location and size of the decant pond; • depositional methodology; and • impoundment operation. Even with the sophisticated numerical modelling tools that are currently available in the geotechnical engineering practice, it is difficult to accurately predict pore water pressures throughout the operational life of a TSF. It is probable that pore pressure conditions within the tailings will fluctuate with variances in deposition or weather conditions. The observational approach is applied in the stability assessment; however, consideration is given to key design features of the TSF. Key differences in the design of TSF5 compared with the previous facilities include the following: • Heel drain: sand drain extends 50 m into the impoundment from the upstream toe over the entire perimeter.

720

OPERATIONAL LIQUOR MANAGEMENT, DRAINAGE IMPROVEMENTS AND INFLUENCE ON BUTTRESS DESIGN • Toe drain: collection system on downstream side of starter embankment to collect seepage. • Underdrain and decant: Y-decant arm from causeway to assist in positioning of the pond. • Increased TSF surface area: larger surface area with greater offset of pond from embankments. A comparison of TSF4 and TSF5 cross-sections are shown in Figures 6 and 7 respectively. The key component of differentiation is the described heel drain constructed of freely draining sand dune material. The benefit of this design feature is described in the subsequent section, detailing the improvement in pore pressures within the embankment and benefits to stability.

Figure 6: TSF4 typical cross-section

Figure 7: TSF5 typical cross-section with heel-drain inclusion

Further differences in TSF5 and TSF4 are detailed in Figure 8, and include the decant causeway design and TSF surface area. TSF5 has seen an increase in surface area by approximately 50%, ensuring pond sizes remains a suitable distance from the embankments. In addition, the introduction of decant “pool arms” has seen better pond orientation control. Key features such as these contribute to the overall liquor balance and management of decant ponds, which has a positive impact on embankment stability.

721

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 8: TSF5 (left) decant with “pool arms” and TSF4 decant (right)

TSF stability considerations TSF 5 introduced several design features that have assisted in the drainage performance of the TSF. Conceptual designs considered the benefit of the heel drain, with Figure 9 detailing a 100% hydrostatic condition (Phreatic Surface 1) and a 75% hydrostatic condition (Phreatic Surface 2), driven by the downward drainage expected to occur with the heel drain. VWPs were installed and have showed the existence of a downward flow, but that downward flow has not been considered in stability calculations so far. Instead a conservative hydrostatic pore pressure profile has been assumed.

Figure 9: TSF5 conceptual design of predice phreatic conditions

722

OPERATIONAL LIQUOR MANAGEMENT, DRAINAGE IMPROVEMENTS AND INFLUENCE ON BUTTRESS DESIGN Recent industry guidelines (ANCOLD, 2019) and standards (GISTM, 2020) have identified the need to identify and address brittle failure modes with conservative design criteria, independent of trigger mechanisms. This has increased efforts within the industry to identify the location of the critical state line of each tailings in the 𝑒~log𝑝′ plane. The critical state line is required so that a state parameter can be determined, as long as the void ratio 𝑒 and mean effective stress 𝑝′ of the tailings are known. That state parameter may then be related to strengths for peak and post-liquefaction conditions, essential to conduct stability assessments for many storages. BHPOD has undertaken a comprehensive program of experimental work to define the critical state line of the tailings in in the 𝑒~log𝑝′ plane, with particular focus on improving the reliability of the undrained strength ratio (USR) for use in post-liquefaction stability assessments. Typically, for a variety of tailings across the world, the USRs relevant to post-liquefaction scenarios may range from 0.06 to 0.20. The second significant figure in the USR is therefore important. It is concerning that the vertical position of the critical state line in the 𝑒~log𝑝′ plane can vary by as much as 0.04 within results generated and interpretations made from one laboratory to the next. This was demonstrated in the round robin of Reid et al. (2023) carried out as part of the TALLIQ (Tailings Liquefaction) project. This will impose a difference of 0.04 in the state parameter. Since state parameter is roughly linearly proportional to USR (e.g., using Figure 6.47 in Jefferies and Been, 2016), it may also result in a 0.04 difference in the estimated USR. Examples of post-liquefaction stability modelling are shown, in which the section conservatively assumes 100% hydrostatic conditions for TSF5, representing a worst case upper bound, reflected in Figure 10a. A sensitivity analysis was performed to explore the relative importance of key input assumptions. The sensitivity analysis modelled the following cases and determined the buttress size necessary to attain a FoS of 1.1 for each case. The cases included the following: 1. TSF5 cross-section with phreatic level assumed to be 100% hydrostatic. 2. TSF5 cross-section with phreatic level assumed to be worse (i.e., TSF4 performance). 3. TSF5 cross-section with phreatic level per case (b) and adjusted tailings USR to attain FoS=1.1. Case (c) revealed that an increased phreatic level, as detailed in Figures 10b and 10c, required an increased tailings post-peak USR of 10% to yield the same result as that of Case (a). The improved drainage condition shown in Figure 10a resulted in the same FoS as the case in Figure 10c. However it was reliant on an increase of the post-peak USR of 10%. Recognising the findings from Reid et al. (2023) and the potential for USR variability, consideration should be given to this uncertainty, and other key design parameters that should be better understood, to ensure sufficient conservatism is applied.

723

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

(a)

(b)

(c) Figure 10: Conceptual stability assessments varying phreatic level and tailings post-peak USR

Conclusion This paper highlights the importance of accurate liquor measurement and model calibration to accurately predict liquor inventory so as to ensure pond sizes can be managed within acceptable limits. In addition, this paper highlights the importance of minimising pore pressures within embankments, as this does have a measurable impact on the tailings post-peak USR that is defined. Reducing pore-pressures or further understanding the hydrostatic conditions in a TSF could translate to a benefit of at least a 10% strength improvement for the example given. CSL elevations of 0.04 for void ratio were determined as part of the TALLIQ program, demonstrating uncertainty specifically when defining a post-peak USR to a second decimal place. Engineers and operators should not underestimate the importance of appropriately investigating and characterising the pore pressure profile within a facility, even when applying deterministic limit equilibrium stability modelling for a post-peak scenario.

724

OPERATIONAL LIQUOR MANAGEMENT, DRAINAGE IMPROVEMENTS AND INFLUENCE ON BUTTRESS DESIGN

Acknowledgements The author acknowledges BHP Olympic Dam Pty Ltd for allowing the sharing of information and lessons learned. The author thanks Professor Adrian Russell from the University of New South Wales (UNSW), Independent Technical Reviewer and Juan “Pepe” Moreno from SRK Consulting and Engineer of Record at BHP Olympic Dam for their continued technical leadership.

References Australian National Committee on Large Dams (ANCOLD). 2012. Guidelines on Tailings Dams: Planning, Design, Construction, Operation and Closure. Australian National Committee on Large Dams. Australian National Committee on Large Dams (ANCOLD). 2019. Guidelines on Tailings Dams: Planning, Design, Construction, Operation and Closure. Australian National Committee on Large Dams. Been, K. and Jefferies, M. 2016. Soil Liquefaction: A Critical State Approach, 1st ed. New York: Taylor and Francis. Global Tailings Review. 2020. Global Industry standard on Tailings Management (GISTM) (online). Accessed April 2023 at https://globaltailingsreview.org/global-industry-standard/ Liu, M. and Ludlow, W. 2023. Calibration of the water balance model for an inactive tailings storage facility using Sentinnel-2 satellite imagery. In Proceedings of the Mine Waste and Tailings Conference 2023. Brisbane, Queensland: 644–650. Reid, D., Fourie, A., Ayala, J., Dickinson, S., Ochoa-Cornejo, F., Fanni, R., Garfias, J., Viana Da Fonseca, A., Ghafgha, M., Ovalle, C., Riemer, M., Rismanchian, A., Olivera, R., and Suazo, G. 2023. Results of a critical state line testing round robin programme. Géotechnique 73(1): 89–90.

725

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

726

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Coke as a Filter in Oil Sands Mining Dams Pathma Wedage, WSP E & I Canada Ltd, Canada Christopher Fortier, Suncor Energy Inc., Canada Anthony Burnett, Suncor Energy Inc., Canada Weidong Li, WSP E & I Canada Ltd, Canada

Abstract A geotechnical study was undertaken to evaluate the feasibility of using coke, a coarse-grained and lightweight by-product of oil sands upgrading, as the filter material in oil sands mining dams. The study involved a review of available empirical filter criteria, laboratory experiments, and field placement trials. Samples of coke collected from various locations were used to assess the potential impact of construction induced stresses on its gradation and to compare its gradation with existing empirical filter criteria. Subsequently, laboratory filter tests were conducted on a screened coke product meeting the empirical criteria to confirm the suitability of the screened coke product as the filter for the dam core. The laboratory test program consisted of a series of slot tests to assess the compatibility of the screened coke filter against the core, and internal stability tests to confirm that the screened coke remains internally stable under more severe hydraulic gradients than would generally be expected in practice. A field trial was undertaken to determine a practical and efficient placement method to follow during the construction stage. This paper presents the steps that may be followed when evaluating potential filter materials for dams, with particular emphasis on dams in the oil sands mining industry.

Introduction Two commonly adopted methods for constructing oil sands mining dams with overburden and oil sands wastes are dams with upstream sand beaches, and dams with internal filter drain systems. In the second category, the designs have relied on a high-specification central core for structural stability and seepage control and a sand filter drain system for defense against piping and to achieve a low piezometric level on the downstream side of the filter (Wedage et al., 2019). Either tailings sand or naturally occurring sands of Pleistocene origin are currently used as the filter sand. Due to the high demand for, and the scarcity of, suitable sands, operators are constantly looking for alternative design methods or for alternative filter materials. Coke has been used as the filter in a limited number of dams in the past. However, documentation

727

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA to validate how the filter criteria were assessed against the oil sands core is not available in the public domain. This paper was written with the objective of sharing the geotechnical characteristics of coke and to provide a method for evaluating coke as a filter. The coke produced at Suncor is a coarse-grained and lightweight by-product of oil sands upgrading. A photo of a stockpile of raw coke, as directly produced, is presented in Figure 1. Raw coke contains numerous cobble-sized and gravel-sized particles, and is segregated upon production. To reduce segregation and provide a more homogeneous product, Suncor crushes the coke to less than about 75 mm particle size (75 mm minus). The gradation of Suncor 75 mm minus coke would classify as well-graded sandy gravel. The Standard Proctor Maximum Dry Density coke varies from about 950 to 1,025 kg/m3. The Optimum Moisture Content varies from about 15% to 25%, with the moisture content of the samples collected from various locations in the range of 5 to 10%. The specific gravity of coke is close to 1.35. Enclosed voids in the larger coke particles make the particle more buoyant and therefore the specific gravity is lower than common soils. The fines content of the 75 mm minus coke produced at Suncor is generally less than 5%, and the hydraulic conductivity is at least an order of magnitude higher than the naturally occurring sands and tailings sands that are typically sourced for filter construction. Coke displays a relatively high frictional strength compared to sand. However, the 75 mm minus coke does not meet geotechnical filter gradation selection criteria for the high-specification dam core, and it exhibits some segregation characteristics during handling and placing with conventional construction equipment. Therefore, despite the favourable hydraulic conductivity and strength characteristics, the 75 mm minus coke is not considered suitable for filter construction from a filter compatibility (retainability) and constructability perspective.

Figure 1: Raw coke as produced at Suncor A filter gradation envelope was developed to satisfy the existing empirical criteria on retainability, hydraulic conductivity, and segregation considering coke as a filter for the high-specification dam core

728

COKE AS A FILTER IN OIL SANDS MINING DAMS materials. Table 1 summarizes the main constraints, and Figure 2 presents the recommended filter gradation envelope. Samples of coke falling within the recommended gradation band have been tested during this study to confirm the validity of the empirical criteria for the specific project application. Table 1: Filter gradation limits for oil sands core based on empirical criteria Reference point in gradation plot

Mass passing (%)

Particle size (mm)

Retainability

A

15

0.25 0.7) or the shear strain deformation level experienced by these materials, residual undrained shear strength was assigned to tailings and lacustrine clays after the seismic event. In the postseismic phase, displacements around 6 m at the toe and 1 m on the shoulder of the south slope were obtained in the horizontal direction. Further, the shear strains increased considerably in the post-seismic phase, indicating a rotational failure of the south and north slopes. The equilibrium could not be reached in FLAC because of mesh limitations to reach large deformations, but these results indicate an ongoing landslide. In this study, two return periods were considered. The analyses corresponding to a 475-year return period show ongoing failure of the north and south slopes, and the model did not stabilize in FLAC. Moreover, it can be observed that the summit is affected by the ongoing south slope deformations. On the other hand, for a 100-year return period, results show a local movement of the south slope, and displacements and deformations stabilize in the post-seismic phase. However, there is still an ongoing landslide on the north slope (~5 m height). The summit and central zones of the TSF are not affected by the south slope displacements and deformations ( -0.05

1

0.05

0.01

60%

0.06

4

2.3

3.7

100%

2

0.06

0.03

89%

0.09

2

1.0

1.0

99%

3

0.00

0.04

89%

0.04

10

8.0

6.2

97%

4

0.06

0.03

38%

0.09

1

1.0

1.6

100%

5

0.01

0.02

56%

0.03

8

6.6

6.6

98%

6

-0.02

0.02

37%

0.00

12

10.0

10.0

98%

7

-0.01

0.04

84%

0.03

11

8.8

6.7

96%

8

0.01

0.04

86%

0.05

9

6.9

5.1

93%

9

0.04

0.03

66%

0.07

5

3.3

2.9

98%

10

0.05

0.02

83%

0.07

3

2.2

3.0

100%

11

0.03

0.04

81%

0.07

7

4.4

3.4

96%

12

0.04

0.02

61%

0.06

6

3.5

4.0

99%

Geometric product Determining the geometric product was a recommendation developed from the ITRB and oversight committee and aimed to provide a geometric comparison of the profiles to relatively rank the major contributors to slope (in)stability. The selected geometric product was as follows:

𝐺𝑒𝑜𝑚𝑒𝑡𝑟𝑖𝑐 𝑃𝑟𝑜𝑑𝑢𝑐𝑡 = 𝑆𝑙𝑜𝑝𝑒 𝐻𝑒𝑖𝑔ℎ𝑡 ∗ 𝑆𝑙𝑜𝑝𝑒 𝐴𝑛𝑔𝑙𝑒 ∗

𝑆𝑎𝑡𝑢𝑟𝑎𝑡𝑒𝑑 𝐻𝑒𝑖𝑔ℎ𝑡 𝑇𝑜𝑡𝑎𝑙 𝑆𝑙𝑜𝑝𝑒 𝐻𝑒𝑖𝑔ℎ𝑡

These values were estimated using topographic data and pore pressure dissipation test data taken from CPT probes and piezometers in each profile. This metric is a straightforward approach to qualitatively evaluating the risk present at each profile. The scaling was set again for 1 to 10, with 1 being the lowest and 10 being the highest relative value. Table 4 shows the range of values for the Geometric Product. Note that uncertainty was not included here because this value was a straight metric of verified conditions.

1236

CASE STUDY: APPROACH TO DETERMINING THE RISK MITIGATION PRIORITY OF A HISTORIC TSF IN NORTH AMERICA Table 4: Geometric product Profile

Geometric product

Ranking

Scaling by best estimate

Straight rank

1

12.4

5

5.6

6

2

13.2

4

6.1

4

3

18.8

2

8.9

3

4

3.2

12

1.0

10

5

5.3

10

2.1

11

6

4.9

11

1.9

7

7

10.7

7

4.8

9

8

8.8

9

3.8

12

9

11.0

6

5.0

8

10

18.0

3

8.5

2

11

21.1

1

10.0

1

12

10.5

8

4.7

5

Undrained Factor of Safety at Residual Strength The Undrained Factor of Safety at Residual Strength was calculated to provide a direct metric of the factor of safety for the estimated geometry and material stratigraphy in a post-liquefaction condition for materials that may be subject to liquefaction (cyclic or static) based on the results of CPT testing. This analysis was repeated using yield and drained strengths for sensitivity, and similar rankings were found. These factors of safety were further calculated under non-conservative assumptions because these data were reflective of an early stage of investigation. In this calculation, “nonconservative” referred to an assumed desaturation and non-liquefaction of un-characterized materials approaching the facility’s downstream extents. It was found that under both conditions, many of the profiles were below industry guidelines (CDA, 2019). Scaling was done from 1 to 10, with 1 being the lowest FOS and 10 being an FOS of 1.2 (the minimum FOS recommended by the CDA for post-liquefaction conditions). Table 5 shows the range of values for the Undrained Factor of Safety at Residual Strength.

1237

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 5: Factor of safety at residual strength

Profile

Factor of Safety

Uncertainty metric (nonconservative FOS)

Uncertainty parameter

Ranking

Scaling by best estimate

Scaling by uncertainty

1

0.26

0.54

0.337

2

1.3

1.8

2

0.24

0.8

0.675

1

1.0

4.2

3

0.24

0.46

0.265

1

1.0

1.0

4

0.59

1.42

1.000

9

7.1

10.0

5

0.56

0.81

0.301

8

6.5

4.3

6

0.76

0.92

0.193

11

10.0

5.3

7

0.32

0.5

0.217

4

2.9

1.4

8

0.32

0.68

0.434

4

2.9

3.1

9

0.52

0.72

0.241

6

7.8

3.4

10

0.61

0.74

0.157

10

10.0

3.6

11

0.42

0.54

0.145

5

6.6

1.8

12

0.53

0.65

0.145

7

10.0

2.8

Presence of a pond or fresh tailings in the impoundment The presence of a pond or fresh tailings in the impoundment metric was a way of quantifying the potential geotechnical stability risk posed by the presence of an ongoing source of saturation in the profile and an increase to the consequence of failure if a failure were to occur. A simple “low”=1, “medium”=2, and “high”=3 value was applied to each section, with high being the ongoing presence of impounded water above the profile, medium being recent deposition or water storage, and low being a decommissioned cell with no water storage in the past several years. Uncertainty was not considered because this metric was directly observable. Table 6 shows the range of values for the Presence of a Pond or Fresh Tailings in the Impoundment.

1238

CASE STUDY: APPROACH TO DETERMINING THE RISK MITIGATION PRIORITY OF A HISTORIC TSF IN NORTH AMERICA

Table 6: Presence of water

Prof.

Water/ recent deposition

Ranking

Scaling by best estimate

Prof.

Water/ recent deposition

Ranking

Scaling by best estimate

1

2

5

5.5

7

3

1

1

2

2

5

5.5

8

1

10

10

3

3

1

1

9

1

10

10

4

2

5

5.5

10

2

5

5.5

5

2

5

5.5

11

1

10

10

6

2

5

5.5

12

1

10

10

Weighting factors Since these criteria cannot be meaningfully combined into a single profile, each metric used a scaling, and then, a final weighting was applied to the metrics. The final weighting is intended to sum to a value of 10 to provide a normalized contribution. The major contributors to the risk profile were the Factor of Safety and the Liquefaction Potential. Since the first three metrics were all related to the liquefaction potential, metrics 2 and 3 were lowered so that those metrics were not over-emphasized. Table 7: Weighting factors #1 - Avg. CRR

2.2

#2 - Avg. Shear Wave Vel.

0.6

#3 - Avg. State Parameter

0.6

#4 – Geometric Product

1.1

#5 - FOS residual

3.3

#6 - Water at Crest/Recent Deposition

2.2

Results and limitations Combined rankings Table 8 shows the combined rankings that were calculated with and without weighting factors to show the general risk profile of each section. The highest rank indicates the lowest priority for mitigation, while the lowest rank indicates the highest mitigation priority. However, the mitigation priority did not necessarily consider which areas could be mitigated the most readily and which could not.

1239

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 8: Combined rankings No weighting factors

With weighting factors

Profile

Scaling rank sum

Scaling overall rank

Scaling weighted rank sum

Scaling weighted rank

1

23.96

2

34.30

1

2

34.13

5

50.41

5

3

32.78

4

37.67

2

4

21.87

1

42.48

4

5

37.14

8

57.81

6

6

34.96

6

66.06

8

7

32.40

3

40.66

3

8

41.87

9

66.40

9

9

42.34

10

77.31

12

10

35.33

7

67.44

10

11

43.64

12

65.58

7

12

43.30

11

73.69

11

Uncertainty Due to lack of information in the complex, uncertainty was a key factor for measuring the validity of each metric and providing an uncertainty-corrected value. Uncertainty was calculated using the standard deviation of each given metric, and the normalized standard deviation was applied across scaling to provide both a standard scaling and an uncertainty scaling. Table 9 shows the uncertainty ranking that was used as an indicator of where to target additional site investigation and engineering analyses. The uncertainty ranking aims to identify areas where additional work may have a greater impact on reducing uncertainty and may reduce remediation efforts. The highest rank indicates the lowest opportunity provided by uncertainty reduction, while the lowest rank indicates the highest opportunity provided by uncertainty reduction.

1240

CASE STUDY: APPROACH TO DETERMINING THE RISK MITIGATION PRIORITY OF A HISTORIC TSF IN NORTH AMERICA Table 9: Uncertainty factors No weighting factors

With weighting factors

Profile

Uncertainty rank sum

Uncertainty overall rank

Uncertainty weighted rank sum

Uncertainty weighted overall rank

1

17.84

3

29.03

6

2

23.18

11

41.21

11

3

21.73

6

26.94

5

4

23.00

10

53.44

12

5

27.61

12

39.81

10

6

22.41

8

37.55

9

7

22.20

7

25.84

4

8

15.97

2

23.05

3

9

15.03

1

19.90

1

10

20.73

5

35.82

8

11

19.14

4

22.61

2

12

22.98

9

34.90

7

To improve the process of targeted investigation, an additional sensitivity metric was prepared to show the potential change in the overall ranking with a change in the level of uncertainty. Table 10 shows that Profiles 8,9, and 11 were most sensitive to changes in uncertainty.

1241

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 10: Uncertainty sensitivity % Change with uncertainty

Rank change with uncertainty

Profile

Scaling rank sum

Scaling weighted rank sum

Scaling overall rank

Scaling weighted rank

1

25.6%

15.4%

1

5

2

32.1%

18.2%

6

6

3

33.7%

28.5%

2

3

4

5.2%

25.8%

9

8

5

25.7%

31.1%

4

4

6

35.9%

43.2%

2

1

7

31.5%

36.5%

4

1

8

61.9%

65.3%

7

6

9

64.5%

74.3%

9

11

10

41.3%

46.9%

2

2

11

56.1%

65.5%

8

5

12

46.9%

52.6%

2

4

Final rankings The final rankings for both the mitigation and the investigation priority were prepared. Following this analysis, the highest-ranking mitigation priorities were combined into construction phases for remediation. Each construction phase was scheduled for approximately one year. During this time, additional site investigations were also planned to begin the process of reducing uncertainties from the analyses.

1242

CASE STUDY: APPROACH TO DETERMINING THE RISK MITIGATION PRIORITY OF A HISTORIC TSF IN NORTH AMERICA

Table 11: Final mitigation priority

Table 12: Final investigation priority

Profile

Priority number

Profile

Uncertainty rank

1

1

9

1

3

2

11

2

7

3

8

3

4

4

7

4

2

5

3

5

5

6

1

6

6

7

12

7

8

8

10

8

10

9

6

9

11

10

5

10

9

11

2

11

12

12

4

12

Conclusions The methodology presented herein can be used to prioritize risk mitigation for a tailings storage facility where little information is available and where mitigation should begin in parallel with additional site investigation due to the potential consequences of a hypothetical failure scenario. The specific metrics and applications of this methodology may vary across sites based on site-specific risks that are identified by the engineer.

Acknowledgments Thank you to Greg Maris and José Luis Morales de la Cruz for their contributions to these analyses. An additional thank you to Dr. Peter Robertson for his contributions as third-party support in review of the CPT data and development of the Geometric Product considered for these facilities.

References Andrus, R.D. and K.H. Stokoe II. 2000. Liquefaction resistance of soils from shear-wave velocity. Journal of Geotechnical and Geoenvironmental Engineering 126(11). Canadian Dam Association (CDA). 2019. Application of Dam Safety Guidelines to Mining Dams. Technical Bulletin. Robertson, P.K. 2009. Interpretation of cone penetration tests – a unified approach. Canadian Geotechnical Journal 46: 1337–1355.

1243

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

1244

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Credible Failure Modes – Summary of 2021 and 2023 Workshops Andy Small, Klohn Crippen Berger, Canada Angela Küpper, BGC Engineering, Canada Tamara Johndrow, Freeport-McMoRan Inc., USA Mohammad (Mamun) Al-Mamun, Tetra Tech Canada Inc., Canada

Abstract This paper provides a discussion on the credible failure modes concept that was included in the Global Industry Standard on Tailings Management (GISTM) issued in August 2020 (GTR, 2020). The interpretation of the credible failure mode concept has been the subject of much discussion and a group of industry leaders convened two workshops (November 2021 and April 2023) to explore the concept and the interpretation of credible failure modes in practice. The key messages from those workshops were as follows: • There are two interpretations when determining credible failure modes; one is based on whether a failure mode is physically possible and the other is based on consideration of negligibility that needs to be clearly defined. • Both interpretations embrace risk assessments and call for a thorough understanding, evaluation, and management of risks associated with potential failure modes. • Both interpretations satisfy the intent of the GISTM. This is an evolving topic, and as the tailings storage facility (TSF) safety community continues to advance the credible failure modes concept in practice, perspectives presented in this paper may also evolve.

Introduction The Global Industry Standard on Tailings Management, issued in August 2020 (GTR, 2020), refers to the term “credible failure mode.” In May 2021, the International Council on Mining and Metals (ICMM) issued the Tailings Management Good Practice Guide (GPG), which provides guidance on how credible failure modes can be determined for a TSF. Although those two documents included a definition of credible failure modes, there is a lack of consensus in the TSF safety community on how best to interpret credible failure modes. A group of industry leaders convened a workshop to discuss interpretations and applications of

1245

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA credible failure modes. The workshop was held on 7 November 2021 as part of the 2021 Tailings and Mine Waste Conference in Banff, Alberta, Canada. A follow-up workshop was convened on 30 April 2023 as part of the 2023 Canadian Institute of Mining Conference in Montreal, Quebec, Canada. The 2021 workshop was co-chaired by Andy Small of Klohn Crippen Berger and Mohammad (Mamun) Al-Mamun of Tetra Tech. Andy chaired the 2023 workshop. A diverse team of professionals representing mine owners, consultants, risk assessment specialists, and regulators was assembled to deliver these workshops, as listed in Table 1. In addition, Dr. Norbert Morgenstern, distinguished university professor (emeritus), University of Alberta (Canada) and consulting engineer, was consulted throughout the planning and development of the 2021 workshop. Table 1: List of key members involved the workshops Name

Affiliation

Workshop(s) involved

Andy Small

KCB, Canada, Co-Chair of 2021 Workshop and Chair of 2023 Workshop

2021 and 2023

Mohammad Al Mamun

Tetra Tech, Canada, Co-Chair of 2021 Workshop

2021

Angela Kupper

BGC Engineering, Canada

2021 and 2023

Tamara Johndrow

Freeport-McMoRan Inc., United States

2021 and 2023

Chris Anderson

Teck, Canada

2021 and 2023

Brett Byler

Newmont, United States

2021 and 2023

Imran Gillani

Rio Tinto, Canada

2021 and 2023

Chad LePoudre

BHP, Canada

2021

Lucy Philip

Stantec, Canada

2021

Rob Schryburt

Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry, Canada

2021 and 2023

Anton Bain

Anglo American, South Africa

2021

Caius Priscu

Independent (formerly Anglo American), Canada

2021 and 2023

Christina Winckler

Anglo American, United States

2021

Yi Zhu

Rio Tinto, United States

2021

Bruce Englesman

SRK, South Africa

2023

Dean Durkee

Gannett Fleming, United States

2023

The 2021 and 2023 workshops were attended by 75 and 40 participants, respectively. Both workshops had designated time for open discussion for the participants to share their thoughts and views. The workshop participants heard perspectives from a member of the Global Tailings Review Expert Panel, the Chair of

1246

CREDIBLE FAILURE MODES – SUMMARY OF 2021 AND 2023 WORKSHOPS ICMM’s writing subgroup for the ICMM Tailings Management Good Practice Guide, mine owners, consultants, and Canadian dam safety regulators. A key point that was agreed to among all workshop team members was that the main objective for the design and governance process throughout the lifecycle of the TSF is to minimize risks and manage potentially catastrophic failure modes using the ALARP (as low as reasonably practicable) approach throughout the lifecycle of the TSF.

GISTM perspectives on credible failure modes The GISTM was prepared by the Global Tailings Review Expert Panel and its Independent Chair. It was co-convened by the ICMM, United Nations Environment Program (UNEP), and the Principles for Responsible Investment (PRI). The GISTM provided the following definition for credible failure modes: “Credible Failure Modes / Scenarios: Refers to technically feasible failure mechanisms given the materials present in the structure and its foundation, the properties of these materials, the configuration of the structure, drainage conditions and surface water control at the facility, throughout its lifecycle. Credible failure modes can and do typically vary during the lifecycle of the facility as the conditions vary. A facility that is appropriately designed and operated considers all of these credible failure modes and includes sufficient resilience against each. Different failure modes will result in different failure scenarios. Credible catastrophic failure modes do not exist for all tailings facilities. The term “credible failure mode” is not associated with a probability of this event occurring and having credible failure modes is not a reflection of facility safety.” During the credible failure modes workshops, Angela Kupper (BGC Engineering), a member of the Global Tailings Review Expert Panel, provided her perspectives on credible failure modes and presented the flow chart shown in Figure 1.

Figure 1: Flow chart to discern credible failure modes

1247

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The following key points were noted by Angela in her presentation at the 2021 and 2023 Workshops: • The Global Tailings Review Expert Panel’s Terms of Reference1 mandated a consequence-based approach for the GISTM2 that was to be used to inform the requirements for design and emergency planning. It is noted that the Consequence Classification approach developed by the water dam industry decades ago is based on the same fundamental concept adopted for risk assessments where consequence and probability of failure are evaluated independently. • All credible failure modes that can lead to a flow failure should be considered for emergency response planning (Requirement 13.1). • The consequence classification for a dam should be based on all credible failure modes (Requirement 4.1 of the GISTM). • Design criteria for new facilities should consider adopting Extreme consequence classification external loading, considering that during the lifecycle of the facility the consequence of a potential failure is likely to increase. This is because it is common for TSFs to be raised beyond the initial plans and there could be development downstream of the TSF. Later in its lifecycle, the consequences of failure could decrease if the pond is removed and if there is sufficient drainage of the deposit to limit the potential for flow failures. If lesser external loading criteria were selected, the Operator should demonstrate the ability to upgrade the design to satisfy Extreme external loading in the future (Requirement 4.2 of the GISTM), if required. For existing facilities, the GISTM requires managing risks according to the principles of ALARP (Requirements 4.7 and 5.4 of the GISTM). • As noted in the definition of credible failure modes above, credible failure modes are based on their technical feasibility. The concept of considering all technically feasible failure modes as credible failure modes was intended to avoid eliminating a potential failure mode too early in the assessment process. It is important to allow time to obtain and analyze sufficient data to support an adequate assessment of each potential failure mode and establish a defendable determination of its level of probability of occurrence. • The credible failure mode definition in the GISTM includes the following sentence, “The term “credible failure mode” is not associated with a probability of this event occurring,” which has led to significant discussion. The intent of the sentence was to clarify that considering a failure mode as

1 From https://globaltailingsreview.org/wp-content/uploads/2019/06/190604_GTR_governance-and-scope.pdf 2 “Building on existing global best practices, the overall scope of the review will be determined by the need to inform the development of a standard that addresses, but is not limited to, the following: A global and transparent consequence-based TSF classification system with appropriate requirements for each level of classification …”

1248

CREDIBLE FAILURE MODES – SUMMARY OF 2021 AND 2023 WORKSHOPS being a credible failure mode is not a statement about its probability of occurrence – its probability could be anywhere from negligible to very high. The determination of the probability of failure is the subsequent step in the assessment process. The process of determining the likelihood of failure of each credible failure mode involves collecting and reviewing the available information, in some cases conducting additional investigations, and discussing the failure mode and scenario, before being able to estimate the probability of failure with sound and reasoned engineering judgement and to support the decision with the data and the results of the analyses. • If a potential failure mode needs action, monitoring, human judgement, procedures to minimize human error, maintenance of field conditions, etc. to manage the likelihood of failure, it is because it is likely credible. • The key goal of the GISTM approach was that there be a thorough consideration of all technically feasible (physically possible) failure modes and that for all of these failure modes, the risks be demonstrated to be low or acceptable and be managed to ALARP. In some cases, these analyses might determine that the risks for the current condition are already negligible and for other cases, some level of action is required to lower the risk level.

Tailings management good practice guide perspectives on credible failure modes Tamara Johndrow (Freeport-McMoRan Inc.), Chair of the Tailings Management Working Group of ICMM, which prepared the Tailings Management Good Practice Guide (GPG), provided her perspectives at the workshop: • The GISTM did not differentiate between Credible Failure Modes and Scenarios, but the GPG provided a distinction between a credible failure mode and scenario as: • Credible failure mode = credible mechanism + credible initiating event + credible failure process (each element needs to be credible for the failure mode to be credible). • Credible failure scenario = credible failure mode + credible consequences (each element needs to be credible for the failure scenario to be credible). •

The GPG proposes beginning the design process for new TSFs assuming “Extreme” loading design criteria (similar to GISTM requirement 4.2 with the criteria in the GISTM’s annex) and proposes being consistent with the principles of ALARP risk for existing facilities.

• The determination of credible failure modes is part of the broader risk management process for a TSF; risk management is embedded in the tailings management system. • Assessment of credibility, uncertainty, and risk occurs throughout the life cycle of a TSF.

1249

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Credible failure modes are defined per GISTM. Potentially credible failure modes may be ruled out categorically during initial screening or if additional investigations and analyses determine from a practical perspective that a failure mode is non-credible. • For those failure modes that are technically feasible (or physically possible), further assessment considering engineering judgement, investigation, and analyses can be undertaken to determine if the failure mode should be considered credible from a practical perspective. • For the purpose of emergency planning, credible failure modes with flow or slumping scenarios that could lead to catastrophic damage downstream should be considered. The definition of catastrophic is provided by the GPG as follows: “... material disruption to social, environmental and local economic systems…Catastrophic events typically involve numerous adverse impacts, at different scales and over different timeframes, including loss of life, damage to physical infrastructure or natural assets, and disruption to lives, livelihoods and social order … Catastrophic failures exceed the capacity of affected people to cope using their own resources, triggering the need for outside assistance in emergency response, restoration and recovery efforts” (ICMM, 2021). • The determination of whether a failure mode is credible or not needs to be reviewed on a regular basis. Thus, the Tailings Management GPG acknowledges that specifying a minimum threshold of probability of occurrence is helpful in determining whether a failure mode is credible or not, but not essential. The goal of the GISTM is for Operators to conduct a thorough consideration of all failure modes and that the risks be managed to an acceptable level. This goal is met by the interpretation provided in the Tailings Management GPG.

Definitions of the term “credible” The term “credible” is not new to the dam safety industry. It is used by the Federal Energy Regulatory Commission (FERC) in relation to the failure modes of water/hydropower dams. The word “credible” has the following meanings in dictionaries: •

Merriam-Webster: offering reasonable grounds for being believed or trusted.

•

Cambridge: able to be believed or trusted. This choice of word is where some of the confusion has arisen in the use of the term credible failure

modes. A dam safety program should not be solely based on belief or trust (per the Merriam-Webster definition) but supported by detailed analyses and engineering judgement that are supported by quality data and analyses. This is an evolving topic, and it is possible that as the tailings storage facility (TSF) safety

1250

CREDIBLE FAILURE MODES – SUMMARY OF 2021 AND 2023 WORKSHOPS community continues to advance this topic, consideration of the credible failure mode concept will also evolve.

FERC perspectives of credible failure mode concept Dean Durkee (Gannett Fleming) is a geotechnical and dam engineering consultant and dam safety risk assessment specialist, and currently serves as Chair of the United States Society on Dams (USSD), Risk Training Subcommittee. As part of the 2023 workshop he shared the following perspectives related to credible failure modes from his experience facilitating risk assessments for hydropower facilities under the jurisdiction of the Federal Energy Regulatory Commission (FERC) in the United States: •

Risk assessments are a valuable tool to identify, characterize, and manage uncertainty and to prioritize dam safety actions relating to monitoring and surveillance, investigation, analysis, design, construction, operation, and modification of dams.

•

FERC’s Engineering Guidelines for evaluating the risk at hydropower facilities (Chapters 17 and 18) is comprised of two parts, the potential failure mode analysis (PFMA) and semi-quantitative risk analysis (SQRA). This approach is often utilized to satisfy the requirements of GISTM.

•

FERCs Chapter 18 states: “For most projects, the results of the PFMA will generate a large number of potential failure modes. In most cases, not all of these potential failure modes need to be evaluated in a risk analysis. Only those potential failure modes that substantially contribute to the risk profile of the project need to be evaluated in a risk analysis. This requires screening the credible potential failure modes identified in the PFMA. The goal of screening is to identify those potential failure modes that need to go into the risk analysis process versus those that do not, which allows the participants to focus their efforts on the potential failure modes that contribute to the risk profile.” Thus, as part of FERCs PFMA process, failure modes are brainstormed and put through a structured screening process referred to by FERC as “Initial Screening.” During initial screening, each PFM is assessed based on physical possibility and negligibility. A PFM is considered credible if it is “considered to be physically possible and the likelihood of the potential failure mode is not considered so remote as to be clearly negligible (FERC, 2021).” In other words, FERC allows the consideration of negligibility in the determination of a credible failure mode, not just physical possibility.

•

FERC Chapter 18 further states that in addition to information gained during the PFMA, screening of credible PFMs requires knowledge and understanding of each PFMs estimated loading likelihood, likelihood of failure, and consequences. In quantitative terms, the annual exceedance probability separating a clearly negligible PFM and non-negligible PFMs could be 1/1,000,000 (defined by FERC as Remote). However, 1/1,000,000 may not be sufficient justification for

1251

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA deeming a PFM as clearly negligible. Consequences and judgment of the subject matter experts must be considered. In the cases where there are large downstream populations, emote Likelihood PFMs are retained for Refined Screening during the semi-quantitative risk analysis •

During the SQRA process, PFMs are further evaluated through “Refined Screening.” Regarding this phase of evaluation FERC Chapter 18 states: “The refined screening of credible potential failure modes will result in the potential failure modes further differentiated into potentially significant and insignificant. Potentially significant potential failure modes are those that contribute substantially to the total project. Insignificant potential failure modes are those that after additional evaluation and discussion are considered to be so remote as to be considered negligible.” Potentially Credible PFMs that are considered to be Urgent or Potentially Significant are estimated utilizing the SQRA process and plotted on the risk matrix. Potentially Credible PFMs that are determined to be Insignificant are not plotted on the risk matrix.

•

Regarding performing breach and consequence analysis, FERC requires breach analysis and consequence analysis for use in the SQRA. In addition, Emergency Action Plans (EAP) are required for all High Hazard Potential and Significant Hazard Potential dams.

Owner’s perspectives of credible failure mode concept Five TSF owners shared their perspectives on how they discern credible failure modes. There were two fundamental approaches: 1. Credibility solely based on whether the failure mode is physically possible. The term “physically possible” was adopted rather than “technically feasible” since the criteria associated with a failure mode being physically possible were interpreted to be clearer than the criteria associated with discerning technical feasibility. 2. Credibility based on whether the failure mode is physically possible and has a likelihood that is non-negligible.

Credibility based on failure modes that are physically possible The determination of whether a failure mode is physically possible can be relatively straight forward. For example, if a TSF is contained by two rockfill dams on rock foundation and the crest elevation of one dam is much higher than the other dam, then the dam with the higher crest elevation can never over top; it is physically impossible. The focus of this approach is considering all physically possible failure modes as credible and designing the TSF such that the probability of these failure modes occurring is reduced to a level considered negligible. This approach can support the establishment of design criteria based on the dam consequence

1252

CREDIBLE FAILURE MODES – SUMMARY OF 2021 AND 2023 WORKSHOPS classification, although some mining companies are selecting more conservative design criteria than the criteria indicated by the TSF consequence classification. The Owners that advocate for this approach suggest the following advantages relative to the second approach (discussed below): • Discerning whether a failure mode is credible or non-credible is simpler and does not require an estimation of likelihood to support it, which is especially challenging when dealing with likelihoods lower than 1 in 1,000,000, thus, avoiding the potential concern of declaring a failure mode as non-credible prior to detailed suitable examination. • Avoids the challenges of the dual consequence classification discussed in the following subsection. With this approach, if a failure mode is non credible (as in not physically possible), it would not need to be included in emergency planning. For the Owners that consider likelihood in emergency planning, there can be a communication challenge with stakeholders, emergency management agencies, and regulators when they are presented with hypothetical failure scenarios in emergency planning documents and simulations. There could be also challenges with stating that a large engineering structure has no credible failure modes.

Credibility based on failure modes that are physically possible and non-negligible In this case, the definition of negligible is critical. Concepts have been developed in the United States for the water-dam industry that potential failure modes with an annual failure probability less than approximately 1 in 1,000,000 and considered negligible (FERC, 2021). Thus, for TSFs, failure modes having annual failure probability less than 1 in 1,000,000 (remote likelihood) may be determined to be noncredible failure modes. Some failure modes can be assigned an estimate of probability while others cannot. For those cases where a likelihood cannot be readily assigned, FERC (2021) provides the following additional statements related to negligibility: The possibility cannot be ruled out, but there is no compelling evidence to suggest it has occurred or that a condition or flaw exists that could lead to initiation. Several events must occur concurrently or in series to cause failure, and most, if not all, have negligible likelihood such that the failure likelihood is negligible. These statements and similar ones are being adopted by some Owners to help in the determination of whether a failure mode is credible or not. In accordance with GISTM, the TSF consequence classification should be based on credible failure modes. Using the approach that considers negligibility, only credible failure modes would be used to define the consequence classification. The TSF consequence classification based on credible failure modes that

1253

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA consider negligibility as well as physical possibility should only be used for disclosure and emergency planning purposes. With respect to establishing design criteria and the stewardship, the TSF can be classified in terms of possible failure modes, either assuming the TSF fails or based on physically possible failure modes. As noted above, the Owner may go beyond the requirements based solely on a consequence classification. This could result in a “dual classification,” one for disclosure and external communication and the second for dam safety design criteria and stewardship. If the definitions of negligibility provided above are adopted, care must be taken to conduct robust and appropriate evaluations, supported by documentation and reviews to make the determination of negligible and, therefore, non-credible according to this approach. The evaluations may include a combination of meeting robust (and “Extreme”) design criteria, reduction of uncertainties due to detailed site investigations, testing and thorough analyses (including sensitivity analyses that bound any remaining uncertainties) and considerations of ongoing/implemented redundant controls that would require multiple failures of existing systems to overcome. The evaluation may also consider a quantitative estimate of probability; event trees may assist in the determination of negligibility. The Owners that advocate for this approach believe that it provides clarity when communicating with external stakeholders. When these Owners communicate to stakeholders, emergency management agencies, and regulators that a particular TSF has credible failure modes, the stakeholders, emergency management agencies, and regulators can understand this concept based on the other emergency plans that they have to deal with. However, there have been mixed experiences. Hence, the Owners need to understand the perceptions and understanding of the stakeholders, emergency management agencies, and regulators to properly convey the message.

Regulator’s perspectives Rob Schryburt (Ontario Ministry of Northern Development, Mines, Natural Resources and Forestry) is the Chair of the Canadian Dam Association’s Mining Dams Regulatory Committee. In the 2021 workshop, Rob noted the following: • The TSF classification that informs the design and stewardship should be based on physically possible failure modes or the assumption that the TSF fails. It should not be based on credible failure modes that consider negligibility. • The Regulators recognize that many in the TSF safety community are moving away from basing their designs and stewardship on the dam classification alone and considering other factors, such as potential impacts to the Owners. This may result in more stringent requirements.

1254

CREDIBLE FAILURE MODES – SUMMARY OF 2021 AND 2023 WORKSHOPS • The Regulators recognize that they will be seeing the two different approaches with respect to credible failure modes (described above) and the Owners are going to have to explain how the approach they are adopting is not compromising the safety of the TSFs or inappropriately reducing the emergency planning requirements.

Summary The discussions and perspectives presented by members of the Global Tailings Review Panel and GPG working group as well as Owners and Regulators gathered from the two workshops are summarized in this paper. Recognizing that the concept of credible failure modes when applied to the safety of the TSFs is evolving, the workshops highlighted the following the key messages: • There are two interpretations of credible failure modes, one based on whether a potential failure mode is physically possible and another based on determining that the probability of failure is considered non-negligible. • Both interpretations embrace risk assessments and call for a thorough understanding, evaluation and management of risks associated with potential failure modes. • Both interpretations satisfy the intent of the GISTM. • Dam safety design criteria and stewardship should continue to be based on Consequence Classifications for technically possible failure modes, either assuming the TSF fails or based on physically possible failure modes, even where the probability of failure is considered negligible. • A key point that was agreed to among all workshop team members was that the main objective for the design and management of TSFs is to minimize risks and manage potentially catastrophic failure modes using the ALARP (as low as reasonably practicable) principle throughout the lifecycle of the TSF.

References Federal Energy Regulatory Commission (FERC). 2021. Chapter 17 – Potential Failure Mode Analysis. In Engineering Guidelines for the Evaluation of Hydropower Projects. Washington. Global Tailings Review (GTR). 2020. Global Industry Standard on Tailings Management (GISTM). International Council of Mining and Metals (ICMM). 2021. Tailings Management, Good Practice Guide.

1255

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

1256

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

The Tailings Safety Case – An Example Jiri Herza, Czech Technical University, Czech Republic Jarrad Coffey, Rio Tinto, Australia Ryan Singh, HATS Consulting, Australia

Abstract Safety Cases have become desirable for water dam owners to demonstrate that the risks associated with their dams are reduced to as low as reasonable and that the dam owner’s duty of care is fulfilled. Additionally, Safety Cases have precedence as a risk control communication tool within other high-risk industries. The Safety Case framework is a valuable tool for risk-informed tailings management that can satisfy the current industry-wide search for a consistent methodology to risk control identification, evaluation, implementation, verification, and communication that the risks posed by a tailings facility are As Low As Reasonably Practicable (ALARP). This paper builds on the paper presenting the key elements of a Safety Case for tailings storages (Herza et al., 2022) and demonstrates the example of a real Safety Case used to demonstrate how risks will be controlled to as low as is reasonably practicable for an iron ore tailings storage facility located in the Pilbara region of Western Australia.

Introduction As proposed in Herza et al. (2022) (referred to as the previous paper), a Safety Case is a document which allows a tailings storage facility (TSF) owner to demonstrate through structured and logical arguments that all reasonably practicable risk controls are either in place or will be implemented using a Tailings Management System (TMS). This demonstration is needed for owners striving to meet their legal duties and conform with industry standards in relation to the TSF safety, which often involves demonstrating that risks are reduced to as low as reasonably practicable (ALARP) or so far as is reasonably practicable (SFAIRP). As discussed in the previous paper, it has been established that these terms share the same purpose and definition. Consequently, the term ALARP has been consistently utilized in this paper. Demonstrating that risks are ALARP is also a requirement of the Global Industry Standard on Tailings Management (GISTM), which all member companies of the International Council on Mining and Metals (ICMM), together with numerous other companies, have committed to comply with.

1257

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The previous paper outlined a conceptual tailings dam safety management framework built open elements (shown in Figure 1), which are also fundamental to the Safety Case.

Figure 1: Key elements of a Safety Case (Herza et al., 2022) This paper describes the implementation of the proposed framework for an iron ore tailings storage facility, owned and operated by Rio Tinto Iron Ore (RTIO) in Western Australia. The proposed framework, together with the discussion in the previous paper, are proposed to be considered as a roadmap for demonstration of responsible tailings management that meets legal and regulatory requirements and current practice. It is emphasized that the Safety Case is the final document for demonstration of the process and its outcomes. Although the Safety Case may have little value in managing risk, it is a powerful communication tool for internal and external stakeholders and decision-makers.

Some further clarifications General The previous paper provided essential clarifications pertaining to the foundation of the framework depicted in Figure 1. Presented below are supplementary explanations that should also be comprehended within the context of the Safety Case framework discussed in this paper.

What can be reasonably practicable? The logic of this paper is founded on the notion that only the risk reduction measures, not the risk itself, could be reasonably practicable as detailed in Herza and Singh (2023). The key argument for this notion is that “practicable” refers to something that is physically possible, while a TSF risk position expresses the risk assessors’ degree of belief of the probability and consequences of various failure outcomes. For clarity,

1258

THE TAILINGS SAFETY CASE – AN EXAMPLE the definition of the term “probability” (provided below) is adopted from ICOLD Bulletin 130 on Risk Assessment in Dam Safety Management (ICOLD, 2005) and ANCOLD Guidelines on Risk Assessment (ANCOLD, 2022): “A measure of the degree of confidence in a prediction, as dictated by the evidence, concerning the nature of an uncertain quantity or the occurrence of an uncertain future event. It is an estimate of the likelihood of the magnitude of the uncertain quantity, or the likelihood of the occurrence of the uncertain future event.” Because “reasonably practicable” refers to risk controls, identification and assessment of these controls should be the focus of the risk assessment and ultimately the Safety Case presented in this paper.

Quantified risk Although risk quantification may neither be necessary nor sufficient for demonstrating that risks are managed to ALARP standards, it presents unparalleled value for comparing assets, hazards, risk controls and other risk-related matters. Risk quantification can also assist in decision making over what risk controls are reasonably practicable through cost-benefit analysis. The example facility presented in this paper was part of a portfolio risk assessment (PRA) where the risk of TSF failure was quantified using a consistent risk analysis tool called Risk Assessment and Management Platform (RAMP). While the quantified risk profile for the example facility from RAMP was not essential for the risk assessment process discussed below, the quantified risk profile has been and will continue to be used to prioritize identified reasonably practicable risk controls both for the facility and amongst the portfolio of facilities managed by RTIO. This use of quantified risk profiles aligns with the principles of equity and efficiency, which are aimed to be optimized through the risk assessment process. Additionally, prioritization of risk controls based on an understanding of the overall risk reduction that can be achieved for a facility or across a portfolio of facilities demonstrates a reasonable approach to implementing a myriad of risk controls within the realities of limited resources and time to do so. Thus, the use of quantified risk profiles aligns with the overarching philosophy and basis that risk management should be carried out in a reasonable and defensible manner.

Current practice Although there is no unanimously agreed level of engineering, management, or risk controls that can be referred to as the current practice, current practice can be defined as a reasoned and negotiated course of action that a given person or entity would be expected to undertake under the given circumstances. This definition relates the current practice with the professional and legal concepts of standard and/or duty of care and aligns with the ordinary meaning of the words “reasonably practicable.” The definition of these words provided in Eckersley v Binnie and Partners (1988, 18 Con LR 44, CA) describes what can

1259

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA reasonably be implemented, which at a minimum, should be what a given person or entity would be expected to do in a given situation (e.g., current practice). There is also legal precedence from other hazardous industries, which indicate the key question being asked of responsible parties after the incident was “did you do everything that would was expected of you?” (e.g., current practice) (IMechE, 2021). It is important to recognize that current practice develops over time, and it is spatially variable. For example, the level of engineering expected in Australia in the 1980’s was different to the level of engineering expected in Brazil in 2022.

Tailings management system While the key objective of the risk assessment outlined in this paper was to identify all reasonably practicable controls, this outcome would provide little value if these controls are not implemented, verified to be effective and improved over time, which can be effectively achieved through a TMS. As discussed in Cook et al., (2022), RTIO’s TMS supports effective tailings management by detailing the processes and procedures to be consciously implemented in achieving the tailings management goals stated in Rio Tinto’s Tailings Policy. Core to the TMS is the principle of Plan-Do-Check-Act which is a commitment to continual improvement of the system and tailings management. Risk controls which are implemented via the TMS will therefore be reviewed, revised and improved to continually reduce risks and improve TSF safety.

Development of a Safety Case General The following case study outlines the application of the risk assessment framework presented in Figure 1, to reach a state where all reasonably practicable controls are identified and have been or will be implemented. The key steps in this process were as follows: • Identifying the causes, triggers, hazards and states leading to the defined unwanted event, selected to be the “uncontrolled release of tailings and/or water.” • Identifying reasonably practicable controls to eliminate those causes, triggers, hazards, and states, and where not possible to eliminate them, mitigate and manage the risks presented by those causes, triggers, hazards, and states through the implementation of reasonably practicable controls. The risk assessment and the outcomes were documented in the Safety Case, which was intended to serve as a document that demonstrates to all stakeholders how RTIO is achieving or will achieve safe operation by using adequate controls and satisfactory management systems per the intent stated in Safe Work Australia (2012).

1260

THE TAILINGS SAFETY CASE – AN EXAMPLE It is also noted that the activities for the case study described below were completed in workshop settings with RTIO’s tailings governance and operations personnel, wider-RTIO stakeholders (e.g., environmental advisors), Engineers of Record (EoR) and independent technical engineers so that the outcomes considered inputs from all personnel identified as accountable, responsible, consulted, and informed within RTIO’s TMS.

Basis of the Safety Case The risk assessment process applied the following hierarchy of documented requirements and guidance: • Applicable legal and regulatory requirements – any such requirements took precedence over all other inputs and considerations in acknowledgement of RTIO being a responsible corporate entity undertaking activities within a legal and regulatory framework. While there were no legal or regulatory requirements for RTIO to develop a Safety Case for their TSFs, the Work Health and Safety Act (2020) provided the requirements on the management of risks to health and safety. • Applicable standards – these included any industry and internal RTIO standards regarding the management of risks. In addition to RTIO’s risk and tailings standards, the requirements of the GISTM were considered. • Applicable guidelines – any guidelines related to tailings management, risk management, or on the development of Safety Cases. Additionally, internal RTIO guidance notes were considered during the process. In particular, the following guidance was considered: o

ALARP for Engineers: A Technical Safety Guide (Institution of Mechanical Engineers Safety and Reliability Group, 2021)

o

Safety Case Guideline 3rd edition (Engineers Australia, 2014)

o

Guide for Major Hazard Facilities: Preparation of a Safety Case (Safe Work Australia, 2012)

o

HSE UK – Safety Case (various guides – see https://www.hse.gov.uk/buildingsafety/safety-cases/index.htm)

TSF overview The need to properly define the system It was critical that the TSF system being assessed was defined and understood by all participants in the process so that all reasonably practicable risk controls that were ultimately identified were applicable and relevant to the facility,

1261

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The definition of a system should include the physical elements of a facility, and also extend beyond to include processes and actions across management, operation, investigations, design, and construction of the facility so that failures in these processes could be captured and controls could be identified. The identification of controls should also be cognisant of any confirmed plans for the facility so that identified controls are relevant and reasonable for the facility. For example, it would not be reasonably practicable to commit to implementing a control that relies on a component of a facility that was planned to be decommissioned. Additionally, acknowledging plans for a facility can allow identification of opportunities to efficiently implement several controls (e.g., including the scope for required geotechnical investigations to support several identified controls within a wider geotechnical investigation scope to support a planned facility expansion).

Example facility – TSF1 The TSF for this case study is the TSF1 at RTIO’s operation, located in the Pilbara, Western Australia. TSF1 was built as a side-hill impoundment to receive waste fines, where the solids settled out and water expelled during settling and rainfall runoff was returned to the process plant for reuse. The facility can contain the 1 in 100 Annual Exceedance Probability (AEP) 72 hour rainfall event without the decant system being operational and it still maintains adequate freeboard as required by the regulator. The facility was built in two stages with the first stage designed to contain 2 years of waste fines production at an embankment level of RL 668 m or approximately 16 m high. Stage 2 consisted of raising the embankments to an elevation of RL 678 m or 26 m high. The facility has the following key features: • Three zoned embankments, consisting of a low-permeability upstream zone and a structural downstream zone. The embankment also featured a cut-off beneath the low permeability core of the embankment, cut to a depth of approximately 3 to 4 m. • An upstream underdrainage system comprising a collector drain system and finger drains within the basin of the system, which feeds to a pumped underdrainage collection tower on the western embankment. The underdrainage system has been blocked and inoperable for several years. • An emergency spillway which was excavated partially in natural ground and partially in embankment fill along the south-eastern boundary of the facility, between the East Embankment and the natural ridgeline to the south. • A tailings delivery system, consisting of a tailings delivery pipeline and perimeter deposition line along the crest of the facility. • A pumped decant return system consisting of a slotted decant tower near the south of the facility, fixed pump and return line, which returns supernatant water back to the process plant. The general arrangement of TSF1 is presented in Figure 2.

1262

THE TAILINGS SAFETY CASE – AN EXAMPLE

Figure 2: General arrangement of TSF1 At the commencement of this process for TSF1, the facility had not been operational for several years due to tailings storage volume being available within mined-out pits and there were no plans to recommission the facility. During the first quarter of 2023, RTIO further committed to not utilize TSF1 for tailings storage in the future. A closure study for TSF1 was being conducted at the time of writing this paper.

Acknowledgement of existing risk controls The risk assessment process was completed acknowledging that RTIO had procedural-based risk controls which were applicable to all their TSFs, with implementation, verification, and improvement activities and outcomes being driven through RTIO’s TMS. As the intent of the risk assessment process and the Safety Case was to identify, assess, and document reasonably practicable risk controls, the TMS-driven risk management actions were recognized to avoid re-identification, re-assessments and re-documentation of these controls for every facility within the portfolio. The risk control identification was developed considering these actions, with the risk control identification process only focussing on identifying instances where these controls were not in place or were found ineffective. It was also acknowledged that RTIO manages their TSFs using systems and processes that are common to all RT operated sites. Similar to the TMS controls, these wider systems and processes were not

1263

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA reidentified, reassessed, and redocumented. Instead, these systems and processes were taken to be a part of how the TSFs were managed. Examples of such systems and processes included training, management of change, incident management, information management, and environment and social management systems.

Risk identification Potential Failure Mode Analysis Potential Failure Mode Analysis (PFMA) is a well-established method of identifying and documenting how loss of control over a system may lead to a material unwanted event (failure). The objective of PFMA is to identify the reason and steps that result in an event that is defined as failure or failure of a system. Essentially, documenting a narrative of how something may go wrong and why. The objective of the PFMA for this assessment was to identify the states or events (and underlying reasons) that would lead to the unwanted event, so that all potential controls could be identified. This objective set the battery limits of the assessment to anything that could impact TSF1 resulting in release of stored material. There are many tools used to conduct PFMA, including Event Tree Analysis (ETA), which utilizes the logic of “if this happens, what would happen next?” and Fault Tree Analysis (FTA) where the focus is on “what are the causes leading to this?” (ICOLD, 2005). FTA was adopted as the method to conduct the PFMA for this assessment.

Physical Fault Tree Analysis The FTA process started with the unwanted event (also called the top event), and then identified the states or processes necessary for the top event to occur using a bottom down, deductive method. The top event for the analysis was defined as “unwanted event of uncontrolled release of tailings and/or water.” To achieve consistency of the PFMA logic, all facility specific fault trees were based on a generic fault tree, which was established to describe the physical events leading to failure (Physical Fault Tree) for all TSFs in RTIO’s portfolio. While the initial Physical Fault Tree (PFT) was generic, it did not diminish the value of site specific PFMA because it reflected the unique features of the TSFs and it was used as prompts to identify and describe the failure modes and hazards specific to each facility rather than limiting the PFMA process. The PFT for each facility covered only the observable or measurable states or steps that could lead to the top, unwanted event. The outcome of this process was a set of bottom events (root causes in Fault Tree nomenclature), which were the fundamental causes leading to failure, also known as hazards. An example of a branch from the generic PFT is presented in Figure 3 which, from left to right, demonstrates how the deductive assessment was used to identify that uncontrolled release of tailings and/or

1264

THE TAILINGS SAFETY CASE – AN EXAMPLE water could be caused by release of a confining structure due to confining structure crest settlement, caused by a buried conduit failure.

Figure 3: Example of a physical fault tree branch The generic PFT was used as a tool in workshops for TSF1 to identify which branches of the PFT were applicable and which branches were not applicable or physically possible. For example, TSF1 did not feature any buried conduit, so the branch of the PFT presented in Figure 3 was removed from the final fault tree for TSF1, with the reasoning being documented. All branches that could physically occur were adopted for the PFT for TSF1 regardless of the assessed probability given the objective of identifying all reasonably practicable controls for the defined risk and a conscious decision was made to not rely on the estimation of likelihood for the identification and decision of reasonably practicable risk controls, due to the considerations regarding the quantification of risk discussed above. However, the relative probabilities of failure were considered later in assisting with the prioritization of the implementation of risk controls. Branches for which it was evident that no controls could be reasonably implemented, such as impact from a meteor strike, were not taken forward given the inability to implement reasonably practicable controls, and were documented as such. The outcome of this process was a numbered list of documented generic bottom events, and associated facility-specific bottom events. A select list of generic and specific bottom events for TSF1 is shown in Table A1 (included in the Appendix).

Process and People Fault Tree Analysis It was identified during the development of the risk assessment process that the Physical Bottom Events could rarely be controlled directly as they were either a feature of the TSFs themselves, or were caused/exacerbated by failures in underlying processes or actions. It was reasoned that to truly meet the principles and objective of reducing risks to as low as reasonably practicable, additional controls were

1265

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA required for potential failures in underlying processes and human actions that lead to Physical Bottom Events even being possible. To identify all applicable Process and People Bottom Events (and ultimately all reasonably practicable controls), separate Fault Trees were developed below each Physical Bottom Event. The branches of each Process and People Fault Tree were defined using the commonly referred to life-cycle stages of a TSF: • inappropriate/insufficient investigation; • inadequate design; • inadequate construction; • maloperation; and • lack of maintenance. An example of a Process and People Fault Tree underlying the Physical Bottom Event of conduit failure is presented in Figure 4. It shows how the deductive method can be used to identify deficient construction, design, and lacking maintenance either singularly or in some combination as the underlying causes for Physical Bottom event of “conduit failure.”

Figure 4: Example of a physical fault tree branch The process described for the identification of the facility-specific Physical Bottom Events was carried out for the identification of facility-specific people and process bottom events. For a given applicable Physical Bottom event, the underlying generic process and people bottom events were assessed as being: • Not possible – the generic failure in processes and people was not applicable for the specific Physical Bottom event. • Possible – however, agreed among all participants that there was sufficient evidence to demonstrate that controls or current practice had been implemented (e.g., all construction documentation was available for a given aspect of TSF1 and no further controls could be reasonably implemented). Given the level of engineering, operation and governance of tailings management required by RTIO’s internal tailings standard, the majority of potential Process and People errors for any given Physical Bottom Event were within this category.

1266

THE TAILINGS SAFETY CASE – AN EXAMPLE • Possible – and it was agreed that there was insufficient evidence to demonstrate that all reasonably practicable controls were in place to address the hazard or that current practice had been met with respect to the potential failure in processes or actions. An example of the outcomes of the assessment for TSF1 is presented in Table A2 (included in the Appendix) with clear links to the applicable Physical bottom events. At the end of this process, the outcome was a list of all potential physical, procedural, or action-based hazards or triggers that could lead to the defined unwanted event based on the information available at the time of the assessment. As discussed further below, some risk controls included further investigations to better understand the facility to reduce uncertainty and therefore risk. It was acknowledged that this updated knowledge may result in a revision to the list of all potential physical, procedural, or action-based hazards or triggers.

Consequence assessment The potential consequences of failure of TSF1 as a result of uncontrolled release of tailings and/or water were assessed as part of an RTIO portfolio-wide dam break study and failure impact assessment. The consequence assessment outcomes were considered for the prioritization of reasonably practicable risk controls and for consequence mitigation. The consequences of failure were also considered in the identification and implementation of consequence mitigation actions for pre and post-failure which were acknowledged to be just as important as risk controls to prevent the unwanted event. The uses of the consequence assessment are discussed further below.

Risk analysis Understanding of risk and risk controls Subsequent to the identification of the physical and process and people bottom events was the identification of potential controls. All potential controls, irrespective of their practicability and practicality were identified and listed to align with the principle of “starting with everything that can be done and only doing less when it is reasonable to do so” (Sheriff, 2011). In addition, without assessing all potential controls within the consistent workflow described below, a pre-emptive “gutfeel” assessment of reasonably practicableness could miss key considerations resulting in different, undefendable outcomes. The generic wording of the various fault trees assisted with the identification the fundamental risk controls by inverting the words of the applicable Physical or Process and People bottom events. For example, for the Physical Bottom Event of “Spillway flows undermining the embankment,” the fundamental control would be “Spillway flows not undermining the embankment.” Similarly, for the Process and People Bottom Event of “Insufficient or inadequate laboratory testing,” the control would be

1267

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA “Sufficient and adequate laboratory testing.” These generic controls were then used as prompts to identify specific controls, considering the facility-specific Bottom Events. The hierarchy of controls was also applied in developing specific controls for each Physical Bottom event: • Elimination of Physical Bottom Events. • Modification controls to physically modify the TSF1 system to mitigate the risk presented by Physical Bottom Events. • Engineering controls to reduce the uncertainty associated with the various Physical Bottom Events and thereby reducing the associated risk. • Procedural controls to increase the confidence in the management of TSF1 being aligned with the design intent, in addition to increased confidence in verification activities. The list of controls also included actions to address any previously identified gaps against current practice. Using this workflow, all potential controls were identified for all Bottom Events that could result in the defined unwanted event. Examples of potential risk controls identified for TSF1 are presented in Table A3 (included in the Appendix). Besides the control type, the identified risk controls were analyzed in the context of the following criteria, which were then used for the selection of reasonably practicable risk controls as part of risk evaluation: • Is having the control implemented current practice? • What is the level of confidence in control effectiveness? • Is there precedence for the control being implemented? • What are the risks introduced due to implementation of control? • What is the sacrifice to implement control? • What opportunities are lost by diverting resources from and/or effecting other committed activities/risk controls and the scale of opportunity cost? • Does the risk control address the dominant failure mode? • What are the means of control effectiveness verification? A tabulated format of the risk analysis scheme addressing the above listed questions is presented in Table A4 (included in the Appendix).

Risk evaluation The objective of risk evaluation was to identify all reasonably practicable risk controls based on the risk controls characterization developed as part of the risk analysis. The identified and analyzed risk controls were decided to be either reasonably practicable to implement, or not reasonably practicable to implement, or it was found that additional work was required to make the decision.

1268

THE TAILINGS SAFETY CASE – AN EXAMPLE The implementation scopes for each control were not necessarily defined during the process, as initial investigations and studies may be required first to effectively implement and verify the control. Instead, the immediate actions that could be defined, were defined. Additionally, where possible, actions for controls were grouped together into scopes, typically structured as a study, to allow effective and efficient undertaking of actions. For controls that were assessed to not be reasonably practicable to implement, it is intended that the discussion and reasoning be reviewed periodically to determine if this assessment is still appropriate in the future. A selection of the controls judged to be reasonably practicable for the TSF1 is presented in Table A5 (included in the Appendix).

Supporting studies Throughout the assessments discussed above, any supporting studies which were identified to be required to confirm, better understand, or confirm if other Bottom Events were applicable, were documented. An example of this was an updated flood hydrology study for TSF1 to support the implementation of all floodrelated controls. Additionally, for Bottom Events where it was obvious that studies or projects should be carried out to address the resulting risk, this was also documented for action as soon as possible. For TSF1, the (at the time) ongoing closure study, which included supporting geotechnical investigations, could be utilized (with its existing scope, or by scope additions) to undertake the various actions required to implement the identified reasonably practicable controls.

Consequence mitigation As stated above, consequence mitigation was acknowledged to be as important for risk management as controls to eliminate or mitigate failure. Consequence mitigation options that were considered for TSF1 fit within three broad categories: • Elimination of the TSF (which is the control initially considered in all cases and was discussed earlier). • Steps toward removal of the hazard, primarily through an acceleration to closure of the facility, or isolation of the inundation effects and sensitive areas. • Emergency planning and response in the event of a failure of the facility developing. Elimination of TSF1 was discussed previously as part of control identification and analysis. In addition, it is noteworthy that the site has employed in-pit deposition since 2016, which included operation in tandem with TSF1 until 2021, which was achieved through successful long-term planning and is an example case of the importance of mine and tailings planning integration presented in Coffey et al. (2021). Adoption of in-pit deposition and avoidance of the creation of dam safety hazards are now central to the

1269

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA company’s TSF development strategy, but these were not formalized at the time of TSF1 being developed initially. TSF1 is being progressed toward closure, with curtailment of deposition in 2021 and closure studies underway to determine the best means of rehabilitating the facility. An option to reclaim the facility and relocate all stored tailings into a pit is included in this study. The site has comprehensive emergency preparedness and response planning in place. These can be summarized as: • Business resilience and response plan: over-arching Company plan, providing clear guidance for the management of emergencies and ensuring the necessary groups and resources are deployed. • Emergency Response Plan: site-based document encompassing response to all identified hazards, incorporating the Tailings Response Plan. • Tailings Response Plan: emergency response procedures including evacuation for TSF1. • Trigger Action Response Plan (TARP): quantified triggers and defined actions required in response to prevent further progression and mobilization of suitable emergency preparedness. • Emergency Action Plan: detailed mitigative response procedures for lower level quantified triggers defined in the TARP. The above is being implemented to meet all GISTM requirements, including consultation with external stakeholders to ensure a joint state of readiness to respond to a failure event, ongoing education, and regular TRP exercises and drills. Due to the very low number of assessed people at risk, consequence mitigation provided by emergency response for TSF1 is relatively limited to other TSFs, to which many more people may be exposed.

Implementation of reasonably practicable controls – the tailings management system In RTIO, the TMS is considered as the implementation of dam safety management, in addition to the implementation of other requirements, such as RTIO’s internal standards and RTIO’s interpretation of the GISTM requirements. All reasonably practicable controls have been or will be implemented within the RTIO TMS framework, such that the actions, activities, evidence and outcomes to implement, verify and improve the controls meet RTIO’s expectations of effective tailings management. The TMS also incorporates reviews of the risk assessment outcomes to assess if all reasonably practicable controls are implemented at any time, given that conditions at TSF1 and current practice for the tailings industry are continually changing and evolving. This continual review and improvement is a significant step in demonstrating that ALARP as a principle is being aligned with at all times, and it is acknowledged that the TMS itself will require changes to accommodate effective implementation of future risk controls.

1270

THE TAILINGS SAFETY CASE – AN EXAMPLE

The Safety Case The Safety Case documents the process and outcomes of safety assessments and management to demonstrate how RTIO manages and plans to manage the risks presented by their TSFs. It should be reiterated that the Safety Case itself does not have inherent value in eliminating or reducing risk, but it is rather a powerful communication tool for internal and external stakeholders and decision-makers. As such, the Safety Case must be presented in a form that allows the intended audience to easily discern and decide if they believe the TSF owner has met their objectives regarding risk management. For RTIO, it was decided that the Safety Case document should be split into two separate documents. The first document, called the Portfolio Safety Case, documented the overarching process described in this paper. The second document, called the Facility Safety Case, stated the outcomes and risk controls measures specific for a given facility. It was reasoned that the Facility Safety Case would allow the message of how RTIO manages a given TSFs risk to be clearly documented for all stakeholders. Stakeholders who wish to understand the basis further could then refer to the Portfolio Safety Case.

Conclusion and discussion The case study presents the risk assessment and management process adopted by RTIO for their TSFs, with the ultimate aim of achieving a state where all reasonably practicable risk controls are identified and either have been or will be implemented within a formal TMS. It has been shown that quantification of risk was not required to make an informed decision regarding what actions should be undertaken to reduce risks to ALARP. The outcomes of the process are documented in a Safety Case that describes not only the outcomes, but also the process itself, so that any reader can decide for themselves whether the process and the outcomes were reasonable and whether RTIO manages the tailings risks appropriately. This documentation is key for responsible ownership as legal precedent has demonstrated that after failure, the focus is on whether the accountable and responsible parties foresaw the potential failure and did everything reasonable to avoid the failure, and where not possible, everything reasonably practicable to reduce the risk of the failure. While the framework and outcomes are presented as an example of how RTIO manages risk to ALARP, it is the underlying philosophy that is paramount. It is recognized that the expectations and current practice for RTIO are not applicable to all mining operations. Ultimately, the steps adopted for a given owner looking to implement the philosophy and framework discussed in this paper should be reasonable and applicable for their operation and context. For example, the development of a generic fault tree for the PFMA was only necessary for RTIO as it manages a portfolio of TSFs and must demonstrate that their risk

1271

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA management and prioritization of risk control implementation is consistent across all their TSFs for the process to be defensible, efficient and equitable. A generic fault tree would not be required for an owner with a single TSF, and it may not even be required/defensible for various reasons that they utilize the fault tree methodology for their PFMA. A critical point to emphasize is that further assessment should not be undertaken just for the sake of assessment. Where risk controls were identified as being reasonably practicable, such as the decommissioning of the tailings line for the inactive TSF1, it was agreed by all parties that no further detailed evaluation was required before the action is taken. It is hoped that the example Safety Case presented in this paper provides a basis for other TSF owners to develop and implement a framework to clearly demonstrate through structured and logical arguments that all reasonably practicable risk controls are either in place or will be implemented, thus assisting in the progression towards the ALARP state for their TSF(s). It is hoped that the example presented in this paper provides the basis for clearly demonstrating how tailings risks can be managed to as low as reasonably practicable.

References Cook, B., Singh, R. and Coffey, J. 2022. Tailings management system – the missing link. ANCOLD Conference 2022. Sydney, Australia. Eckersley and Others v Binnie & Partners and Others. 1988. EWCA Civ J0218-8. Herza, J. and Singh, R. 2023. The risk of quantifying risk. ICOLD Annual Meeting 2023. Gothenburg, Sweden. International Commission on Large Dams. Herza, J., Coffey, J. and Singh, R. 2022. Key elements of a tailings dam safety case. In Proceedings of Tailings and Mine Waste 2022. Denver, Colorado. November 2022. IMechE (Institution of Mechanical Engineers). 2021. ALARP for Engineers: A Technical Safety Guide. Institution of Mechanical Engineers. London, UK. Lee, D-J., Bates, D., Dromey, C. and Xu, X. 2003. An imaging system correlating lip shapes with tongue contact patterns for speech pathology research. In Proceedings of the IEEE Symposium on Computer-Based Medical Systems. Source: IEEE Xplore. Zielinski, P.A. 2014. Event trees in the assessment of dam safety risks. ANCOLD Conference on Dams, Canberra 2014.

1272

THE TAILINGS SAFETY CASE – AN EXAMPLE

Appendix Table A1: Generic and facility-specific Physical Bottom Events for TSF1 No.

Generic bottom event

Facility-specific bottom event/commentary

2

Underdrainage system inadequate/ineffective.

Upstream underdrainage system consists of a herringbone drain arrangement within the basin, feeding to a drainage sump at the upstream toe of the Western Embankment. Pumped riser on upstream face of the Western Embankment was installed to pump water from the underdrainage system.

7

Spillway active, with flow undermining the Eastern Embankment.

A training bund has been constructed downstream of the spillway to direct flows away from the Eastern Embankment; however, it is foreseeable that spillway flows may overtop the bund and undermine the Eastern Embankment.

9

Alteration of structural zone materials (e.g. strain softening, structural change).

Saturation of foundation zones that may result in structural change of material, specifically mischaracterized materials, leading to reduction of structural zone resistance, leading to loads exceeding strength of structural zone, leading to embankment failure.

18

Spillway active, with flow undermining the Eastern Embankment.

Spillway blocked by debris/objects including failed delivery line supported on box culverts across spillway, leading to spillway capacity being exceeded, leading to flood handling capacity being exceed, leading to embankment being overtopped.

19

Failure of the spillway structure – failure of spillway culverts.

If the culverts across the spillway that are used for the tailings line crossing fail, it could lead to the spillway being compromised or a reduction in the spillway capacity.

Table A2: Process and people bottom event and control assessment for TSF1 Possible process and people bottom event

No.

Facility-specific bottom event/commentary

2

Upstream underdrainage system consists of a herringbone drain arrangement within the basin, feeding to a drainage sump at the upstream toe of the Western Embankment. Pumped riser on upstream face of the Western Embankment was installed to pump water from the underdrainage system.

Design – Inadequate design of underdrainage presented in Design Documentation.

Assessed and agreed that available design documentation presented appropriate design details of underdrainage system. No additional controls required for this bottom event.

7

A training bund has been constructed downstream of the spillway to direct flows away from the Eastern Embankment, however, it is foreseeable that spillway flows may overtop the bund and undermine the Eastern Embankment.

Construction – Inadequate QA/QC presented for construction of training bund.

Assessed and agreed that appropriate evidence demonstrating types and levels of QA/QC for the bund was available. No additional controls required for this bottom event.

9

Saturation of foundation zones that may result in structural change of material, specifically mischaracterized materials, leading to reduction of structural zone resistance, leading to loads exceeding strength of structural zone, leading to embankment failure.

Investigation – Incorrect derivation of structural zone material characteristics presented in Interpretive geotechnical investigation report.

Insufficient evidence available demonstrating that identified deep clay layers had been appropriately tested, investigated and characterized. Action required to implement control of appropriate derivation of foundation material characterization

1273

Assessment

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Possible process and people bottom event

Assessment

Spillway blocked by debris/objects including failed delivery line supported on box culverts across spillway, leading to spillway capacity being exceeded, leading to flood handling capacity being exceed, leading to embankment being overtopped.

Maintenance – Inadequate details of maintenance presented in Operations, Maintenance and Surveillance Manual.

Assessed and agreed that suitable maintenance tactics were available and implemented. No additional controls required for this bottom event.

If the culverts across the spillway which are used for the tailings line crossing fail, it could lead to the spillway being compromised or a reduction in the spillway capacity.

Design – Inadequate structural design of culverts.

Assessed and agreed that available design documentation presented appropriate design details. No additional controls required for this bottom event.

No.

Facility-specific bottom event/commentary

18

19

Table A3: Potential control identification Bottom event 2 – Underdrainage inadequate/ineffective, leading to excess pore pressures, leading to reduction of structural zone resistance, leading to loads exceeding strength of structural zone, leading to embankment failure.

Potential control

Control type

Assess/investigate underdrainage system.

Process and People control

7 – Spillway flows undermining the eastern embankment, leading to embankment undermined, leading to reduction of structural zone resistance, leading to loads exceeding strength of structural zone, leading to embankment failure

Backfill and close spillway.

Elimination

9 – Saturation of foundation zones that may result in structural change of material, specifically mischaracterized materials, leading to reduction of structural zone resistance, leading to loads exceeding strength of structural zone, leading to embankment failure

Geophysics to identify local saturation in foundation.

Process and People control

18 – Spillway blocked by debris/objects including failed delivery line supported on box culverts across spillway, leading to spillway capacity being exceeded, leading to flood handling capacity being exceed, leading to embankment being overtopped

Remove portion of tailings delivery line and blank off.

Elimination

19 – Failure of culverts across spillway supporting perimeter tailings deposition line, leading to spillway capacity being exceeded, leading to flood handling capacity being exceed, leading to embankment being overtopped

Remove portion of tailings delivery line and blank off.

Install trash racks to prevent debris blocking spillway.

1274

Modification

Elimination

THE TAILINGS SAFETY CASE – AN EXAMPLE

Table A4: Control evaluation framework Criteria/aspects

Options/semi-quantitative scale

Comments

Control type

Elimination Modification Engineering Administrative

Level of confidence in control

1 – Precedence of control being always effective 2 – Precedence of control being effective most of the time 3 – Precedence of control being effective some of the time 4 – Precedence of control being ineffective most of the time 5 – Precedence of control being ineffective all of the time

Based on precedence and considering context of control for the TSF being assessed.

Is there precedence for the control being implemented?

Yes No

Precedence of control considering similar context of facility and RTIO’s operations

New risks introduced due to implementation of control

HSE risks Environmental risks Legal/regulatory risks Dam safety risks Community and Social Performance risks Operations

Qualitative description against risk type to give an indication of magnitude of consequences, likelihoods and exposure time of risk.

Sacrifice to implement control

Time to implement the controls in years The number of Full-Time-Employee Years for the personnel in the Process Engineering – Tailings Team. Order of magnitude total cost to implement the control

All considered and described separately.

Lost opportunity by diverting resources from and/or effecting other committed activities/risk controls

Examples include: EoR/Consultant time Loss of storage Committing a dedicated contractor to implement risk control

Scale of opportunity cost

0 – No opportunity cost / control already committed to as part of current plans 1 – Negligible opportunity cost 2 – Little opportunity cost 3 – Some opportunity cost 4 – Considerable opportunity cost 5 – High opportunity cost

Does the risk control address the dominant failure mode?

Yes No

Risk control verification

Examples include: Independent design review Construction QA/QC Instrumentation

Does the control align with current practice?

Yes No

Based on outcomes of semi-quantitative risk assessment.

Negotiated position based on discussion amongst participants.

1275

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table A5: Reasonably practicable assessment Control

Control type

Physical Bottom Event

Reasonably practicable to implement?

Remove portion of tailings delivery line and blank off

Elimination

18 and 19

Yes

Facility is inactive with no confirmed plans for reactivation and as a result, the perimeter tailings delivery pipeline is not required.

No further studies/assessments required as an input to implement this control

Backfill and close spillway

Elimination

7

No

Having an emergency spillway is current practice. Removal of spillway would breach licence conditions and introduce/increase dam safety risks.

Implement control

Install trash racks to prevent debris blocking spillway

Physical modification

18

No

Installation of trash racks in front of emergency spillways is not current practice. Ongoing work/maintenance would be required to clear the trash rack and confirm effectiveness which introduces health and safety risks to operators. The storm event sufficient to mobilize debris would likely prevent operators from accessing and clearing trash racks prior to and during such a storm event.

Assess/investigate underdrainage system

Process and people control

2

Yes

Review and assessment of the underdrainage system should be carried out as a first step to closing/managing the risk of the underdrainage system

Geophysics to identify local saturation in foundation

Process and people control

9

No

No precedence and little confidence in geophysics being appropriate to identify saturated structural zones

1276

Summary of decision making

Next steps

Carry out review and assessment of underdrainage system either as part of a standalone study, or the Closure Study

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Risk-informed Weighting and Communicating Uncertainties for Tailings MAAs Kate Patterson, Klohn Crippen Berger, Canada Jamie Engman, Klohn Crippen Berger, Canada Laura Wishart, Klohn Crippen Berger, Canada Len Murray, Klohn Crippen Berger, Canada James Penman, Klohn Crippen Berger, Australia

Abstract This paper provides an overview of optional tools that can be applied during a conventional tailings management multi-criteria alternatives assessment (MAA). Use of MAAs for selecting tailings management strategies have become a standard part of doing business for mining companies since the Global Industry Standard on Tailings Management (GISTM) was published, which included the requirement in Principle 3. An MAA framework is commonly used to provide a clear, transparent, and defensible methodology intended to eliminate bias, identify stakeholder values, and achieve a relative ranking of various alternatives. However, application of the MAA process can be challenging because of (1) poorly defined purpose and battery limits, (2) embedded uncertainties in alternative characterization and scoring, and (3) stakeholders’ diverse backgrounds and perspectives, which often result in a wide range of opinions for objective weightings used for ranking of alternatives. The optional tools presented in this paper include: (1) confirm overall purpose and plan for the MAA; (2) understand and communicate uncertainties; and (3) risk-informed objective weighing (which helps participating stakeholders align on weightings).

Introduction This paper provides an overview of optional tools that can be applied during a conventional tailings management multi-criteria alternatives assessment (MAA) to: confirm overall purpose and plan for the process; understand and communicate uncertainties; and help participating stakeholders align on objective weightings. An MAA includes a ledger, made up of objectives for selecting a tailings management strategy. Some of the objectives will be more important than others; for example, safety related objectives are more

1277

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA important than cost related objectives. Objective weightings represent the importance of an objective compared with other objectives based on stakeholder’s values, typically on a 1 to 10 scale. The tools presented in this paper do not need to be applied, and potentially should not be applied, to every tailings management MAA. Rather they are potential value-adding tools that companies can use, when appropriate, to make the best decision given the available information.

MAA framework for tailings siting and technology studies A framework for tailings management MAAs was originally proposed by Robertson and Shaw in 1999 and further described in Environment Canada “Guidelines for the assessment of alternatives for mine waste disposal” in 2016. Evolutions of the framework, with additional considerations and guidance, are summarized in: 1. A Guide to the Management of Tailings Facilities (v3.2) Mining Association of Canada (MAC) (MAC, 2021) 2. Tailings Management Handbook: A Life-Cycle Approach (Society for Mining, Metallurgy & Exploration [SME], 2022) 3. Use of Multi-Criteria Alternatives Assessment for TSF Option Selection: Lessons Learned from Recent Applications (Penman and Casey, 2023) All these frameworks are valid, similar in purpose, and provide good information that one should review prior to initiating a tailings management MAA. This paper will not repeat a description of a MAA framework in detail; rather it will use the Penman and Casey (2023) framework steps and calculations to clarify the optional value-add tools presented herein. The framework steps from Penman and Casey (2023) are summarized below and an example populated MAA Ledger with calculations is included in Figure 1. •

Confirm Overall Study Purpose and “the Plan”

•

Step 1: Establish Design Basis

•

Step 2: Establish Knowledge Base

•

Step 3: Facility Location Assessment

•

Step 4: Alternatives Identification

•

Step 5: Preliminary Option Screening

•

Step 6: MAA Framework Development

•

Step 7: Alternatives Characterization

•

Step 8: Apply and Review MAA Framework

•

Step 9: Sensitivity, Threats, and Opportunities Assessment

1278

RISK-INFORMED WEIGHTING AND COMMUNICATING UNCERTAINTIES FOR TAILINGS MAAS Notes on steps:

1. The italicized text prior to Step 1 has been added to indicate a pre-step that will be discussed in this paper. 2. Documentation is not included as a specific step; however, it is an important part of the process at every step.

Figure 1: Example of populated MAA ledger (Penman and Casey, 2023) The optional value-add tools that can be utilized during a tailings management MAA include additions at the following steps: •

Prior to Step 1: Confirm Overall Purpose and “the Plan.”

•

Step 6 to Step 9: Understand and Communicate Uncertainties.

•

Step 6, Step 8 and Step 9: Risk-informed Objective Weighting.

Confirm overall purpose and “the plan” As engineers and scientists, we tend to dive into technical details and try “solve” problems right away; however, selecting a tailings storage facility site and technology is not solely a technical issue. When initiating a tailings management MAA, one should first clearly and succinctly define what they are trying to achieve with the MAA in a statement. Given the importance of this statement in representing the overall purpose of the decision analysis at hand and goal of the tailings facility, it should be authored or approved by the tailings facility Accountable Executive. Some examples of a “Purpose Statement” are given below.

1279

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Develop a short-list of the best tailings storage facility alternatives (combination of site and technology) to be progressed to the next phase of characterization and design that will manage tailings for the project with a maximum production rate of 20,000 tpd. • Determine the optimal tailings management strategy for the project for 10 years at 50,000 tpd after the current tailings storage has reached its current design capacity. • Determine the best tailings storage facility alternative that reduces risk to the project. Once the Purpose Statement is clearly defined and approved, “the plan” for completing the tailings management MAA should be developed. This should include the following: • Facilitation – tailings management MAAs are not purely engineering studies, and as such, a purely engineering approach is not likely to be successful. A facilitator should be engaged to ensure the process is successful. For example, an engineer may ask pointed, direct questions and expect direct answers in return. Often, important input from stakeholders may not come in such a direct and quantitative form. An experienced facilitator will navigate the process and gather stakeholders’ input with open-ended questions initially, then frame considerations/issues for the group to finally gain alignment. • Roles and Responsibilities – tailings management MAAs are multi-disciplinary and incorporate technical, risk, environmental, social, operational, closure, and economic considerations. Therefore, representatives of these considerations should be included to some level during the process. Several structures for roles and responsibility definition can be used, for example, a RACI (responsible, accountable, consulted, informed)-type structure could be useful for large, complex projects. • External stakeholder – external stakeholders (e.g., indigenous, impacted communities, regulators) will be represented in the MAA, either passively (e.g., through community engagement professionals being involved in the process) or actively (e.g., traditional owner groups being directly involved in the process). The method for incorporating their input should be confirmed during the initial phases of the MAA to confirm their input can be meaningfully incorporated within the schedule and framework used. • Schedule – once the Purpose Statement, roles and responsibilities, and external stakeholder plan are developed, the execution schedule can be developed (including workshops with the required participants) such that those involved can understand when and how they will be involved. • Economics – one should consider keeping economics separate to the MAA scoring. Although not always possible or necessary with every tailings management MAA, the authors have found that separation of economics from the MAA score is useful in communicating results both internally and externally. It is helpful to plot MAA results compared to economic metrics.

1280

RISK-INFORMED WEIGHTING AND COMMUNICATING UNCERTAINTIES FOR TAILINGS MAAS • Importance of weighting –The weighting process is an important part of achieving clarity and understanding among stakeholders. More important than scoring, which should be transparent and as quantitative as possible. When there are divergent views on weighting, there are often very good reasons for those views that need to be heard. Adequate time and facilitation are key to building consensus on weighting selections. The sensitivities and risk-informed objective weighting processes presented in this paper can help stakeholders agree on weightings.

Understand and communicate uncertainties Uncertainties are embedded in assumptions included in an MAA and are particularly high at an early stage of mine conceptualization. To understand and communicate uncertainties, a Monte Carlo sensitivity module was programmed in GoldSIMTM (GoldSIM, 2019). In this module, stochastic relationships can be applied to scoring and/or weighting and many realizations can be run to determine the sensitivity of the MAA outcome to a range of potential inputs. The application of this tool can give confidence that the MAA results are one of two things: not sensitive and the tailings management alternative(s) is(are) obvious; or very sensitive and more work is needed to reduce uncertainty to progress the preferred tailings management alternative(s). In the Monte Carlo sensitivity module a base case and sensitivity case can be defined for each objective weighting and alternative score (see Fig. 2). The module randomly selects either base case or sensitivity objective weighting and alternative scores for each realization, with the ability to run any number of realizations. See Figure 2 for an illustration of the process. The user can also increase the likelihood of the base case weighting or scoring being selected or define a stochastic distribution for weighting or scoring if there is enough information to warrant that level of detail.

Figure 2: Monte Carlo sensitivity module To illustrate this process, consider the results in Table 1 and Figure 3, and remember, “rank” is explained in Figure 1 as the ranked MAA scores (best being ranked 1 and increasing to worst) for the

1281

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA alternatives considered. For this case study example, the Monte Carlo sensitivity module was run for 1,000 realizations and the MAA scores (Figure 3) and ranks (Table 1) were calculated for each alternative.

Figure 3: Example of MAA score results using the Monte Carlo sensitivity module Table 1: Example of rank results using the Monte Carlo sensitivity module Alternative 1

Alternative 2

Alternative 3

Base case rank

3

2

1

Minimum rank

3

2

2

Maximum rank

3

1

1

Average rank

3

1.6

1.4

0

400

600

Count of #1

ranks1

Note: 1. Total of 1,000 realizations

It is clear from the results in Table 1 and Figure 3 that Alternative 1 is not the preferred alternative. However, it is not as clear if Alternative 3 or Alternative 2 should be progressed. Below is some additional discussion on this scenario: • There is high confidence that Alternative 1 is not the preferred alternative and should be removed from further consideration. This can be communicated to stakeholders that under all possible circumstances/scenarios evaluated, Alternative 1 is never the preferred. • Uncertainty remains if Alternative 3 or Alternative 2 should be progressed. Depending on the stage of the project or owner’s wishes, the path forward could include the following: o

Review the MAA ledger and identify the alternative scores and objective weightings that are creating the most uncertainty. Can anything be completed now to further decrease the

1282

RISK-INFORMED WEIGHTING AND COMMUNICATING UNCERTAINTIES FOR TAILINGS MAAS ranges that are being considered in the sensitivity analysis? For example:

o

§

More community or indigenous engagement?

§

Better understanding of consequence of failure and risks?

§

More site or tailings characterization?

Carry both options to the next stage and update the weighting/scoring and sensitivity analyses after further development of alternatives.

Risk-informed objective weighting Stakeholder’s diverse backgrounds, perspectives, and biases often result in a wide range of opinions for objective weightings (see Figure 1 for how objective weightings impact the MAA score). There are many available methods for selecting weightings, and those should be reviewed prior to selecting the best process for your situation. A process for risk-informed objective weighting can be used to help gain stakeholder alignment on objective weighting selection. However, prior to implementing this value-add tool, consider if the stakeholder group requires this process. Typically, this process is more effective when a smaller number of alternatives are being considered.

Figure 4: Example statistics of individual objective weightings

1283

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Consider the following example: the authors commonly have each of the MAA participants provide individual objective weightings prior to the objective weighting workshop. These objective weightings are collected and statistically assessed, for an example see Figure 4. Under this scenario, it is clear that the entire group is not aligned on what weighting to apply for each objective. Average weightings could be used as a starting point with a facilitated discussion during a workshop to align on a base case objective weighting. Or a risk-informed objective weighting process could be implemented. Another method to identify if bias exists in the collective individual input is to group stakeholders by their perspective and plot their weightings on a web chart. In Figure 5, each axis represents a perspective (e.g., engineering personnel, board personnel, safety personnel) and each data series represents the average objective weighting for the account (account is the overall category: technical, social, environmental). Figure 5 shows an unbiased set of results in (a) and a biased set of results in (b). After completing these plots (or something similar), one could decide whether implementing a risk-informed objective weighting approach is warranted.

(a) Relatively unbiased weighting

(b) Biased weighting

Figure 5: Example distribution of weightings

The risk-informed weighting process includes the following steps: 1. Identify the risk (or consequence) of not meeting each of the objectives within the MAA Ledger. For example: a. Objective: Minimize dust generation potential (by minimizing exposed area to high velocity winds).

1284

RISK-INFORMED WEIGHTING AND COMMUNICATING UNCERTAINTIES FOR TAILINGS MAAS b. Risk of not meeting objective: Dust cannot be managed effectively and tailings facility construction results in excessive dust that impacts air quality and air permit exceedances. 2. For each alternative and objective, assign a likelihood and consequence category consistent with a risk matrix. Examples of category levels are presented in Table 2. This can be completed by the individual stakeholders prior to the workshop or during a facilitated discussion during the workshop. 3. If step 2 was completed prior to the workshop, a facilitated discussion would occur with the aim of gaining alignment for each likelihood and consequence category from the workshop participants. 4. The risk-informed weighting is calculated as the product of the consequence and likelihood scoring for each objective and each alternative, normalized to a scale of 10 (as shown in Table 2). Each alternative may have a different risk-informed weighting for one objective, for example, see Figure 6. For each objective, the maximum alternative risk-informed weighting could be selected (or the average, but the maximum would highlight its importance). If combining with sensitivity analyses, the bounds of the risk-informed weightings can be used. 5. Through open dialogue, the final “base case” objective weighting is agreed to by the workshop participants for each objective. It should be noted that this approach is not intended to replace a formal dam safety risk assessment, rather it is meant to be a qualitative, informative tool used to align stakeholders in selecting objective weightings. The authors have found this provides a valuable framework for normalizing risks across a diverse set of objectives and perspectives. Another benefit from this process is that it often results in a larger spread of objective weightings, thus highlighting the most important objectives. This process also often triggers discussion for future planning, analyses and risk mitigation plans. To illustrate example results of Step 4, Figure 6 shows the risk-informed weightings for three alternatives and 21 objectives. The greater the risk, the higher the risk-informed weighting and therefore the greater potential to impact the project and MAA results.

1285

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 2: Example of risk-informed objective weighting matrix Consequence 1

2

3

4

5

Low

Significant

High

Very High

Extreme

Descriptions/quantification can be adopted from industry guidance (e.g., Canadian Dam Association (CDA), 2007, 2023) or project-specific metrics (e.g., an appropriate scale for the project scale on social, environment, economics, schedule, etc.)

5

Likelihood

4

3

2

1

Likely

There is direct evidence or substantial indirect evidence to suggest it will occur or is likely to occur in the near future.

2.0

4.0

6.0

8.0

10.0

Possible

There has been convincing evidence to show this will likely occur; indirect evidence suggests it is plausible; evidence is weighted more towards likely than less unlikely.

1.6

3.2

4.8

6.4

8.0

Unlikely

There has been evidence to show this may occur; indirect evidence suggests it is plausible; evidence is weighted more towards unlikely than less likely.

1.2

2.4

3.6

4.8

6.0

Rare

The possibility cannot be ruled out, but there is no compelling evidence to suggest it would occur.

0.8

1.6

2.4

3.2

4.0

Extremely Rare

Several events must occur concurrently or in a series for the result to happen, and most, if not all, have negligible likelihood such that failure likelihood is negligible.

0.4

0.8

1.2

1.6

2.0

Notes: 1.

Numerical values for the risk-informed objective weighting presented are based on consequence and likelihoods categories being given a value from 1 to 5, filling the matrix by multiplying those values, then normalizing to a scale with a maximum value of 10.

2.

Likelihood descriptions are slightly modified from United States Bureau of Reclamation (USBR, 2019).

1286

RISK-INFORMED WEIGHTING AND COMMUNICATING UNCERTAINTIES FOR TAILINGS MAAS

Figure 6: Example of risk-informed objective weighting results matrix

Conclusions This paper gives practitioners optional value-add tools that can be applied during a conventional tailings MAA to achieve the following: confirm overall purpose and plan for the process; understand and communicate uncertainties; and help stakeholders participating in an MAA align on objective weightings. The authors feel that these tools can be extremely useful when applied appropriately. However, they are not always required and may not be useful in every situation. It is best to work with a professional to find the best framework for each situation.

1287

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Acknowledgements The authors would like to thank Andy Small, Dr. Norbert Morgenstern and our MAA clients – Andy for making the time in his very busy schedule to talk passionately about the decision-making process (at any time of day or at the drop of a hat), and who provided essentially all the content in the “Confirm Overall Purpose and the Plan” section as advice during these talks; and Dr. Norbert Morgenstern for his steadfast belief in the MAA process and his eloquently worded challenges wrapped in encouragement (or vice versa). We would also like to thank all the wonderful and patient clients we get to work with during these projects. They are challenging, but fun, and it helps when the client is engaged and ready to do the work needed to make them a success.

References Canadian Dam Association (CDA). 2007. Dam Safety Guidelines (with 2013 revision). Canadian Dam Association (CDA). 2023. Technical Bulletin: Revision to Consequences of Failure Environmental Consequence Classification. Environment and Climate Change Canada. 2013. Guidelines for the Assessment of Alternatives for Mine Waste Disposal. Updated December 23, 2016. Accessed 26/06/2023 at: www.canada.ca/en/environment-climatechange/services/managing-pollution/publications/guidelines-alternatives-mine-waste-disposal.html. Global Tailings Review (GTR). 2020. Global industry standard on tailings management. Accessed 26/06/2023 at: https://globaltailingsreview.org/global-industry-standard/ GoldSim Technology Group. 2019. GoldSim (Computer Software). Penman, J. and J. Casey. 2023. Use of multi-criteria alternatives assessment for TSF option selection: lessons learned from recent applications. Robertson, A. and S.C. Shaw. 1999. A multiple accounts analysis for tailings site selection. Accessed 28/06/2023 at: https://pdf.library.laurentian.ca/medb/conf/Sudbury99/MiningSociety/MS5.PDF Society for Mining, Metallurgy & Exploration (SME). 2022. Tailings Management Handbook: A Life-Cycle Approach. The Mining Association of Canada (MAC). 2021. A guide to the management of tailings facilities. Version 3.2. Accessed 28/06/2023 at: https://mining.ca/resources/guides-manuals/a-guide-to-the-management-of-tailingsfacilities-third-edition/ United States Bureau of Reclamation (USBR). 2019. Semi-quantitative risk analysis. Accessed 27/06/2023 at: https://www.usbr.gov/damsafety/risk/BestPractices/Chapters/A4-Semi-QuantitativeRiskAnalysis.pdf

1288

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

A Comprehensive Risk Management Strategy Michael James, Agnico Eagle Mines, Canada Michel Julien, Agnico Eagle Mines, Canada Jessica Huza, Agnico Eagle Mines, Canada Edouard Masengo, Agnico Eagle Mines, Canada Thomas Lepine, Agnico Eagle Mines, Canada

Abstract The socially and economically responsible stewardship of critical mine waste infrastructure (tailings storage facilities, water management infrastructure and rockfill storage facilities) requires comprehensive understanding and management of the associated risks. This paper presents a risk management strategy that combines quantitative risk assessment with potential failure mode analysis to provide the information required for the identification and characterization of risks as well as the prioritization, selection, and implementation of risk mitigation measures in a well-informed manner. Risk assessment is the first element of the strategy. The method used here applies the Level of Practice (LOP) and the factor of safety to estimate an annual probability of failure (APF), which is then combined with the potential consequences of failure to quantify risk. The LOP is derived from an evaluation of design, construction, operation, maintenance, and observed performance of the structure. The second element is potential failure mode analysis (PFMA), where the likelihoods of credible failure modes under static conditions are assessed. The third element, reconciliation, consists of three parts, static, flood, and seismic. For the static reconciliation, the PFMA is reconciled with the APF to produce discrete APF values for each credible failure mode. The reconciliations for flood and seismic conditions are based on the expected performance of the structure with respect to events of different recurrence intervals. The results of the reconciliations are APF and risk quantification for credible failure modes in static, flood, and seismic conditions. For credible failure modes in static conditions, the LOP can be managed to improve the APF or the FS and thus the risk. Mitigation measures can be considered in revised analyses and the residual risks estimated. The application of this strategy is demonstrated.

Introduction Due to the physical and chemical characteristics of the waste materials, mining is inherently adverse to the

1289

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA health and safety of the environment and communities (HSEC). How these waste materials are managed determines the potential for failure and adverse consequences. The management of the risk posed by the potential for adverse consequences is a primary responsibility for owners, and in recent years has become as important as profitability. Effective management of these risks is possible if they are well understood and can be quantified and clearly communicated to technical professionals and corporate administrators, who can then make informed decisions with respect to priority, methodology, and resource allocation. This paper presents a comprehensive risk management strategy that combines risk assessment with potential failure mode analysis, with the objective of identifying and quantifying the global and discrete risks posed by critical mine waste infrastructure, as well as the sources of risk. This information can then be used to identify and assess specific risk mitigation measures in terms of better practices (e.g., engineering or operation), improved stability (e.g., berms), alleviation of consequences (e.g., protection of public infrastructure), control of initiating mechanisms (e.g., inverse filters), and so on.

Quantitative risk assessment This work uses the risk assessment method of Chovan et al. (2021), which is a comprehensive, quantitative approach based on empirical relationships developed by Silva et al. (2008) between the factor of safety, the level of engineering and the probability of failure. In brief, the risk assessment consists of: 1. Selecting a critical section of a structure for assessment and conducting a stability evaluation. 2. Quantifying the engineering. 3. Calculating the probability of failure. 4. Evaluating the consequences of failure. 5. Estimating the risk. The stability evaluation produces a factor of safety (FS) under static conditions. The engineering is quantified using the Level of Practice (LOP), which consists of an objective evaluation of 45 elements divided into three main categories, Design, Construction as well as Operation & Monitoring. Design is further divided into subcategories of Investigation, Laboratory Testing, as well as Analysis and Documentation. Operation and Monitoring includes Performance. The purpose of the LOP is to quantify the quality and completeness of the engineering, which determines the level of uncertainly associated with the FS. The LOP varies from I to IV, respectively, Very High, High Good, Poor, with intermediate levels, e.g., Ib, IIb. The FS and LOP are used to calculate the annual probability of failure (APF) using Figure 1 or corresponding equations. As shown, the APF increases with LOP and FS. These relationships are based the

1290

A COMPREHENSIVE RISK MANAGEMENT STRATEGY analyses of over 75 structures, including TSF dikes (Silva et al., 2008). The application of this method to TSF is verified in Chovan et al. (2021). This method also provides scores for the categories and subcategories that are quite useful for determining which elements of LOP can be improved to reduce the APF. Alternatively, stabilization measures could be used to increase the FS and thereby reduce the APF. Table 1 is used to provide a Probability Rating (PR) from the APF. The potential consequences of failure are evaluated based on the HSEC Risk Analysis Chart (Agnico Eagle Mines 2020), which classifies the consequences in terms of the effects on health, safety, environment and community, and a Consequence Rating (CR) is selected. CR values range from 1 (Negligible) to 5 (Extreme). The Risk Rating (RR) is a product of the PR and the CR and is used to inform the corporation and communities of interest of the risk associated with critical mine waste infrastructure in a general manner. First, a Risk Factor is calculated at the product of PR and CR (Risk Factor = PR · CR). Secondly a Risk Rating, RR, is selected per Table 2. The RR varies from 0, Negligible to 4, Very High.

Annual Probability of Failure (APF)

1.0E-08 1.0E-07 Level I

1.0E-06

Ib

II

Dike 1

1.0E-05

IIb

1.0E-04 III 1.0E-03

IIIb

1.0E-02 Level IV

1.0E-01 1.0E+00 1.00

1.10

1.20

1.30

1.40

1.50

1.60

1.70

1.80

1.90

2.00

2.10

Factor of Safety (FS)

Figure 1: APF as a function of LOP and FS (from Chovan et al., 2021)

1291

2.20

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 1: Probability categories and ratings (Chovan et al., 2021) Probability category

Estimated annual probability of occurrence

Probability rating, PR

Negligible

£ 1×10-6

0

Very low

>

1×10-5

1

Low

> 1×10-5 to 1×10-4

2

Moderate

> 1×10-4 to 1×10-3

3

High

> 1×10-2 to 1×10-2

4

Very high

1×10-6 to

>

1×10-2

5

Table 2: Risk ratings (Agnico Eagle Mines, 2020) Probability Rating (PR) Consequence rating (CR)

Negligible (0)

Very low (1)

Low (2)

Moderate (3)

High (4)

Very high (5)

Extreme (5)

–

5

10

15

20

25

Major (4)

–

4

8

12

16

20

Moderate (3)

–

3

6

9

12

15

Minor (2)

–

2

4

6

8

10

Negligible (1)

–

1

2

3

4

5

RR

0 Negligible

1 Low

2 Medium

3 High

4 Very High

Potential failure mode analysis The purpose of potential mode analysis (PFMA) is to identify specific mechanisms that, if initiated, could adversely affect the stability of a structure, assess the credibility of these mechanisms with respect to structure-specific characteristics and conditions, and to estimate the relative probabilities of occurrence of these mechanisms. The PFMA method used here is based on World Bank (2021) and consists of the assessment of potential failure modes based on initiating mechanisms. Failure modes are the general means by which structures can fail. In the case of TSF these are mainly slope instability, internal erosion, overtopping and surface erosion. However, there may be other modes specific to a structure or site. Initiating mechanisms are the physical phenomena that can induce specific failure modes. For example, slope instability (a failure mode) may be induced under static conditions by the rupture of a soft foundation layer (an initiating

1292

A COMPREHENSIVE RISK MANAGEMENT STRATEGY mechanism). A failure mode may have multiple initiating mechanisms. The method used here consists of three steps: 1. Identification of possible failure modes. 2. For each failure mode: a. Identification of the potential initiating mechanisms. b. Assessment and classification of the initiating mechanisms in terms of credibility, estimated magnitude of consequences should failure occur, and adequacy of available information per Table 3. If a mechanism is deemed credible, then Step 3 is conducted. If a mechanism is deemed not to be credible, then the analysis of that mechanism is terminated. If additional information is required to assess a mechanism, recommendations to obtain this information are formulated and this mechanism may be excluded from analysis for the time being. However, this exclusion should be noted. c. Factors contributing to or inhibiting the mechanisms are evaluated and an appropriate Probability Ratings (Table 1) are selected. 3. Select a Probability Rating for the failure mode. This may be equal to the highest PR of the associated mechanisms or may be the next highest PR when more two or more mechanisms have the highest PR but not more than 5. For the PFMA, when applying the Probability Ratings in Table 1, the mechanisms are rated relative to other mechanisms of concern, so consider the Probability Categories rather than the Estimated Annual Probabilities of Occurrence. Major mechanisms, Category I per Table 3, should have PR values of 3 to 5. Minor mechanism should have PR values of 0 to 2.

Table 3: Initiation mechanism categories (based on World Bank, 2021) Category

Characterization

Description

I

Major mechanism

Considers potential for occurrence, magnitude of consequence, and likelihood of adverse response (physically possible, fundamental flaw or weakness identified, conditions reasonable and credible).

II

Minor mechanism

Of lesser significance and likelihood than category I.

III

Additional information or analysis required

IV

Not credible

Lacked information for confident judgment and because action may be required, these may be highlighted. May be that physical possibility does not exist, information eliminated concern, or so remote a possibility as to be noncredible or not reasonable.

1293

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA As noted in Table 3, an initiating mechanism may be considered not credible when it is physically impossible or an unreasonable possibility. Act of war or terrorism can generally be deemed not credible. In any event, there are no reasonable mitigation measures for this mechanism.

Reconciliation of the RA and PFMA The risk assessment produces the following for each structure considered: 1. A LOP score that can be considered in its totality or per elements of Design, Construction, etc. 2. An APF and a PR based on the static factor of safety. The PFMA produces, for each structure considered: 1. Potential failure modes. 2. Credible initiating mechanisms for each potential failure mode. 3. Relative probabilities of occurrence for the initiating mechanisms. 4. Relative probabilities of occurrence for the potential failure modes. These analyses can be reconciled in terms of static, flood and seismic conditions. The risk assessment is mainly associated with static conditions, so flood and seismic conditions are treated separately because of their association with event recurrence intervals.

Static conditions To reconcile the risk assessment with the PFMA, the APFi for each potential failure mode (where there are N modes) is estimated as the sum of the PR values for the credible mechanisms for that mode (SPRMODEi), divided by the sum of the PR values for all failure modes (SPRMODE1-N), times the APF for the structure from the risk assessment: 𝐴𝑃𝐹$ = ∑

∑ '()*+, . '()*+, /01

∙ 𝐴𝑃𝐹 (Equation 1)

Subsequently, new PR values for the failure modes can be selected using the annual probabilities of occurrence in Table 1, then RR for the failure modes can be selected from Table 2. Note that the CR values of the failure modes are constant.

Flood and seismic conditions Under flood conditions, the typical potential mode of failure is slope rupture due to: 1) overtopping caused by an event exceeding the design capacity of the spillway; 2) elevated phreatic levels in a downstream slope; or 3) an engineering defect. Assuming that failures due to the later mechanism are captured in the risk assessment (LOP), the first two are of concern here. If it is assumed that a storm event that causes some level overtopping of the dike due to insufficient spillway capacity will result in failure, then the APF of the

1294

A COMPREHENSIVE RISK MANAGEMENT STRATEGY structure due to this mechanism is the reciprocal of the recurrence interval of that event. For many structures, this recurrence interval is less than the Probable Maximum Precipitation (PMP) as initially designed or as modified by climate change. Using sensitivity analysis, it is possible to estimate the phreatic levels in the downstream slope of a dike associated with different storm events and the resulting factors of safety. The APF of this mechanism is the reciprocal of the recurrence interval of the storm event associated with a FS of 1. These two mechanisms (overtopping and elevated phreatic surface) are considered independently. See example in Figure 3. The FS values are for the downstream slope and vary with storm recurrence interval, due to elevated phreatic levels, and in this example instability (FS=1) occurs at a recurrence interval of 6,500 years. Slope failure due an elevated phreatic surface will occur before overtopping. Instability, due to overtopping, occurs when the basin level attains the crest (el. 318 m), which occurs at a recurrence interval of 8,000 years. The APF associated with the flood condition is thus the inverse of the lower recurrence interval, so 1.54 × 10-4 in this example.

1.75

320 319

1.50 FS

Crest Elevation 318 m

318

1.25 317

Bassin Elevation 1.00 1

10 100 1,000 Storm Recurrence Interval

Basin Elevation (m)

FS

316 10,000

Figure 3: Factors of safety and basin elevations associated with storm recurrence intervals

Figure 4 is an example for seismic conditions, pseudostatic factors of safety (FSPS) are shown for different recurrence intervals. The value of FSPS is 1 at about 4,000 years. However, given that pseudostatic analysis is inherently inaccurate, numerical analysis was also conducted for different recurrence intervals and the corresponding horizontal displacements of crest are also shown. Assuming that the maximum allowable horizontal displacement of the crest is 80 cm, then the APF of this mechanism is less than (1/10,000), so 1×10-4.

1295

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA In this example, the pseudostatic FS was used because liquefaction was not an issue. Had liquefaction been an issue, the post-shaking FS could have been used when the seismic event was sufficient to induce significant strength loss in any liquefiable materials present.

100 90

Pseudostatic FS

1.50

FS

1.25

80 70 60

1.00

50

0.75 0.50 Displacement

0.25 0.00 1

10 100 1,000 Earthquake Recurrence Interval

40 30 20

Crest Displacement (cm)

1.75

10 0 10,000

Figure 4: FS and crest displacement associated with different earthquake recurrence intervals

Example application – Dike 1 The following example is based on an actual structure. However, several elements have been modified for the purposes of this paper. Dike 1, a perimeter embankment of a TSF, is shown in Figure 5. At a critical section, Dike 1 is 6-m-high and occupies a low-lying depression in the natural terrain. It is a rockfill dike with an upstream bituminous membrane for impermeability. There is a 1-m-thick transition zone (0 to 50 mm) between the membrane and rockfill (0 to 200 mm). The upstream and downstream slopes are inclined at 1.5:1 (H:V) and 2:1, respectively. The crest is at elevation 318 m, is 8-m-wide and is topped with fined crushed rock (0 to 75 mm). The foundation consists of up to 2.5 m of medium stiff silty clay of moderately high plasticity overlying bedrock of excellent quality. The upstream geomembrane is embedded into the silty clay to a depth of 1 m. Dike 1 retains tailings, and the deposition plan indicates a final tailings elevation of 317.5 m against the dike. The pond of the TSF is located at a nominal distance of 200 m from Dike 1. However, during extreme storm events, there may be temporary ponding against the dike. The operational and emergency spillways of the TSF were designed to safely evacuate the PMP. However, recent analysis considering the probable effects of climate change indicate that the PMP will

1296

A COMPREHENSIVE RISK MANAGEMENT STRATEGY increase appreciably in the coming years. Additionally, it was revealed that the watershed reporting to the TSF was actually underestimated in the design of the spillways.

Figure 5: Photograph of Dike 1

Risk assessment of Dike 1 The risk assessment for Dike 1 was completed by the Engineer of Record in collaboration with site personnel. The assessment considered a FS (static) of 1.55 and the LOP given in Table 4. The LOP for the different elements vary from Good to Very High, with a total LOP of High-Very High. A few elements that negatively affected the LOP were: 1) The investigation and laboratory testing of the clay layer in the foundation were not comprehensive; 2) Only weekly inspection was provided during construction; 3) Asbuilt plans and reports were not complete; 4) Climate change was not considered in the water management; and 5) Revised analysis of the capacity of the emergency spillway indicates that its capacity is not sufficient for the PMP due to an error in the calculating the area of the watershed. The resulting APF is 2.31×10-5 (see Figure 1) and gives a PR of 2 (Low) per Table 1. Based on the potential consequences of failure (not described here), the Consequence Rating (CR) is 5 (Extreme) and the resulting Risk Factor is 10, indicating a Risk Rating (RR) of 2 (Medium) per Table 2. Although the FS meets regulatory requirements, the acceptance of this level of risk is a corporate decision. Based on the risk assessment only, the level of risk associated with this structure may be reduced by increasing the FS (e.g., constructing a berm) or improving the LOP (e.g., additional sampling and testing of the clay layer). However, conducting a PFMA may provide additional information for risk management.

1297

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 4: Summary of risk assessment for Dike 1 Variable

Value

FS (Static)

1.55

Level of practice

Ib (1.69)

Notes High – Very High

Design: investigation

I

Very High

Design: Laboratory testing

I

Very High

Design: Analysis/Documentation

Ib

High – Very High

Construction

II

Good

Operation and maintenance

I

Very High

APF

2.31×10-5

Potential failure mode analysis of Dike 1 The PFMA was conducted by the Engineer of Record, revised by the Design Engineer, and reviewed by the Independent Review Board. It is partially summarized on Table 5. The information in this table is abbreviated due to space limitations. The actual PFMA report for a site with seven distinct structures is over 30 pages including text and tables. From Table 5, one can ascertain which modes and mechanisms of failure are credible, and why so, as well as their relative probabilities of occurrence. As shown, there is some probability of internal erosion due to seepage though gap in the liner and a lesser probability of erosion due to seepage below the liner and through the clayey foundation layer. The factors that would contribute to or inhibit the initiating mechanisms are listed and relative probabilities are assigned using the PR values of Table 1. As noted, what is important for the PFMA are the relative probabilities of the initiating mechanisms (using the Probability Categories) not the estimated probabilities in Table 1. Overtopping was identified as a potential failure mode due to the spillway not having sufficient capacity for the design PMP or the possible PMP caused by climate change or the possibility that the spillway would be blocked by debris. However, there is a lower possibility that the spillway will be blocked due to daily inspection by site personnel. Another potential failure mode was slope failure initiated by rupture of the silty clay below the downstream of the dike. Recall that the risk assessment indicated that additional sampling and testing of this clay was warranted. Surface erosion was not identified as a credible mode of failure because of the nonerodable nature of the rockfill given the inclination of the downstream slope and the relatively low height of the dike, 6 m.

1298

A COMPREHENSIVE RISK MANAGEMENT STRATEGY Table 5: Summary of PFMA of Dike 1 Potential failure mode

PR mechanism

Initiating mechanisms

Notes

Seepage through possible gap in liner.

Contributing: Sufficient gradient, no full-time construction observation, transition layer does not meet filter requirements. Inhibiting: No gaps observed at end of construction inspection. Class: II (Table 3).

2

Seepage through foundation below liner.

Contributing: High gradients at bottom, no fulltime construction observation. Inhibiting: Silty clay of moderate plasticity not readily eroded, embedment of liner into clay. Class: II

1

Storm exceeding spillway capacity.

Class: I

3

Spillway blocked by debris.

Class: II

2

Settlement of crest.

Class: Not credible

0

Seismic densification of rockfill shell.

Class: Not credible

0

Slope Instability

Failure through clay at toe of dike.

Class: I

3

Surface Erosion

Erosion of rockfill shell.

Class: Not credible

0

Internal erosion

Overtopping

Reconciliation of the risk assessment with the PFMA To reconcile the PFMA with the risk assessment, the PR values for each potential failure mode are summed, the APF partitioned per Equation 1, and mode-specific PR and RR values are determined from Tables 1 and 2, respectively. Recall that the CR of Dike 1 is 5 for all potential modes of failure. The results are given in Table 6. Note, for internal erosion, the APF is (3/1) · 2.31×10-5, so 6.3×10-6. As mentioned, flood and seismic conditions are treated separately. For flood conditions, assume the Figure 3 applies to Dike 1, except that the FS is not affected by the basin level due to the presence of the liner. Then the only credible failure mechanism is overtopping due to the 8,000-year event. The APF of this mode is thus 1.3×10-4 and thus PR of 3, CR of 5 and RR of 15 (Moderate). Figure 4 can be used for the seismic reconciliation of Dike 1, specifically the horizontal displacements of the crest. The maximum allowable displacement of the crest is 80 cm and dynamic numerical analysis indicated a displacement of 75 cm due to the 10,000-year event. Given the approximate nature of numerical analysis, 75 cm and 80 cm are equivalent and the APF of Dike 1 in seismic conditions is 1×10-4, giving a PR of 2, CR of 5 and RR of 10 (Low). Note that overtopping occurred as a potential failure mode in the static and flood analyses. In the

1299

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA risk assessment, it was due to engineering and the in the PFMA it was due engineering and climate change. In this case the analysis with the lower APF should be discarded. Table 6 provides general information for the prioritization of risk management strategies, while Tables 4 and 5 provide specific information on how risks can be managed. Table 6: Reconciliation for Dike 1 Potential failure mode

SPR (mechanisms)

APR (mode)

PR (mode)

RR (mode)

Static – Internal erosion

3

6.3×10-6

1

5 (Very Low)

Static – Overtopping

5

1.1×10-5

2

10 (Low)

Static – Slope instability

3

6.3×10-6

1

5 (Very Low)

Flood – Overtopping

–

1.3×10-4

3

15 (Moderate)

Seismic – Crest displacement

–

1×10-4

2

10 (Low)

Strain-softening and static liquefaction There are some advanced soil behaviours that can be very important for stability. These are implicit in the risk assessment where the technical review of the field investigation, laboratory testing and design by an Engineer of Record and an Independent Review Board are part of the LOP evaluation and thus the APF. They are explicit in the PFMA, which is also subject to technical review. If of concern, such phenomena could be credible initiating mechanisms and must be dully considered in the PFMA.

Conclusion As an essential tool for the responsible management of critical mine waste infrastructure, it is very important that comprehensive risk assessment strategies be implemented throughout the industry. It is also very important that such tools be transparent and readily available. The authors welcome comments and suggestions for the improvement of this strategy.

References Agnico Eagle Mines. 2020. HSEC Risk Analysis – Consequence and Probability Criteria. Chovan, K.M., Julien, M.R., Ingabire, E.-P., James, M., Masengo, E., Lépine, T. and Lavoie, P. 2021. A risk assessment tool for tailings storage facilities. Canadian Geotechnical Journal 99(999): 1898–1914. Silva, F., Lambe, T.W. and Marr, W.A. 2008. Probability and risk of slope failure. Journal of Geotechnical and Geoenvironmental Engineering 134(12): 1691–1699. World Bank. 2021. Technical Note 5 – Potential Failure Mode Analysis. 25 pages.

1300

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Did Someone Say ALARP? If So, How Do We Get There? Malcolm Barker, GHD, Australia Manoj Laxman, GHD, Australia Jayamini Methiwala, GHD, Australia

Abstract The International Council on Mining and Metals (ICMM) publication, “Tailings Management Good Practice Guide (ICMM May 2021)” and the Global Industry Standard on Tailings Management (GISTM), require consideration of the As Low As Reasonably Practicable (ALARP) principle for risk management when designing new tailings storage facilities and for continuing construction, operation, and maintenance of existing facilities. How are risk assessment techniques being applied in the application of ALARP? Is there a way to formalize the process for ALARP justification using quantitative, semiquantitative, qualitative, or experiential methods? According to ICMM, credible failure scenarios that have elevated levels of risk may require mitigation measures to reduce risk. The level of acceptable risk is defined by each Operator using ALARP or by local regulatory requirements, as applicable. ALARP requires that all reasonable measures be taken with respect to “tolerable” or acceptable risks to reduce them even further until the cost and other impacts of additional risk reduction are grossly disproportionate to the benefit. This paper will explore the application of the ALARP principle in both the ICMM publication and GISTM, together with the ANCOLD 2022 Risk Assessment Guidelines and other sources where the ALARP principle is considered in order to provide a process for tailings dams.

Introduction According to the Collins dictionary, “a standard is a level of quality or achievement, especially a level that is thought to be acceptable.” The Global Industry Standard on Tailings Management (GISTM) has the ultimate goal to achieve zero harm to people and the environment with zero tolerance for human fatality. Operators are to take responsibility and prioritize the safety of tailings facilities, through all phases of a facility’s lifecycle, including closure and post-closure. It also requires the disclosure of relevant information to support public accountability.

1301

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA In order to achieve a level of acceptability, both GISTM and ICMM require consideration of the As Low As Reasonably Practicable (ALARP) principle for risk management when designing new tailings storage facilities and for continuing construction, operation, and maintenance of existing facilities. There are various methods for presenting risk, including quantitative, semiquantitative, qualitative, or experiential methods that mine operators can use to evaluate risk reduction to achieve an ALARP position for their facilities. The question is: “Can any one of these methods be used for facilities that may range from Low to Extreme consequence, or is there a way of logically applying the methods to provide the best outcome for achieving cost-effective ALARP facilities, considering both the uncertainty and the complexity of the facilities?” It is important to note that there is no clear guidance in GISTM or ICMM as to which method of risk assessment should be used for evaluating ALARP. The following is aimed at providing a framework for ALARP, including establishing basic understanding of risk assessment and ALARP, from which to formulate a matrix that can be used for evaluating ALARP. Much has been written on the use of risk assessment for reducing risks; however, in order to evaluate whether the risk for a facility has been reduced to ALARP, it is important to understand the inputs and process that can be used to evaluate ALARP. This paper will not discuss the inputs and process in detail, but the following are briefly presented as the basis and process for ALARP.

Aggregated versus disaggregated Given the complexity of dam systems and the many structures and sub-structures considered in undertaking failure mode analysis for a dam system, multiple techniques are required in considering ALARP. ANCOLD (2022) considers there to be two fundamental approaches: 1. A disaggregated approach; and 2. An aggregated or global approach. The aggregated approach considers the risk profile as a whole and is appropriate where the risk analysis provides quantified data for both the financial and life risk, i.e., annual probabilities of failure with corresponding loss of life and financial/economic losses for each failure mode, which can be aggregated and used as follows: • Evaluation of the “level of risk” in comparison to the limit or threshold of tolerability using the Societal Risk (FN) plot. • Estimation of the Cost to Save a Statistical Life (CSSL) economic metric, and thereby assessment of gross disproportion in cost of risk reduction. • Estimation of other Cost/Benefit (C/B) ratios.

1302

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE? It should be noted that risk analysis methods that are not quantitative are not amenable to aggregating the risks as they do not provide a means of combining the risk ratings. Only quantitative risk analysis methods can be aggregated for comparison with existing guidelines. The disaggregated approach is the fundamental approach that can be used for both qualitative and quantitative risk assessment techniques when evaluating whether risks are ALARP. Disaggregation aims to break down the total risk profile to identify controls or risk mitigations to ensure that all reasonably practicable measures to reduce risk are considered in assessing ALARP. ANCOLD (2022) identifies two forms that disaggregation can take: 1. Through development of event or fault trees or other representations of failure mechanisms. The overall system should be broken down into components and the ALARP test applied to each component and the corresponding failure modes and mechanisms in turn to ensure that all reasonably practicable measures to reduce risk have already been taken or proposed. 2. Through Prevention/Control/Mitigation (bow-tie) analysis. For a risk to be ALARP it is necessary to demonstrate that all reasonably practicable measures have been taken under each of prevention, control, and mitigation. In the disaggregated approaches, judgement that risks are or would be ALARP is often but not necessarily qualitative (ANCOLD, 2022). This involves identifying: • What are the risks? • What could be done to eliminate or reduce the risks further? • What are the constraints on the practicability of reducing risks further? This paper provides guidance regarding the approach for documenting the aggregated and disaggregated approaches when assessing whether risks are ALARP.

Common risk assessment methods According to the ICMM (2021), “risk analysis involves the characterization of what is known and what is uncertain about the present and future performance of an existing or planned tailings facility. During risk analysis, the likelihood of the specific potential failure mode loading condition, the likelihood of an adverse structural response, and the magnitude of the consequences are estimated for each potential failure mode.” Note: Emphasis added by author for to highlight the triplet required to evaluate risk. There are a number of methods for risk assessment including the following.

Failure modes effects analysis and failure modes effects and criticality analysis Failure modes effects analysis (FMEA) is an inductive method of analysis where particular faults or

1303

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA initiating conditions are postulated and the analysis reveals the full range of effects of the fault or the initiating condition on the system. FMEA can be extended to perform what is called failure modes, effects, and criticality analysis (FMECA). In a FMECA, each failure mode identified is ranked according to the combined influence of its likelihood of occurrence and the severity of its consequences. Additional input for controls, means of detection, and comments provide useful data for evaluation of ALARP. The FMEA or FMECA must be used as the basis (foundation) for all risk analysis, in order to determine the failure modes with the highest risk for which more detailed analyses are completed for the system in quantified approaches. Furthermore, it is important for the personnel completing the FMEA or FMECA to conduct a site inspection, otherwise the risk assessment may be fatally flawed (ANCOLD, 2022). Table 1: Risk assessment methods Risk assessment method Risk matrix

Bow tie

Fault tree analysis (FTA)

Event tree analysis (ETA)

Pros

Cons

-

Used across many industries

-

Simplicity using limited data

-

Forms the basis for the semi-quantitative method

-

Graphical representation of the dynamics of the hazard

-

Useful for evaluating controls

-

Structured thinking

-

Good for communicating and training

-

Can be used for Safety Case

-

Structured thinking

-

Qualitative and quantitative analysis

-

Useful for evaluating controls

-

When quantified valuable to define the minimal cut set defining failure for evaluation of ALARP

-

Can be used for Safety Case and for semi quantitative risk assessments

-

Accounts for common cause and common mode failures

-

Useful graphical presentation

-

Can be used for identifying controls

-

Structured logic for the failure pathway

-

Incorporates aggregated and disaggregated approaches for Risk

-

Can be used for the full range of hazards

1304

-

Most of the matrices do not extend the likelihood to events rarer than the 1 in 100 or 1 in 200 AEP event

-

Cannot be used to aggregate risks

-

Not normally quantified

-

Aggregates failure modes but can be broken down to disaggregate failure modes

-

Can be complex development when using commercial software and for considering multiple failure modes

-

Failure modes when aggregated can result in loss of transparency

-

Does not capture the full range of hazards

-

Errors can be introduced when using Excel spreadsheets for calculation

-

Common cause and common mode failures for combined failure modes incorporated at the end of the analysis

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE? Risk assessment method

Pros

Cons

Semiquantitative (SQRA)(1)

-

Inherently considers uncertainty

-

Failure modes cannot be aggregated

-

-

Figure

Confidence can be assigned in the analysis

Generally descriptors for failure modes and not quantified

-

Can be used with numeric values for evaluating risk.

-

Only disaggregated approaches for risk

Hybrid risk assessment (HRA)

-

Limits calculation of failure modes to critical event

-

Does not provide best estimates for risk but rather a lower bound

-

Provides societal and individual risk data for ALARP evaluation

-

-

Incorporates aggregated and disaggregated approaches for risk

Must be used with caution for detailed analysis and remedial options design for ALARP

-

Uncertainty dealt with using sensitivity analysis and not explicitly

-

Full range of hazards evaluated

-

Complex mathematics

-

Can incorporate uncertainty

-

-

Provides societal and individual risk data for ALARP evaluation

A lot of analysis is required for conditional probabilities and downstream consequences

-

Incorporates aggregated and disaggregated approaches for risk

-

Requires a level of experience and expertise that is limited at present

-

Not generally accepted by owners and practitioners

-

Must be completed by a team with a minimum of ITRB, ROE, Designer, Client

Quantitative methods (QRA)

Note: 1. USBR/USACE July 2016 – See Figure 1 for an example

Figure 1: Incremental risk matrix example (USBR/USACE 2016)

Risk evaluation Risk evaluation compares the outcomes of risk analysis for existing conditions to determine if risks are within acceptable limits, whether present risk measures and controls are adequate, and what additional

1305

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA alternative risk reduction measures could be considered (ICMM, 2021). According to ICMM, the following typical “criteria” should be used to evaluate the existing risk, in order to determine what risks are within normal operating limits, and to consider any mitigation options and recommendations for actions deemed justified. These form part of the ALARP evaluation for the facility: • Robustness of design. • Past and future performance monitoring. • Site context. • Practicality of any remediation. • Guidelines from regulatory agencies, governing bodies, and other industries associated with tailings facility safety. • Corporate governance for normal operating limits. • Environmental, social, cultural, ethical, political, and legal considerations.

Uncertainty Uncertainty is the result of imperfect knowledge and is not the result of risk assessment. There are two main sources of uncertainty with respect to risk analysis, as shown in Figure 2, which need to be considered when evaluating whether the risk for a facility is ALARP. Aleatory uncertainty (natural variability) and epistemic (knowledge) uncertainty are the principal forms of uncertainty of concern to engineers because they pertain to matters of engineering (Hartford et al., 2008).

Figure 2: Sources of uncertainty (ICOLD, 2023)

1306

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE? For the purpose of ALARP evaluation, these can be translated into confidence levels, as shown in Table 2 (FERC, 2021). These levels of confidence are also used by FERC for Dam Safety Risk classification and ALARP evaluation, as shown in Table 2. It is of note that FERC will only consider ALARP when the risks are low, which is an onerous requirement showing risk aversion and precautionary principle. Sources of uncertainty and potential actions that could be taken to reduce that uncertainty should both be identified when evaluating whether the risk for facilities is ALARP. Table 2: Confidence descriptors for ALARP (FERC, 2021) High Confidence: The individual/team is confident in the assigned order of magnitude descriptor and it is unlikely that additional information would change the estimate to the point where the decision to take (or not take) action to reduce risk or reduce uncertainty would change Moderate Confidence: The individual/team is relatively confident in the assigned order of magnitude descriptor, but key additional information might possibly change the estimate to the point where the decision to take (or not take) action to reduce risk or reduce uncertainty may change Low Confidence: The individual/team is not confident in the assigned order of magnitude descriptor and it is entirely possible that additional information would change the estimate to the point where the decision to take (or not take) action to reduce risk or reduce uncertainty could change.

Controls It is important when considering ALARP to provide controls that are evaluated for each failure mode. A control is any measure (process, device, practice, or required action) that directly enables an opportunity, or prevents or mitigates a threat, to meet the objectives. Performance of the control is specifiable, measurable, and verifiable (auditable). The last-mentioned is important for understanding whether a control is effective. According to GISTM Principle 6, Critical controls are to be in place for planning, building, and operating tailings facilities to manage risk at all phases of the tailings facility lifecycle, including closure and post-closure.

ALARP evaluation The objective for the design process throughout the lifecycle of a tailings facility should be to limit credible failure modes, either to having no credible failure modes or, where credible modes cannot be eliminated, ensuring that potentially catastrophic credible failure modes are managed using the ALARP approach through the phases of the facility’s lifecycle where they are present (ICMM, Section 3.2.4).

What is ALARP? ALARP requires that all reasonable measures be taken with respect to “tolerable” or acceptable risks, so as to reduce them until the cost and other impacts of additional risk reduction are grossly disproportionate to

1307

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA the benefit (GISTM, 2020). The level of acceptable risk is defined by each Operator using ALARP or by local regulatory requirements, as applicable. It is noted from ICOLD (2023) that there are two fundamental principles from which tolerable risk guidelines are derived, as follows: • Equity. The right of individuals and society to be protected, and the right that the interests of all are treated with fairness, with the goal of placing all members of society on an essentially equal footing in terms of levels of risk that they face. • Efficiency. Efficiency is the need for society to distribute and use available resources so as to achieve the greatest benefit. It is important to consider both of these principles when establishing whether risks are reduced to ALARP. According to ICMM, factors involved in applying ALARP after a remedial action and/or enhanced operational practices or surveillance have been implemented include: • Application of relevant good practice. This can be applied to all aspects of the facility including operation, maintenance, surveillance, construction, and emergency planning. • Evaluation of the level of incremental risk in relation to the established risk guidelines. • Evaluation of the cost-effectiveness of the risk reduction measures in relation to likelihood and/or consequence. • Taking into account the remaining life of the facility and potential alignment with closure planning, which may reduce likelihood and/or consequence. • Acknowledging societal concerns as revealed by consultation with the community and other stakeholders. • Other factors such as: o

consideration of standards-based approaches;

o

benchmarking;

o

direct business impacts;

o

constructability;

o

implementation schedule; and

o

environmental consequences.

Each of these factors is considered in the ALARP approach proposed for facilities varying from Low to Extreme risk.

1308

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE?

ALARP documentation requirements The following documentation is recommended as a minimum to develop a sound basis for the ALARP process (Robilliard and Sih, 2018). While the referenced paper deals with water dams, the documentation requirements are nevertheless valid for tailings facilities and have been modified as appropriate. Documentation type Dam safety

Adequacy of documentation (Robilliard and Sih, 2018) -

Assessment of existing risk

-

Defensible documents

-

Risk reduction options

-

Is there a well-documented tailings facility management system that is regularly reviewed? Is the safety management system effective, including the Trigger Action Response Plans (TARPs)? Is the tailings safety management system sustainable? (e.g., is there training? Is there succession planning?) Have all credible failure modes been identified? Was there a screening process to include/exclude failure modes and has this been documented? Was the assessment made on sound information? Did the analysis team include members with specialist expertise in accordance with the requirements of GISTM? For semi quantified, hybrid and quantified risk analyses, was the risk analysis (probability and consequence) rigorous? For qualitative risk matrices, is the matrix appropriate for low likelihood probabilities applicable to the risk assessment for the credible failure modes? Is there a current safety review of the tailings facility? Does the safety review address all credible failure modes? Has there been an independent technical review of the safety review? Are there detailed assessments that defend items that are taken as ALARP? (e.g., slope stability analysis for a slope that meets current design practice)? Have all potential critical controls for preventing failure modes been considered and documented? What is the level of confidence in these controls? Have all documents to defend the ALARP position for individual failure modes been collated? Are the regulator or other state guidelines available for consideration in the ALARP process? Have structural and non-structural risk reduction options been identified and documented for each credible failure mode? Has each option been evaluated for practicability? (Note: for impracticable options, reasons for not continuing must be documented.) Has the cost for each option been estimated? This is appropriate when using a quantified risk assessment approach. Has the risk reduction for each option been quantified? This can be for any of the risk assessment approaches.

The process from the documentation review to selection of the risk approach is shown in Figure 3 and entails a decision as to whether to proceed with the ALARP evaluation, or whether the ALARP evaluation is to be qualified owing to limited data. This is followed by an initial risk position from which to determine whether to proceed with the qualitative or quantitative methods of risk analysis.

1309

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 3: Selection process for risk assessment approach

Risk assessment approaches The following matrix provides an indicative way forward for the use of the various risk assessment methods when evaluating ALARP. This is the authors’ viewpoint and is not detailed in current literature and does not negate the use of more detailed risk analyses for the low risk facilities where data is available and there is a requirement to comply with the tailings storage facility owner’s risk management framework or a regulator’s requirements. The more detailed risk analyses are preferred where there is a population at risk; however, where a tailings facility has no population at risk, then the simplified analyses are appropriate, particularly since the cost benefit ratios are limited in value for the ALARP evaluation. Table 3: Proposed risk assessment approaches for ALARP evaluation Consequences Likelihood Description Very High

High Moderate Low Remote

Likelihood AEP Range 1E-2 to 1E-3 1E-3 to 1E-4 1E-4 to 1E-5 1E-5 to 1E-6 1E-6 to 1E-7

Low

Significant

High

Medium SQRA

High Hybrid/ QRA

Medium SQRA

Medium SQRA

High Hybrid/ QRA High Hybrid/ QRA

Low Qualitative Matrix Low Qualitative Matrix Low Qualitative Matrix

Medium SQRA Low Qualitative Matrix Low Qualitative Matrix

Medium SQRA Medium SQRA Low Qualitative Matrix

Very High

Extreme

Extreme QRA

Extreme QRA

Extreme QRA

Extreme QRA

High Hybrid/ QRA High Hybrid/ QRA Medium SQRA

Extreme QRA High Hybrid/ QRA High Hybrid/ QRA

Note: Consequence categories based on Table 1: Consequence Classification Matrix (GISTM 2020)

1310

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE?

ALARP factors The process for the ALARP decision is shown in Figure 4 and entails consideration of the factors described below.

Figure 4: ALARP decision factors

Cost effectiveness The analysis of cost effectiveness can only be evaluated when a quantitative risk assessment is used for the ALARP evaluation. The use of cost benefit analysis (CBA) or Cost to Save a Statistical Life (CSSL) ratios are used to determine whether the costs are grossly disproportionate to the benefits achieved. See ANCOLD (2022) for details of calculations. It should be noted that these ratios are often not satisfied and more reliance is placed on the remaining factors when considering whether the risk for a facility is ALARP. It is important to consider the effects of future potential increase in downstream consequence and external drivers such as climate change that could increase the risks and change the ALARP justification.

Good practice and precedent • Assess the dam and its components for compliance with current industry standards based on good engineering practice. Use an aggregated and disaggregated approach. Where there are differences from current industry standards and good engineering practice, document these deficiencies and the engineering reasons that good practice is not met. Confirmation is needed that additional investigation or analysis will not markedly reduce the risk and improve the ALARP. • Are there measures which may not greatly affect the risk position but can be implemented at relatively small cost to satisfy good engineering practice?

1311

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Determine whether Critical and Mitigating Controls have been identified, practicable and verifiable. • Is there precedence of other dams with comparable conditions and existing risk where a decision was made not to implement remedial works? A similar approach for benchmarking against other projects if possible to evaluate the effectiveness of the ALARP measures. • The level of confidence in the Risk analysis results will provide clear guidance as to whether the facility risk is ALARP. Where there is a low confidence, then this is very likely to result in additional investigation and analysis to improve the uncertainty and confidence in the analyses used to evaluate ALARP.

Existing level of risk In the case of quantified risk analysis, the societal and individual risk data is used to evaluate the existing risk prior to any consideration of ALARP. If the risk is above the limit of tolerability for the societal risk plot or individual risk appropriate to the regulator or national guidelines, this is not acceptable and is a strong justification for work to reduce the risks to ALARP. It is useful to compute the level of risk (R) below the limit of tolerability, which is defined in the equation below. For societal risk: 𝑅 = −log

𝑊𝑜𝑟𝑠𝑡 𝑟𝑖𝑠𝑘 𝑝𝑜𝑖𝑛𝑡 𝑜𝑛 𝐹 − 𝑁 𝐶𝑢𝑟𝑣𝑒 (𝐹 × 𝑁)BCD = −log 𝐿𝑖𝑚𝑖𝑡 𝑜𝑓 𝑇𝑜𝑙𝑒𝑟𝑎𝑏𝑖𝑙𝑖𝑡𝑦 10GH

Where, • R is the level of risk expressed in orders of magnitude below the Limit of Tolerability • (F×N)MAX is the product of the worst risk pair of F and N • F is the cumulative annual probability of failure with expected loss of life ≥N

For individual risk: 𝑅 = −log

𝐼𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑅𝑖𝑠𝑘 𝐿𝑖𝑚𝑖𝑡 𝑜𝑓 𝑇𝑜𝑙𝑒𝑟𝑎𝑏𝑖𝑙𝑖𝑡𝑦

This value is also used in assigning the disproportion factor for estimation of CSSL defined by ANCOLD (2022). Recommendations for the interpretation based on the considerations above are shown in Figure 5.

1312

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE?

Figure 5: Sources of uncertainty (ICOLD, 2023) In the case of semi-quantified or qualitative risk assessment the level of risk acceptance is defined by the owner of the facility using the risk matrix and can be used to judge the risk levels for each failure mode and implement measures to either reduce the likelihood by remedial works or reduce the consequences using defensive means. Efficiency of risk reduction for upgrades can be evaluated using a chart of cost of upgrade versus risk. The chart can also be used to evaluate whether there is a distinct point at which the efficiency of risk reduction reduces significantly. Direct business impact where ongoing operation of the facility may be jeopardized or there would be national or international repercussions for the business. Percent compliance with regulation standards or fallback criteria for floods or earthquake loading can be used as a means to evaluate the urgency of any remedial works or to confirm that the facility is in compliance with the regulations and that ALARP remediation is not needed.

Societal concerns Community consultation needs to carried out to evaluate acceptance of the following: • High consequences – what affect will the consequences, particularly life, social, and environmental consequences, have on the affected community and the community at large? Will the society have greater leverage for reducing risks ALARP? • Vulnerable people – who are they and where are they with respect to any potential failure scenario, and can they be moved to another location in order to eliminate the risk? • Critical infrastructure – Is there any critical infrastructure that would have far reaching consequences if impacted, for example, main highway bridges or hospitals outside of the tailings facility boundaries, etc.?

1313

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Risk equity across the system – this is a decision for the tailings facility owner to evaluate where best to implement changes in operation of the infrastructure to achieve equitable risk reduction across the system. • Environmental consequences – these are important, particularly where there are endangered species in the path of a potential failure flood zone. • Climate change – has climate change been considered in the ALARP decision with respect to changes in floods and rainfall distribution?

Upgrade implementation Consideration needs to be taken of the remaining life of facility and closure planning as to the duration and acceptability of risks. The question to be asked is: “What is the likelihood of failure within the remaining life rather than an annual probability of failure?” If the risk is close to the limit of tolerability, is there an appetite for acceptance of the risk or is there a precautionary approach that requires risks to be driven down lower, particularly for low-risk facilities? Increased risk after implementation may occur in certain instances where, for example, a raising of a dam allows passage of the design flood; however, the damage and resulting risk from the increased capacity results in a higher risk than the increased flood capacity. Increased risk during construction is an important consideration where works are likely to extend into a rain period or where the works require a lower factor of safety to be accepted while the existing infrastructure is prepared for the remedial works. For example, excavation at the toe of an embankment required for buttressing increases likelihood of slope failure and another option is needed for ALARP. Introduction of new risks where the remedial works change the configuration of a site or result in placement of material over areas that may be affected in future operations. This is particularly important where a “temporary fix” becomes a “permanent fix” and the resulting effects on the structure can lead to failure if the failure modes are not well understood. Urgency to achieve risk reduction where it is clear that a failure mode is imminent and remedial works are required to prevent the failure mode. An additional consideration is the turn out cost (TOC) that can be presented with the level of risk reduction for upgrade options, as shown in Figure 6 for an example dam. The cost effectiveness and risk reduction can be compared and used in the decision making process to determine the ALARP option.

1314

DID SOMEONE SAY ALARP? IF SO, HOW DO WE GET THERE?

Figure 6: Level of risk versus cost for each example – dam upgrade option and existing dam

Conclusion This approach for evaluating whether ALARP is met for a tailings facility has been developed with a view to using the most simplistic risk analysis approach, using a risk matrix to a fully quantified risk analysis that can include uncertainty analysis. In all of the ALARP evaluation, the question needs to be asked: “If the dam failed and there were lifethreatening consequences, could you confidently answer the question that ALARP has been achieved, or were all reasonably practicable actions completed with respect to the known event?”

References ANCOLD. 2022. Risk Assessment Guidelines. FERC. 2016. Dam Safety Risk Classification (DSRC) Table 4-1. FERC. 2021. Engineering Guidelines for the Evaluation of Hydropower Projects. Chapter 18 – Level 2 risk analysis. Hartford et al. 2008. Handling uncertainty in dam safety decision making. In USSD Conference Proceedings 2008. ICOLD. 2023. Workshop on Risk Analysis for Tailings Dams.

1315

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA International Council on Mining and Metals (ICMM). 2021. Tailings Management Good Practice Guide. Robilliard, K. and Sih, K. 2018. A logical ALARP risk procedure. Australian National Conference on Large Dams Conference 2018.

1316

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Application of Monte Carlo Simulation to the Probability Assessment in FMEA Geinfranco Villalta, Stantec, Peru Raquel Borja, Stantec, Peru

Abstract The mining industry requires risk assessment of major facilities such as tailings dams (TD) at each stage of project development. Various methodologies have been used depending on the needs, with one of the most popular applications being the failure mode and effects analysis (FMEA). The FMEA assesses the likelihood of occurrence (L) and the consequence (C) of each identified failure mode to determine the risk level (R) associated with each failure mode (FM). Subsequently, the critical FMs are selected based on specific criteria for risk classification in order to guide risk reduction efforts. The criteria evaluation form is the qualitative rating standard (e.g., low, medium, high). Two primary issues emerge. First, judgment errors may arise during the workshop-based assessment of Ls and Cs due to potential overconfidence exhibited by the experts. Second, the averaging of the scores performed by each expert to derive a final value is a deterministic analysis that overlooks the variability in expert judgments. The variability includes individual biases, such as risk aversion or risk-taking tendencies, as well as uncertainty among the scores provided by the team. It does not consider a weighted average of individual assessments. It is important to emphasize that some relevant information may be lost by considering only the averages of expert analysis. Given the limitations of deterministic analysis, several researchers have proposed tools to calibrate the behaviour of individual experts within a team. Calibration tests are utilized to measure and address overconfidence exhibited during the estimation of L or C factors. This can be fitted to obtain a stochastic representation of each of them, and then the Monte Carlo simulation can be applied to gain more information about the uncertainty in the outcomes (R). The decision maker then has a range of plausible outcomes and the probabilities of occurrence of them.

1317

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction The current practice of TD design or design of any other mining facility requires risk analysis at every stage of the design and life of the facility. Over the years, numerous methodologies have been developed, categorized as Qualitative and Quantitative, and Semi-quantitative. Lacasse et al. (2019) extensively documented these methods, while the United States Bureau of Reclamation (USBR, 2015) introduced the concept of semi-quantitative risk analysis (SQRA). SQRA employs a risk matrix approach to assess potential failure modes (PFMs), assigning likelihood and consequences categories to each mode. Quantitative methods often face resistance from practitioners due to perceived complexities and requirements for extensive data and statistical knowledge. In contrast, SQRA has gained popularity as it allows for a more flexible assignment of probabilities using verbal equivalence. However, this approach relies heavily on expert judgment, disregarding the biases inherent in estimations and employing a deterministic evaluation of risk levels. In this study, we propose a methodology that addresses these limitations by mathematically modelling bias in score assignments, incorporating uncertainty through Monte Carlo simulation, and integrating it into the determination of risk levels.

FMEA background An FMEA is a logical step-by-step process that allows a systematic identification and analysis of PFMs and their associated consequences (Robertson et al., 2006), aiming to identify all of them, which involves the mechanics of failure, the consequences of these failure modes, and how the risks associated with the failures can be avoided or controlled (Valis et al., 2009). This can be done by screening every PFM according to the severity risk classification. Based on Hartford et al. (2004), the following structure can be defined: 1. Define the system, including the components to be analyzed. 2. Identify the critical elements of the system. 3. Analyze the PFM of the different elements. 4. Assess the likelihood of occurrence of the failure (L) and consequences level (C) of the various elements. 5. Set the (L, C) into a Risk Matrix to interpret the severity of the risk. 6. Summarize the findings. Special attention is required in Step 4 because this step involves the assessment of L and C of every identified PFM and is subject to the bias of judgment experts like the overconfidence of the stakeholders.

1318

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA The most common method to cope with this issue is to get an average rating. However, any measure of the uncertainty of the scores is assessed, making all the ratings assigned to each PFM comparable.

Likelihood of occurrence (L) and consequence level (C) Once a PFM is identified, every stakeholder should assign L and C values according to some scale, which the owner can define. Usually, the scales range from 1 to 5 (USACE/BOR, 2019); see Tables 1 and 2. Table 1: Failure likelihood descriptions (USACE/BOR, 2019)

Rate

Likelihood

1

Rare

2

Unlikely

Annual failure

Descriptor of evidence

likelihood < 1x10-6 1x10-5 to 1x10-6

Several events must occur concurrently or in series to cause failure. The possibility cannot be ruled out, but there is no compelling evidence to suggest it has occurred or that a condition or flaw exists that could lead to initiation. The fundamental condition of defect is known to exist; indirect evidence suggests

3

Possible

1x10-4 to 1x10-5

it is plausible; and key evidence is weighted more heavily toward less likely than more likely. The fundamental condition or defect is known to exist, indirect evidence suggests

4

Likely

1x10-3

to

1x10-4

it is plausible, and key evidence is weighted more heavily toward more likely than less likely.

5

Very likely

There is direct or substantial indirect evidence to suggest it has initiated or is

> 1x10-3

likely to occur soon.

Table 2: Consequences rating (USACE/BOR, 2019)

Rate

Category

Incremental life

Descriptor of evidence

loss

1

Low

1,000

Extreme economic losses affecting critical infrastructure or services

Loss of recreational facilities, seasonal workplaces, or infrequently used transportation routes. High economic losses affecting infrastructure, public transportation, or commercial facilities

1319

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Risk category Once the L and C of a PFM are determined, these scores must be combined to determine the risk, which can be obtained by summation or product among L and C. After that, the risk should be categorized by levels. According to Table 3, the risk was classified into four levels. It is common practice for risk to be managed using the following principle: As Low As Reasonably Practicable (ALARP). All risk reduction measures should be employed in ALARP if the cost of implementing them is reasonably practicable, considering cost-effectiveness (Aven, 2016). The high-risk category is defined with consideration of the ALARP principle. In the extreme and high-risk category, the risks are undesirable and must be reduced using ALARP. Table 3: Risk category (Schafer et al., 2021) Risk category

Description of risk category

Low

Risk minimal. Monitor risks.

Moderate

Risk tolerable with controls. Assess risk mitigation options and monitor these risks.

High

Risk undesirable. ALARP should employ risk mitigation to reduce the risk category.

Extreme

Risk intolerable. Risk mitigation is required immediately to reduce the risk category. Requires more detailed risk analysis.

Risk matrix The risk matrix is a colour code arrangement (see Figure 1), where the vertical axis usually corresponds to the likelihood rating, and the horizontal axis will be the level of consequence rating. This arrangement combines estimates of L with estimates of C to arrive at an expression of risk (Porter et al., 2019). Several methods have been proposed to define a good colour arrangement in the matrix, and a good compilation of the methods has been developed by Schafer et al. (2021). This study employed basic arithmetic based on ordinal numbers assigned to each L and C category (Duijm, 2015).

Monte Carlo simulation The Monte Carlo simulation is a model that makes decisions with repeated evaluation, with the input being a set of random numbers (Arend et al., 2019). Such a method often uses random scenarios for comparatively detailed evaluation, nonlinearity, or parameters with more than two uncertainties. A simulation model can contain more than 10 million simulation evaluations for representativeness. Much software is available in

1320

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA the industry to conduct the simulation. For the current study the program @Risk was employed.

Figure 1: Example of a risk matrix

Uncertainty model There are several approaches to modelling uncertainty for expert valuation in many respects. According to Haimes (2008), one way to model uncertainty about assessing scores from the expert is not to ask to assess probabilities. Instead, only three assessments of outcomes are solicited from the experts: lowest value, highest value, and most likely value. However, this kind of distribution considers a continuous random variable. It does not consider overconfidence because the mean value has the highest probability of occurrence, and the others are the lowest and highest value according to the expert judgment. A non-uniform discrete function was considered to represent the uncertainty better because it considers unequal intervals, assigning each interval a probability level. The intervals are designed to match the L and C levels, such as their probability, respectively.

Subjective judgment One critical component that needs to be tested would be the human experts. The risk matrix relies heavily on human expert input. However, although subjective judgment is unavoidable, we cannot just rely on it to perform subjective judgments. What is needed is objective assessments of subjective estimates. It is necessary to “calibrate” experts (Hubbard, 2020). Several features of subjective judgments were identified by Kahneman (2009) and are listed below: • Experience is a non-random, non-scientific sample of events throughout our lifetime. • Experience is memory-based, and we are very selective regarding what we remember. • What we conclude from our experience (or at least the part we choose to remember) can be full of logical errors. • Unless we get reliable feedback on past decisions, there is no reason to believe our experience will tell us much. • No matter how much experience we accumulate, we seem to be very inconsistent in its application.

1321

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The most pervasive, exhaustively researched conclusion reached by psychologists is that almost everyone is naturally overconfident in their prediction.

Overconfidence People are frequently overconfident in the accuracy of their estimates of uncertain quantities. Overconfidence can be measured using a straightforward method. Researchers like Lichtenstein et al. (1980) show how often someone is right about an estimate or forecast, and compare that to how often they expect to be correct. However, one or two forecasts are not enough. For instance, was someone overconfident if he says he is 90% confident in a prediction and is wrong on the first try? Not necessarily. That is why we must ask a large number of questions to be sure. After asking a subject many questions, researchers compute the expected number of correct answers. In decision analysis, the word expected usually means “probability-weighted average.” If someone makes fifty predictions and is 70% confident in each one, he expects to get thirty-five predictions right.

Calibration After extensive research, Lichtenstein and Fischhoff (1980) compiled a series of methods to calibrate overconfidence through general knowledge tests. There are two types of tests to calibrate overconfidence, which are described below.

The true/false calibration test This type of test is used to measure the expert’s confidence through a series of questions where the answer must be true or false, and assigning a confidence level to the answer, such that 50% confidence is when someone is just guessing and 100% is when the expert is sure of his answer. Figure 2 (top) shows a compilation of several outcomes of the test by Hubbard (2020), where it can be observed clearly that overconfidence is substantial precisely at a 90% assessed chance when people believe it is correct.

The confidence interval calibration test This is another type of calibration test because overconfidence also appears when experts are asked about range estimates. In this type of test, a series of questions should be answered, given the lower and upper boundary of a 90% confidence interval of their answers. Again, Hubbard (2020) shows a compilation of several outcomes (see Figure 2, bottom) where it was identified that the most common result if everyone were calibrated is to get a score of 9 out of 10; that is, ten of the answers fell within the stated range. Appendix A contains the two types of calibration tests used for this study, extracted from Hubbard (2020).

1322

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Figure 2: (Top) Comparison of uncalibrated and calibrated curves. (Bottom) Score distribution of actual ten question 90% confidence interval tests prior to training compared to Ideal calibration (Hubbard, 2020)

Stochastic modelling of expert judgment For the current study, calibration tests were conducted to obtain an actual calibration curve for overconfidence for expert valuation. A group of 15 people were evaluated using the tests shown in Appendix A. The test result is shown in Figure 3, where scattered data can be seen around the average of the uncalibrated curve proposed by Hubbard (2020), probably because of the small number of tested people. However, it can be used as an excellent reference to measure the overconfidence of the experts. Based on Figure 2, the average uncalibrated curve has uncertainty at which every level of “assessed chance” can be assigned an uncertainty and an average value, as shown in Table 4. Every expert should be assigned a confidence percentage to be calibrated according to the test result, using the Table 4 values. For instance, if someone is 90% for the estimation of L or C, the actual probability is around 65%, with surrounding boundaries ±18%. So, if a score for L or C is assigned, the mean value

1323

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA would be the score, and the lower and upper values for this score should be the values ±1. Then, the probability mass function for one expert should be as in Figure 4.

Figure 3: Calibration curve for participants Table 4: Boundaries for overconfidence level Assessed

Actual mean

Lower bound for

An upper bound for

chance

estimation

estimation

estimation

50%

50%

25%

25%

60%

52%

24%

24%

70%

56%

22%

22%

80%

60%

20%

20%

90%

65%

18%

18%

Figure 4: Schematic representation of non-uniform distribution for one expert

1324

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Case study The proposed methodology evaluated the FMs on a TD. The purpose was to demonstrate the utility and the additional information that this can add to the final analysis of failure modes by applying FMEA. As part of the design review of a tailings dam, a risk assessment was performed to address the potential FMs for the next design stage.

Site description The current TD is a sand dam with a starter dyke constructed of rockfill. The facility was constructed with a downstream construction method. The facility is in a high seismic hazard area, and a weak layer in the foundation was identified. Moreover, there is a pyrite cell embankment upstream of the TD, which could cause an overtopping of the main dam if the pyrite cells fail.

Failure description A workshop was held among 15 people (stakeholders), and a summary of the TD design was provided. The first task performed was to brainstorm a list of PFMs. Following the development of the initial list, the FMs that had been submitted by FMEA attendees via questionnaires prior to the workshop were reviewed and added to the list, if they had not already been identified by the group. The group then screened the list of PFMs to determine which PFMs to carry forward to the FMEA, and which ones to screen out. PFMs which were carried forward were each developed and further analyzed by the group. Development of each PFM consisted of listing the sequence of events in a step-by-step progression from the initial loading to the failure event. Relevant information for each PFM was discussed, captured, and input into the following categories: additional information, positive and adverse factors, potential surveillance and monitoring, data information needs, and potential risk reduction measures. Appendix B shows the description of the five PFMs considered for this study. The final step in the development of each PFM was to estimate the L and C of each PFM. This process with each PFM is further discussed in the following section.

Estimation of likelihood and consequence Previous to the workshop, a calibration test was performed on the participants, using the two types of tests (see Appendix A). For the current workshop, the participants estimated the L and C using an electronic polling method (Google forms), which asked them to state their confidence for both estimations (L and C) based on whether additional information was necessary. Additionally, based on the measure of the overconfidence of the group, all the estimates were adjusted according to the proposed values in Table 4.

1325

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Appendix D shows the probability mass function (PMF) of L and C after the overconfidence adjustment.

Simulations Firstly, the risk matrix was determined using a determinist approach (the common practice) to compare later with stochastic results. Figure 5 shows the 5 FMs after discussion in the workshop. The likelihood of occurrence and level of consequence were determined using central tendency statistic parameters like mode and median.

Figure 5: Risk matrix for case study Then, a PMF was defined for L and C obtained for every FM (see Appendix C), expressed as a nonuniform distribution. Once the inputs were defined, a Monte Carlo simulation was applied to calculate the R by 10,000 simulations. Table 5 shows the outcomes after simulations, comparing deterministic and probabilistic approaches. Table 5: Comparison of deterministic and probabilistic approaches Failure

Risk

Exceedance

severity

probability

2

5

14% (p > 5)

3

3

6

0.2% (p > 6)

FM-03

2

3

5

42% (p > 5)

FM-04

2

4

6

12% (p > 6)

FM-05

4

4

8

0% (p > 8)

Mean L

Mean C

FM-01

3

FM-02

mode

Discussion By employing Monte Carlo simulation, more comprehensive information regarding the probabilities of risk severity can be obtained. As shown in Table 5 and Figure 5, uncertainties arise regarding the need for risk reduction in FM-02, FM-04, and FM-05, as all failure modes should ideally be within the yellow zone. Additionally, FM-01 and FM-03 might potentially be elevated to the orange zone. Based on the outcomes, the following conclusions can be drawn:

1326

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA • FM-01 has a probability of approximately 14% to exceed the current risk level of 5. This finding is noteworthy as it suggests a possibility of moving to the orange zone, warranting additional cost considerations. • FM-02 has a probability of approximately 0% to exceed the current risk level of 6. This indicates that the risk level will likely remain constant, even with potential changes in expert judgment. • FM-03 has a probability of approximately 42% to surpass the current risk level of 5, indicating a high likelihood of this failure mode being classified as high-risk. A recommended risk level adjustment may be suggested, even with a conservative approach. • FM-04 has a probability of approximately 12% to exceed the current risk level of 6. However, due to its location in the risk matrix, the risk level would remain within the same orange zone. • FM-05 has a probability of approximately 0% to exceed the current risk level of 8. With this classification, it can be concluded that the risk level has reached its maximum value.

Conclusion Despite the criticisms surrounding the use of risk matrices, they remain prevalent in mining risk assessments. Therefore, the proposed methodology enhances the existing deterministic methodology used in the industry, providing stronger support and robustness. The present framework uses risk knowledge in the finance and nuclear industry. This paper shows how overconfidence and stochastic representation of the main inputs of FMEA analysis can give more information than just a risk level to inform a decision. The limitation of the methodology proposed lies in the tremendous amount of information required to get a more accurate curve for uncalibrated people. It is recommended to start the calibration process among professionals reporting to the stakeholders of mining companies, since more reliable information is needed to assess scores. Additional future studies were identified that could potentially reduce uncertainty and increase the group’s confidence in the risk estimate. Where confidence in the estimation is low, additional benefits may come from obtaining more information through additional studies or analyses. Much of the additional information will come in future design stages, during construction, and during the facility’s operation. Moreover, it is essential to reduce the biases in estimation probabilities, not only due to overconfidence but following the estimation of individual changes. The current approach can be adapted to conduct a more sensitive analysis by incorporating variations in the uncalibrated curve or probability distribution of expert judgments.

1327

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

References Arend, M.G. and Schäfer, T. 2019. Statistical power in two-level models: a tutorial based on Monte Carlo simulation. Psychological methods 24(1): 1. Aven, T. 2016. Risk assessment and risk management: review of recent advances on their foundation. European Journal of Operational Research 253(1): 1–13. Duijm, N.J. 2015. Recommendations on the use and design of risk matrices. Safety Science, pp. 76, 21–31. Haimes, Y.Y. 2005. Risk Modeling, Assessment, and Management. John Wiley & Sons. Hartford, D.N. and Baecher, G.B. 2004. Risk and Uncertainty in Dam Safety. Thomas Telford. Hubbard, D.W. 2020. The Failure of Risk Management: Why it’s Broken and How to Fix It. John Wiley & Sons. Kahneman, D. and Klein, G. 2009. Conditions for intuitive expertise: a failure to disagree. American Psychologist 64(6): 515. Lacasse, S., Nadim, F., Liu, Z., Eidsvig, U., Le, T.M.H., and Lin, C.G. 2019. Risk assessment and dams – recent developments and applications. In Proceedings of the XVII ECSMGE 2019: Geotechnical Engineering foundation of the future: European Conference on Soil Mechanics and Geotechnical Engineering. Lichtenstein, S., Slovic, P. and Fischhoff, B. 1980. Facts and Fears: Understanding Perceived Risk, pp. 181–216. Springer USA. Lichtenstein, S. and Fischhoff, B. 1980. Calibration training. Organizational Behavior and Human Performance 26(2): 149–171. Martins, E.F., Lima, G.B., Santana, A.P. and Fonseca, R.A.D. 2017. Stochastic Risk Analysis: Monte Carlo Simulation and FMEA. Revista Espacios 38(04). Porter, M., Lato, M., Quinn, P. and Whittall, J. 2019. Challenges with use of risk matrices for geohazard risk management for resource development projects. In MGR 2019: Proceedings of the First International Conference on Mining Geomechanical Risk pp. 71–84. Australian Centre for Geomechanics. Robertson, A. and Shaw, S. 2006. Mine Closure. Info Mine E-Book: Vancouver, BC, Canada. Schafer, H.L., Beier, N.A. and Macciotta, R. 2021. A failure modes and effects analysis framework for assessing geotechnical risks of tailings dam closure. Minerals 11(11): 1234. Sun, J.J., Yeh, T.M. and Pai, F.Y. 2022. Application of Monte Carlo simulation to study the probability of confidence level under the PFMEA’s action priority. Mathematics 10(15): 2596. Tversky, A., Brenner, L.A., Koehler, D.J. and Liberman, V. 1996. Overconfidence in probability and frequency judgments: a critical examination. Organizational Behavior and Human Decision Processes 65(3): 212–219. United States Army Corps of Engineers/Bureau of Reclamation (USBR). 2019. Best Practices in Dam and Levee Safety Risk Analysis, Version 4.0, July 2019. Valis, D. and Koucky, M. 2009. We selected overview of risk assessment techniques. Problemy eksploatacji (4): 19–32.

1328

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Appendix A Supplementary Calibration Test: Range #

Question

1

What percentage of bronze is typically made of copper?

2

How many countries have at least one McDonald's?

3

How many employees did eBay have in the first quarter of 2006

4

What was the population of Miami (within the city limits, not the entire metropolitan area) in 1990?

5

How many casualties did the French suffer in the Battle of Waterloo?

6

What is the range in miles of a Minuteman Missile?

7

What is the percentage of I.T. jobs in the U.S. were unfilled in 1997?

8 9 10

The Supremes' (with Diana Ross) song “Stop! In the Name of Love” was how long? (minutes, seconds) How many undergraduates attended Cambridge in 1990? If you could jump 50 feet straight into the air, how many seconds would you be airborne before landing?

11

How many gallons are in a bushel (they are both measures of volume)?

12

How many sovereign rulers has England had in the last thousand years?

13

14

15

If the air temperature was 5 degrees below zero (Fahrenheit) and the wind speed was 15 mph, what would the temperature adjusted for wind-chill be? The average cost of testing in software development is what percentage of total project costs? On average, if a software development project was projected to take 17 months, it takes how many months?

16

How many meters tall is the Sears Tower?

17

How many gold medals did Jesse Owens win at the 1936 Berlin Olympics?

18

19 20

In 2005, the average combined MPG for all U.S. cars and light trucks on the road was how much? The average house in the United States uses how many gallons of water per day? What was the average price in the United States of a house sold in 2001? .

1329

Lower Bound

Upper Bound

(95% chance value

(95% chance value

is higher)

is lower)

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Supplementary Calibration Test: Binary #

Statement

Answer

Confidence that you are correct

(T/F)

(Circle one)

1

The melting point of tin is higher than the melting point of aluminum.

50% 60% 70% 80% 90% 100%

2

In English, the word “quality” is more frequently used that the word “speed”.

50% 60% 70% 80% 90% 100%

3

Any male pig is referred to as a hog.

50% 60% 70% 80% 90% 100%

4

California's giant sequoia trees are named for an early 19th century leader of the Cherokee Indians.

50% 60% 70% 80% 90% 100%

5

The Model T was the first car produced by Henry Ford.

50% 60% 70% 80% 90% 100%

6

When rolling 2 dice, a roll of 7 is more likely than a 3.

50% 60% 70% 80% 90% 100%

7

No one has ever been reported to have been hit by any object that fell from space.

50% 60% 70% 80% 90% 100%

8

Sir Christopher Wren was a British anthropologist.

50% 60% 70% 80% 90% 100%

9

Pakistan does not border Russia.

50% 60% 70% 80% 90% 100%

10

The Navy won the first Army-Navy football game.

50% 60% 70% 80% 90% 100%

11

The paperback version of the book “The Da Vinci Code”, as of July 2007, still ranks in the top 500 bestselling books on Amazon.

50% 60% 70% 80% 90% 100%

12

Italian has more words than any other language.

50% 60% 70% 80% 90% 100%

13

The month of August is named after a Greek god.

50% 60% 70% 80% 90% 100%

14

The deepest ocean trench is deeper than the Grand Canyon.

50% 60% 70% 80% 90% 100%

15

Abraham Lincoln was the first president born in a log cabin.

50% 60% 70% 80% 90% 100%

16

As of July of 2007, more people search Google for “Harry Potter” than “Hillary Clinton” (according to Google Trends).

50% 60% 70% 80% 90% 100%

17

The population of Alabama is higher than the population of Arizona.

50% 60% 70% 80% 90% 100%

18

No category 5 hurricane hit the U.S. in 2004.

50% 60% 70% 80% 90% 100%

19

The U.K. is among the top 10 largest economies in the world (by GDP).

50% 60% 70% 80% 90% 100%

20

The movie Forest Gump has grossed more to date than E.T. The Extra Terrestrial.

50% 60% 70% 80% 90% 100%

1330

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Appendix B Failure mode 1 Slope instability through the foundation at the main dam due to a weak foundation layer. The potential consequence of this failure mode was rated significant with the following justification: • The runout with this type of failure is not expected to extend very far. The nearest Population at Risk (PAR) is approximately 4 miles from the impoundment, with the potential exception of operators on site. • The tailings from this failure are expected to stay in the Dripping Springs Wash. • The primary concern is environmental damage. A failure will impact habitat and vegetation damage would be expected.

Failure mode 2 Internal erosion through the foundation at the starter dam. The potential consequence of this failure mode was rated significant to high with the following justification: • Because the starter dam will impound a larger pond, closer to the exterior slope during initial operation, the consequences from this failure were seen as potentially greater than for the previously postulated failures that occur later in operation. • The tailings from this failure are expected to stay in the Dripping Springs Wash. • The primary concern is environmental damage, and with the further runout the environmental damage could be high. A failure will impact habitat and vegetation damage would be expected.

Failure mode 3 Slope instability through the tailings dam during a seismic event. The potential consequence of this failure mode was rated significant to very high with the following justification: • The runout for failure is expected to be similar for the normal case, but some of the group felt it may extend further due to the liquefaction of the tailings and could impact the Gila River. • Both a potential for loss of life and environmental damage contributed to the consequence estimate. A failure will impact habitat and vegetation damage would be expected. • Potentially higher consequences could occur if the PAG cell embankment were to fail concurrently with the Main dam.

1331

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Failure mode 4 Slope instability at the main dam due to high porewater pressure in the dam following a hydrologic event. Due to the presence of the pond the potential consequence of this failure mode was rated high to very high with the following justification: • Because the starter dam will impound a larger pond, closer to the exterior slope during initial operation, the consequences from this failure were seen as potentially greater than for the previously postulated failures that occur later in operation. There will also be additional water in the impoundment from the flood. • The tailings from this failure are expected to stay in the Dripping Springs Wash. • The primary concern is environmental damage, and with the further runout the environmental damage could be high to very high. • A failure will impact habitat and vegetation damage would be expected.

Failure mode 5 Slope instability through the foundation at the PAG dam due to a weak layer. The potential consequence of this failure mode was rated very high with the following justification: • There is more water in the PAG Embankment that will be released and will result in further runout, thus failure releases are expected to reach the Gila River. • Pyrite tailings will have higher environmental consequences. The estimate is based primarily on the environmental and economic damages. • The potential for direct loss of life is considered very low to remote, but there may be some indirect loss of life.

1332

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Appendix C Assessment of the experts FM-01

FM-02

FM-03

FM-04

FM-05

Expert O

C

O

C

O

C

O

C

O

C

1

3

1

3

1

3

3

2

3

3

4

2

3

1

3

1

3

3

2

4

3

4

3

3

2

3

1

2

2

2

3

3

4

4

3

2

3

2

2

2

2

2

3

4

5

2

2

3

2

2

2

2

3

4

4

6

2

2

3

3

2

2

1

4

4

4

7

2

2

3

3

2

2

3

4

4

4

8

1

2

2

3

2

3

3

4

4

4

9

4

3

2

3

2

3

3

4

4

4

10

3

3

2

3

2

3

3

3

4

3

11

3

2

2

3

4

3

2

4

4

3

12

4

2

2

3

4

3

2

3

4

3

13

4

1

2

4

1

3

2

4

4

3

14

1

2

1

4

1

3

2

4

2

3

15

2

2

4

4

1

3

2

4

2

3

Mean

2.7

1.9

2.5

2.7

2.2

2.7

2.2

3.5

3.5

3.6

Median

3

2

3

3

2

3

2

4

4

4

Mode

3

2

3

3

2

3

2

4

4

4

1333

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Appendix D Non-uniform distribution of experts for every failure mode

Figure D.1: Non-uniform distribution of likelihood and consequence for FM-01

Figure D.2: Non-uniform distribution of likelihood and consequence for FM-02

Figure D.3: Non-uniform distribution of likelihood and consequence for FM-03

1334

APPLICATION OF MONTE CARLO SIMULATION TO THE PROBABILITY ASSESSMENT IN FMEA

Figure D.4: Non-uniform distribution of likelihood and consequence for FM-04

Figure D.5: Non-uniform distribution of likelihood and consequence for FM-05

1335

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Appendix E Simulation outcomes

Figure E.1: Exceedance curve for all the analyzed failure modes

1336

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Operational Control and Trigger Action Response Plan for a Tailings Storage Facility Alejandro Calvo, SRK Consulting, Colombia Manuel Cervantes, SRK Consulting, Mexico Laura Moreno, SRK Consulting, Colombia Diego Cobos, SRK Consulting, Colombia

Abstract Tailings storage facility failures often result in the release of large volumes of tailings and water, directly impacting the environment and the economy, and pose a high risk to the lives of workers and communities in the areas downstream of the facilities. The United Nations reported about forty such incidents in the past decade, causing alarm in communities near Tailings Storage Facilities (TSFs), due to the potential scale of these events. The Trigger Action Response Plan (TARP) methodology aims to identify key monitoring parameters relating to dam safety. The pre-defined trigger levels are provided for performance criteria and describe actions to be taken if these values are exceeded. Risk levels classify events where these values trigger a response to mitigate or eliminate risk. These parameters are usually monitoring of deformations, settlements, pore pressures, and phreatic levels. Specialized instrumentation, such as piezometers, inclinometers, shape arrays, and survey monuments, are used to obtain those parameters. Based on the instrumentation available at each facility, a series of crucial cumulative information is collected, allowing improved knowledge of the monitored TSF structure. This also enhances risk-based decision-making by increasing the understanding of uncertainties and the potential for expected deviations during operations. This document presents the recommended operating parameters based on historical data, current instrumentation readings, and model analysis of a TSF built in the 1980s, recently reinforced with a buttress and instrumented with 13 piezometers in 12 locations, and 13 survey monuments with one control point. Finally, conclusions and lessons learned are provided, aiming to give a framework of where to take the future of instrumentation and TARP, presenting potential options such as automatization of instrument readings, generation of graphs and models in real-time, and telemetry, among others.

1337

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction The failure of a TSF can have adverse impacts on the surrounding environment and population. To prevent failure, it is important that TSFs are properly designed, constructed, operated, and monitored. Triggers, actions, and response plans (TARPs) are a critical component of tailings management. TARPs are designed to identify and respond to potential tailings dam failures. They typically include a set of triggers, which are specific conditions that, if exceeded, indicate, for example, that a failure may be imminent. When a trigger is exceeded, a set of actions are taken, which are designed to mitigate the risk of failure. A robust monitoring system and associated comprehensive TARPs are a key part of any successful tailings embankment operation. Tailings management systems range from simple to complex, with designs tailored to each TSF’s unique characteristics and expected performance (Morrison, 2022). The TARPs specify the necessary actions for set trigger levels, which may relate to hydraulic heads, fluxes, deformations, solute concentrations, etc. When those levels are reached, a predefined response (action) should be taken, such as increased monitoring or, at more critical levels, the implementation of mitigating measures. These levels should be frequently compared to pre-defined values, taking into account both regulatory requirements and criteria based on the expected performance and design intent of the tailings storage facility. The TARP initiates with the identification of an unusual event and is followed by an evaluation of the emergency level and an appropriate response. TARPs include actions for various triggers that may employ traffic light evaluation methods, together with the essential inspection and monitoring items for TSFs. Site-based workers (such as TSF operators) should have access to clear and succinct decision-making tools through TARPs. In all mine locations that operate TSFs, TARPs are advised based on governance requirements and international standards such as GISTM, ICOLD, ANCOLD, and CDA. This document presents a brief introduction to TARP methodology, and the recommended operating parameters for a tailings storage facility (TSF) that was built in the 1980s and recently reinforced with a buttress. The TSF is instrumented with 13 piezometers in 12 locations and 13 survey monuments with one control point. The recommended operating parameters are based on historical data, current instrumentation readings, and model analysis.

TARP methodology Triggers are one of the key components of TARPs. They are specific conditions that, if exceeded, constitute an early warning, or even warn that a tailings dam failure is imminent. Triggers can be based on a variety

1338

OPERATIONAL CONTROL AND TRIGGER ACTION RESPONSE PLAN (TARP) FOR A TAILINGS STORAGE FACILITY (TSF) of factors, including: Geotechnical conditions: These include the stability of the tailings dam, the presence of faults or other geological features, and the seismicity of the region. Hydrological conditions: These include the water content of the tailings, the rainfall and runoff in the area, and the presence of groundwater. Instrumentation data: This includes data from piezometers, inclinometers, and other monitoring devices. The selection of triggers is a critical step in the development of TARPs. Triggers should be based on a thorough understanding of the tailings dam and the potential failure mechanisms. They should also be sensitive to changes in the conditions that could lead to a failure. When a trigger is exceeded, a set of actions are taken. These actions are designed to mitigate the risk of failure and protect people and the environment. The actions that are taken will vary depending on the specific situation. However, they typically include: Monitoring: This involves increasing the frequency of monitoring and collecting additional data. This can help to identify the cause of the trigger and to assess the risk of failure. Intervention: This may involve dewatering the tailings, stabilizing the tailings dam, or evacuating people from the area. Intervention can help to reduce the risk of failure and to protect people and the environment. Emergency response: This includes planning for and responding to a tailings dam failure. Emergency response plans should detail the specific actions that will be taken in the event of a failure. Response plans should be developed for each tailings dam. These plans should detail the specific actions that will be taken in response to each trigger. Response plans should be regularly reviewed and updated to ensure that they are up-to-date and effective. Monitoring systems are an important part of emergency action plans and the development of TARPs. Triggers can be established based on a threshold value or action level. Threshold value: A specific numerical value that initiates a systematic response when exceeded. For example, a threshold elevation may be established for an individual piezometer. If the piezometer elevation exceeds the threshold value, further investigation and analyses are warranted. Action level: A prescribed change that occurs over a discrete period. For example, a piezometer may read a specific elevation that is well below the established threshold elevation. However, if the piezometer elevation increases over several days and exceeds the action level, pore pressures within the tailings embankment are rapidly increasing and warrant further investigation.

1339

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA TARPs are a key part of the monitoring plan and need to be carefully formulated to be effective. The first part – setting the trigger – should be validated and reviewed often, as this is the primary component of a TARP. If the trigger is incorrect or inappropriate, the action and response will not occur. The level of the trigger should be related to the responses required. For example, a yellow trigger can indicate conditions outside normal operations and warrant further investigation, but may not result in a large-scale evacuation. However, a red trigger indicates conditions close to failure, and evacuation should occur. The designer or EoR (engineer of record) should understand the failure mode being considered, identify what is being measured, and set a threshold that reflects a factor of safety (when considering stability) or value that would be cause for concern. For example, a cumulative movement of 200 mm over 6 months may be set for an inclinometer. A phreatic surface elevation may be set for slope stability, with various elevations indicating relevant factors of safety values (typically 1.5 and 1.3). The triggers should be based on design analyses and supported by field data, where possible. Sufficient sensitivity analyses should be undertaken in the analyses to support identification of the conditions under which failure could be indicated, to support identification and selection of appropriate triggers.

Case study The TSF selected as a case study to show the parameters selected based on a piezometric historical data, is an earthen tailings dam in Mexico. It was approximately 40 m high and built downstream. The outer slope of the dam was over steepened at about 1.4H: 1V. The toe of the dam had been flooded by tailings deposited into the dam directly downstream. There were no design reports or construction records available for the dam; however, based on information from site staff and a review of available historical images, the dam was built after the year 2000. In 2021, the owner carried out a project to upgrade the dam that included the follow activities: 1. Excavation of tailings at the toe of the dam. 2. Installation of a drain and filter layer. 3. Decant penstock decommissioning. 4. Buttress construction. 5. Spillway upgrade. 6. Monitoring instrumentation installation. With the new dam configuration, deterministic stability analyses were performed using the numerical software Slide 2018 (Rocscience, 2018) to evaluate the Factor of Safety (FoS) and its associated slip surface. The FoS was calculated using Spencer’s method (Spencer, 1967). The water level upstream of the

1340

OPERATIONAL CONTROL AND TRIGGER ACTION RESPONSE PLAN (TARP) FOR A TAILINGS STORAGE FACILITY (TSF) dam was associated on 1,915.9 m, which corresponds to the maximum pond operating level. Both wet toe (dewatering dyke reservoir) and dry toe conditions have been tested during this assessment to evaluate variable elevation of the downstream pond. Only failure surfaces that involve at least half of the dam crest were considered.

Figure 1: Locations used to estimate the phreatic levels (wet toe condition)

Establishing triggers For dam operation, after buttress construction, operating values (triggers) of the instrumentation were defined to assure the integrity and structural safety of the dam. These values were defined using three main states for each instrument: the historical values, the design values, and the theoretical failure values. To evaluate the impact of varying phreatic levels in the reduction of the FoS in the dam against slope instability, the phreatic levels were modified manually in a 2D limit equilibrium stability model used during buttress design. The phreatic level was raised to investigate the point at which the static FoS could be reduced to 1.3, 1.1 and 1.0, with 1.0 being the theoretical onset of failure and 1.3 and 1.1 being arbitrary values between the design criteria (static conditions) and the theoretical failure.

Historical and design values The historical data of the phreatic level inside the dam was collected since 2015, using standpipe piezometers. Since 2020, vibrating wire piezometers were installed in new locations and the standpipes were automated, also using vibrating wire piezometers (VWP). Figure 2, Figure 3, and Figure 4 show the plots of the data of the piezometers installed in the dam crest, the intermediate berm, and at the toe, respectively.

1341

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 2: Dam crest VWP water pressure elevation

Figure 3: Dam berm VWP water pressure elevation

1342

OPERATIONAL CONTROL AND TRIGGER ACTION RESPONSE PLAN (TARP) FOR A TAILINGS STORAGE FACILITY (TSF)

Figure 4: Dam toe VWP water pressure elevation

Based on the phreatic surface used in the stability analyses, design phreatic levels were estimated at three locations in the embankment cross-section where piezometers are installed. The phreatic levels used in the stability analyses when designing the stabilization buttress for the three selected locations are presented in Table 1. The design levels are plotted with the actual measured phreatic levels until January 2022. It can be observed that the measured piezometric levels are lower than the design values, except for the piezometers installed at the dam toe, which are affected by the water conditions at the downstream pond (i.e., due to the dewatering dike reservoir level variations).

TARP levels The acceptable condition has been defined as levels below historical levels measured in nearby piezometers or during and immediately after construction of the dam upgrades. The early warning is reached when the levels surpass the historical levels. Subsequently, the corrective action level is triggered once the piezometric levels are such that the Factor of Safety in static conditions is below 1.3. Finally, the critical condition is reached when the piezometric levels are such that the Factor of Safety in static conditions is below 1.0. .

1343

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 1: Recommended VWP TARP Levels, high toe water level TARP Trigger Level Green

Yellow

Orange

Red

Acceptable condition

Early warning

Corrective action

Critical condition

PZM 01

x < 1,901.0 m

1,901.0 m < x < 1,911.0 m

PZM 02

x < 1,909.0 m

1,909.0 m < x < 1,911.0 m

PZM 03

x < 1,908.0 m

1,908.0 m < x < 1,911.0 m

SRK20-D4i-P01

x < 1,904.0 m

1,904.0 m < x < 1,911.0 m

1,911.0 m < x < 1,914.2 m

x > 1,914.2 m

SRK20-D4iP01A

x < 1,893.0 m

1,893.0 m < x < 1,911.0 m

SRK20-D4i-P04

x < 1,886.0 m

1,886.0 m < x < 1,911.0 m

SRK20-D4i-P07

x < 1,883.0 m

1,883.0 m < x < 1,911.0 m

SRK20-D4i-P02

x < 1,889.0 m

1,889.0 m < x < 1,897.5 m

SRK20-D4i-P03

x < 1,889.0 m

1,889.0 m < x < 1,897.5 m

1,897.5 m < x < 1,902.2 m

x > 1,902.2 m

SRK20-D4i-P08

x < 1,883.0 m

1,883.0 m < x < 1,897.5 m

SRK20-D4i-P05

x < 1,881.0 m

1,881.0 m < x < 1,883.0 m x > 1,883.0 m

x > 1,883 m and any of the above piezometers in Red level

Instrument

Crest piezometers

Berm piezometers

Toe piezometers

SRK20-D4i-P06

x < 1,882.0 m

1,882.0 m < x < 1,883.0 m

SRK20-D4i-P09

x < 1,882.7 m

1,882.7 m < x < 1,883.0 m

1344

OPERATIONAL CONTROL AND TRIGGER ACTION RESPONSE PLAN (TARP) FOR A TAILINGS STORAGE FACILITY (TSF) Table 2: Recommended VWP TARP Levels, low toe water level TARP Trigger Level Green

Yellow

Orange

Red

Acceptable condition

Early warning

Corrective action

Critical condition

PZM 01

x < 1,901.0 m

1,901.0 m < x < 1,912.0 m

PZM 02

x < 1,909.0 m

1,909.0 m < x < 1,912.0 m

PZM 03

x < 1,908.0 m

1,908.0 m < x < 1,912.0 m

SRK20-D4i-P01

x < 1,904.0 m

1,904.0 m < x < 1,912.0 m

1,912.0 m < x < 1,913.9 m

x > 1,913.9 m

SRK20-D4iP01A

x < 1,893.0 m

1,893.0 m < x < 1,912.0 m

SRK20-D4i-P04

x < 1,886.0 m

1,886.0 m < x < 1,912.0 m

SRK20-D4i-P07

x < 1,883.0 m

1,883.0 m < x < 1,912.0 m

SRK20-D4i-P02

x < 1,889.0 m

1,889.0 m < x < 1,898.3 m

SRK20-D4i-P03

x < 1,889.0 m

1,889.0 m < x < 1,898.3 m

1,898.3 m < x < 1,902.5 m

x > 1,902.5 m

SRK20-D4i-P08

x < 1,883.0 m

1,883.0 m < x < 1,898.3 m

SRK20-D4i-P05

x < 1,874.7 m

1,874.7 m < x < 1,877.0 m

SRK20-D4i-P06

x < 1,875.0 m

1,875.0 m < x < 1,877.0 m

1,877.0 m < x < 1,880.2 ,m

x > 1,880.2 m

SRK20-D4i-P09

x < 1,875.2 m

1,875.2 m < x < 1,877.0 m

Instrument

Crest piezometers

Berm piezometers

Toe piezometers

Conclusion Effective tailings management systems are crucial for safe mining operations. The utilization of Triggers, Actions, and Response Plans (TARPs) plays a vital role in ensuring the proper function and operation of TSFs. TARPs are specifically designed to identify and respond to potential failures of tailings dams by defining triggers, which are specific conditions that, if surpassed, could indicate an early warning or an imminent failure. When a trigger is exceeded, a series of actions are implemented to mitigate the risk of failure. The TARP methodology determines the necessary actions based on trigger levels, which can involve factors such as hydraulic heads, fluxes, deformations, and solute concentrations. When these levels are

1345

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA reached, predefined responses or actions should be taken, such as enhanced monitoring or, in more critical situations, the implementation of measures to mitigate risks. In the case study, TARPs are of particular significance in preventing potential failures, considering the recent upgrade of the dam with a buttress, which has led to increased height and stresses at the foundation. The TARP was important for its change in the slopes, which reduced the risks of failure. TARPs were developed based on the Factor of Safety (FoS), calculated using 2D limit equilibrium models that incorporate different phreatic levels and correspond to the locations of various piezometers. On the downstream toe of the case study, the water level in the dewatering dike reservoir can rise to the level of the buttress lower bench. The higher-than-design phreatic surface at the toe is not currently problematic; however, the water level at the toe becomes an important variable for control. As it lowers, the piezometric levels at the toe need to lower in conjunction to keep the slope stable. This paper presented what TARPs are trying to control and why these factors were chosen. However, future efforts may show what would happen if others were chosen, what external conditions influence the stability of the dam, and what uncertainties are guiding the definition of critical controls. In summary, TARPs serve as a crucial tool in preventing failures in the studied case, especially considering the dam’s recent upgrades and associated risks. The installation of multiple piezometers and survey monuments allows for comprehensive monitoring, and the establishment of TARPs based on FoS calculations and phreatic levels further enhances the risk mitigation efforts.

Acknowledgements The authors acknowledge Ignacio Garcia, SRK Practice Leader, head of the project presented in this paper, for his continuous support.

References CDA – Canadian Dam Association. 2019. DRAFT Revision – Application of Dam Safety Guidelines to Mining Dams – Section 3.5.4.1 Targets for Factor of Safety. MAC – The Mining Association of Canada. 2019. Developing an Operation, Maintenance and Surveillance Manual for Tailings and Water Management Facilities, Second Edition Morrison, Kimberly. 2022. Tailings Management Handbook: A Life-Cycle Approach. Rocscience. 2018. Slide software. Spencer, E. 1967. A method of analysis of the stability of embankments assuming parallel inter-slice forces. Geotechnique 17(1): 11–26. https://doi.org/10.1680/geot.1967.17.1.11

1346

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Extending a Mining Company’s Risk Assessment Framework to Tailings Facilities Andy Small, Klohn Crippen Berger, Canada Johanna Barbaran, Klohn Crippen Berger, Canada Ogechi Mary Ileme, Klohn Crippen Berger, Canada

Abstract A typical risk assessment framework that is used by mining companies is a 5 × 5 matrix that includes the two main components of risk: likelihood and consequences, plotted on two axes. The matrix is further subdivided into risk levels that define whether a risk is: (i) acceptable to the mining company; (ii) requires further reduction; or (iii) is not acceptable and cannot be allowed to continue. This approach has been effective for mine operators to identify, characterize, and manage their risks. However, the typical risk matrix developed by a mining company is not suitable for tailings storage facility (TSF) safety, which typically has risks with much lower likelihood events (e.g., 10,000-year return periods) and higher consequences, than other risks at a mine site. A number of alternative frameworks to the typical risk matrix have been developed that are better suited to TSF risks. However, using an alternative framework can make it difficult for the TSF team to communicate and prioritize risks within the mining company. Some mining companies have already modified their corporate risk assessment framework to address or incorporate TSF risks. This paper discusses an approach utilized by the authors and colleagues at Klohn Crippen Berger for mining companies that have not already made this modification to consider. It extends the typical 5 × 5 risk matrix to address the unique aspects of tailings facilities; enabling integration within the corporate risk management system. This framework has assisted clients with: (i) meeting the requirements of the Global Industry Standard on Tailings Management (GISTM); (ii) categorizing and communicating TSF risks within project, site and corporate teams; (iii) assisting the TSF safety teams to gain a thorough understanding of the facility risks; (iv) identifying the critical controls and surveillance activities; (v) identifying and prioritizing additional risk mitigation opportunities, and; (v) supporting the development of trigger, action, and response plans (TARPs) and emergency response plans (ERP).

1347

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Recent tailings dam failures around the world are encouraging mining companies to improve their tailings storage facility (TSF) safety assessment approaches as part of their commitment to protecting the public and environment by identifying risks in a timely manner. Typically, the risk assessment involves characterizing the existing risks and controls to determine if they are tolerable or if additional controls need to be implemented to further reduce the level of risk. This paper provides an approach that can be used by mining companies to extend their corporate risk assessment to address tailings storage facility risks.

Risk assessment A risk assessment involves three key elements: risk identification, risk analysis, and risk evaluation (ICMM, 2021). Risk identification involves identifying potential failure modes and the events that could lead to a failure. The credibility of the potential failure mode is considered at this stage. Risk analysis develops an estimate of the likelihood of occurrence of the failure modes and severity of the consequences of the potential resulting events. Risk evaluation considers the outcome of risk assessment and other factors (such as cost and timing) to determine if the risk is tolerable and support decisions on risk reduction measures.

Typical mining company risk assessment Mining companies have multifaceted risks due to complex human engineering systems (Tubis et.al., 2020), infrastructure, and operations. The outcome of these risks, if not adequately controlled, could be catastrophic and could lead to loss of life, adverse effects on human health and well-being, detrimental impacts on the environment, and financial and negative reputational impacts. This demands that mining companies adopt risk-based management strategies to help to reduce risks to acceptable levels. Figure 1 presents a typical mining company risk framework based on the author’s experience working with mining companies on their risk assessments. The likelihood descriptors shown (Figure 1) are representative of typical criteria used by mining companies in the author’s experience. Descriptors for each consequence level and category, can be quite variable between mining companies and therefore no examples are shown. The risk matrix is further categorized into three overall “Risk Ratings” that are considered during the risk assessment stage. Some companies use up to five rating categories.

1348

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES

Figure 1: Typical mining company risk framework Common risk assessment techniques include, but are not limited to, the following: failure mode and effect analysis, bow-tie analysis, fault tree analysis, fish bone analysis, pareto analysis, consequence/likelihood matrix. While most of these methods are based on general risk assessment concepts proposed by the International Standard Organization (ISO) and others (i.e., identification of hazards, the cause (trigger), the impact of the hazard (consequence), probability of occurrence (likelihood) etc.), the framework for communicating the risks and supporting risk reduction decisions can vary. The typical corporate framework shown in Figure 1 has been effective for several aspects of mining projects such as mine planning, operations, health and safety, etc. However, the framework is not suitable for tailings facilities that have risks with lower likelihood events and higher consequences than most other risks on a mine site.

Terminology Agreement on terminology is important to provide clarity in the risk assessment process. A list of terms used by the authors is provided at the end of the text.

Objectives of tailings storage facility risk assessment The main objectives of a TSF risk assessment are as follows: 1. Gain a thorough understanding of the facility: Risk assessments can guide the TSF safety team to analyse and thoroughly understand key components of tailings facilities, the system within which they function, the process of operation and the risks associated with the facility. 2. Categorize and communicate tailings risks within the project, site and corporate teams: Risk assessment provides a means of categorising risks and informing decision on whether or not risk reduction is needed.

1349

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA 3. Identify critical controls: In general, the design controls are considered critical controls and, where applicable, should be clearly identified for each particular failure mode. Critical mitigative controls should also be noted (such as evacuation). 4. Identify critical surveillance activities: The risk assessment can identify potential failure modes and provide an understanding of performance indicators associated with the potential failure modes. By recognizing the potential failure modes, surveillance activities can be identified, and an evaluation undertaken to determine the critical surveillance activities. This can be achieved by asking some simple questions: What would happen if the surveillance activity was suspended for an extended period of time? Could a failure mode develop without detection and possible intervention? If so, then this surveillance activity could be deemed as critical. 5. Identify further risk reduction opportunities: Risk assessment can identify additional risk reduction actions to reduce risks to tolerable levels and can support prioritization of the projects or the implementation of actions based on the risk level, the mining company’s requirements, and other considerations. 6. Support the development of trigger, action, and response plans (TARPs) and emergency response plans (ERPs): Credible failure modes identified in the risk assessment are considered for TARPs and ERPs are developed. 7. Inform the assessment of as low as reasonably practicable (ALARP): For each failure mode, an assessment can be undertaken to determine if the risk level associated with a particular failure mode can meet the conditions of ALARP. Note that these conditions are typically established by the mine owner. 8. Compliance with GISTM: The recently published GISTM sets out standards for management of tailings facilities to achieve zero harm to people and the environment. The standard requires TSF owners use risk assessments to inform design, to conduct risk assessment at a minimum of three years or when there is a material change to a facility or to social, environmental, and economic context.

Modifications to typical mining company risk assessments Key modifications / clarifications to the typical mining company’s risk assessment framework are presented in this section.

1350

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES

Likelihood categories “Bolt-On” Tailings Storage Facilities are designed and constructed under various guidelines and regulations, often with stringent design criteria (e.g., probable maximum flood or maximum credible earthquake). These have generally lower likelihoods categories than those typically presented in a mining company’s typical risk framework (Figure 1). Therefore, two additional likelihood categories are included: Extremely Rare and Negligible making a 5 × 7 risk matrix. The Extremely Rare likelihood describes a risk in which a “fundamental condition or defect is known to exist” (FERC, 2021b, Level FL3), but a combination of very low probability circumstances would be required for both an initiating event and a failure mechanism to develop concurrently for the failure mode to occur. This is associated with equivalent return periods of 1 in 10,000 to 1 in 100,000 years. The Negligible likelihood describes a risk in which “the possibility cannot be ruled out, but there is no compelling evidence to suggest it has occurred or that a condition or flaw exists that could cause initiation” (FERC, 2021b, Level FL2). In this case, several events must occur concurrently or in series to cause a failure and most, if not all, of the events have negligible likelihoods (FERC, 2021b, Level FL1). This is associated with equivalent return periods greater than 1 in 100,000 years. The authors have combined Federal Energy Regulatory (FERC) Levels FL1 and FL2 into a single likelihood category, called Negligible. When considering events of this low likelihood, it is a challenge to accurately define the likelihood. Even though these likelihoods seem very small, failure modes in these categories need to be considered for management purposes, risk informed decision making, adequate closure planning and passive care closure. Figure 2 presents the mining company risk framework presented in Figure 1 with the two additional likelihoods discussed above. An additional column entitled “Additional Descriptors” was added to provide a better description of the conditions to support the determination of the likelihood. The risk levels range from High (red), Medium (orange), to Low (green). The bottom right cell of the risk matrix could be colourcoded orange or green, depending on the mining company’s risk tolerance. When presenting the risk assessment results in reports, the authors have included the “corporate heat map” and the “dam safety heat map”. The “corporate heat map” is like the one shown in Figure 1, and often the failure modes form a cluster in the lower right hand corner of the heat map. The “dam safety heat map” allows the Owner to better understand the basis for these risks in the lower right hand corner.

1351

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 2: Extended mining company risk framework

Use of annual exceedance probability When establishing the likelihood of a potential failure mode, it is not always possible to arrive at an annual exceedance probability such as for static liquefaction of an upstream constructed TSF. Judgement is required in concert with the additional risk descriptors shown in Figure 2 to arrive at the likelihood. For failure modes such as overtopping due to flooding and instability due to a seismic event, it is possible to estimate the probability of the initiating event (flood or earthquake) and this can be used to derive numerical estimates of the annual probability of failure by considering the chain of events (event tree) that can lead to failure. This section discusses the author’s approach to extending the additional risk descriptors provided in Figure 2 to equivalent annual exceedance probabilities (AEP). As indicated in Figure 2, the words likely, could, unlikely, and very unlikely are used in the additional risk descriptors to indicate the possible occurrence of the failure mode over the time frame that has been indicated. These have been defined as follows: • Very Unlikely to occur: Probability of event occurring over specified time frame is less than 5% • Unlikely to occur: Probability of event occurring over specified time frame is between 5 and 25% • Could occur: Probability of event occurring over specified time is between 25% and 60% • Likely to occur: Probability of event occurring over specified time frame is greater than 60% These are generally based on the results of a survey reported by Reagan (1989). A binomial distribution is used to calculate the AEP as follows:

1352

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES

Where: Pe = Probability of one or more events occurring during a specified time period. n = specified time period, in years T = return period which is the inverse of the AEP The AEP is then calculated as follows: AEP = 1 – (1-Pe)^(1/n) Table 1 presents the resulting AEPs and return periods.

Table 1: Redefinition of AEP for tailings storage facility risk assessments Likelihood category

Annual Exceedance Probability (AEP)

Highly Likely

1/1 to 1/2

Likely

1/2 to 1/10

Probable

1/10 to 1/100

Unlikely

1/100 to 1/1,000

Highly Unlikely

1/1,000 to 1/10,000

Extremely Rare

1/10,000 to 1/100,000

Negligible

0 to 1/100,000

Note that in Table 1, the negligible likelihood threshold is set at below 1/100,000. FERC (2021a) indicates that a negligible failure mode is associated with an AEP of less than 1/1,000,000. When dealing with likelihoods that are this low, it can be a challenge to discriminate between a likelihood of 1/100,000 or 1/1,000,000. The AEP of 1/1,000,000 is considered further in the discussion on credibility in discussed below.

Confidence scale Another key aspect of a TSF risk assessment is the assignation of a confidence level to each assessed likelihood. Likelihoods are based on the information available for the risk assessment process and are analysed in a case-by-case basis. The authors use a three-level confidence scale, including high, medium, and low confidence levels, and its assignation is based on the confidence definition by FERC (2021b).

1353

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA In the risk assessments, if a low or medium confidence level is assigned, the likelihood category may be increased by at least one level. Depending on the risk category for that failure mode, it could trigger a risk reduction measure that may involve further studies and investigations to improve the confidence level.

Credible failure modes An important part of the risk assessment process for tailings facilities is the assessment of the credibility of the failure modes. This concept was described by the GISTM and there is no clear guidance on how to discern a credible failure mode. Small et al (2023) has provided some guidance on this topic. After the failure modes have been identified and physically impossible failure modes have been eliminated, a credibility assessment is conducted to determine the failure modes that will be carried further into the risk assessment step. As noted in the GISTM (GTR, 2020): “Credible failure modes refer to technically feasible failure modes given the materials present in the structure and its foundation, the properties of these materials, the configuration of the structure, drainage conditions and surface water control at the tailings’ facility, throughout its lifecycle. The likelihood of credible failure modes is not considered so remote to be negligible (FERC, 2021a). The term ‘credible failure mode’ is not associated with a probability of the event occurring and having credible failure modes is not a reflection of a facility safety.” As noted in Small et al (2023), this definition is not clear and discerning credible and non-credible failure modes has been the subject of much discussion since the release of the GISTM. There are two interpretations to the definition of credible failure modes: one is based on whether a failure mode is physically possible and the other includes consideration of negligibility that is clearly defined. The risk assessment process described in this paper can be applied for both definitions of credible failure modes. Regardless of the definition used, it is important to document the credibility assessment and basis for the determination so it can be reviewed and scrutinized in the future. The risk assessment is an effective place to document this, so the knowledge is not lost or forgotten. For companies that use physical possibility to discern between credible and non-credible failure modes, the failure mode and associated preventative controls are described in detail. If the failure mode is considered non-credible, then the risk assessment does not proceed further. For companies that consider physical impossibility and negligibility, if a failure mode meets the “negligibility” criteria noted above, a further, rigorous examination is required to discern if the failure mode meets the test of non-credibility. FERC (2021a) provides additional guidance: that a failure mode can be considered non-credible if the likelihood of the potential failure mode leading to a failure is minute; in quantitative terms, an AEP of less than 1/1,000,000. It is possible to have a failure mode with a negligible likelihood but is considered credible, and another failure mode with a negligible likelihood that is

1354

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES considered non-credible. The former failure mode may have an AEP of less than 1/100,000, but there is some uncertainty in the analysis or investigations that precludes the failure mode from being called noncredible. The latter failure mode would have an AEP of less than 1/1,000,000 and has a high degree of confidence in the analysis and investigations.

Risk assessment flow chart To summarize the approach proposed in this paper, Figure 3 presents a flowchart on the proposed tailings facilities risk assessment process.

TARPs and ERPs For mining companies that consider physical impossibility and negligibility when identifying credible failure modes, then these failure modes can be used to inform the TARPs and ERPs. For mining companies that consider only physical impossibility when identifying credible failure modes, then the physical possible failure modes need to be considered for TARPs and ERPs. Some companies are basing their TARPs and ERPs on all physically possible failure modes, while others establish separate approaches to identify the failure modes that require TARPs and ERPs.

Approach for different stages of a tailings storage facility Unlike a typical mine operation or planning risk, it is expected that risk assessments will be conducted at different stages – design, construction, operation, and closure – and although risk assessments cannot be properly conducted for design stages earlier than a feasibility level, risk principles can still be used to inform the design by considering the failure modes and the appropriate controls. Once a risk assessment is completed, it is good practice to review and update every two to three years or when there is a material change to a facility or to social, environmental, and economic context (GTR, 2020). A risk assessment should reflect the conditions at the time that the risk assessment is conducted, or it is possible to consider additional “temporal scenarios” that consider the different stages of a facilities development and life, including closure. Two points need to be considered: • The risk assessment framework described above is based on the typical life of a facility, i.e., less than 100 years. • When considering closure risks, a TSF may remain in place for a very long time, then using the Poisson formula above, the likelihood of failure modes that are physically possible will become one and the risk assessment reduces to a consideration of consequences for physically possible failure modes.

1355

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA It is often difficult to properly assess potential risks more than 20 to 50 years into the future based on uncertainties associated with climate change, potential downstream development, reliance on the TSF being developed as assumed at the time, advancements in engineering and technologies that may change the understanding of facility designs. Hence, the authors use the “temporal scenario” concept.

Figure 3: Tailings facilities risk assessment process

1356

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES For example, consider a mine that has been operating for 20 years with a life of mine ending in the next 10 to 15 years. There are four additional TSF dam raises planned, followed by removal of the tailings pond at closure. A risk assessment should be conducted on the existing conditions with the understanding that the risk assessment is only relevant for a specific timeframe. Separate “temporal scenario” risk assessments should be considered prior to the design of each raise, the end of operations, and to reflect the closure conditions. These allow a clear determination of the risks for each of these temporal scenarios.

Semi-quantitative and quantitative risk assessments The risk assessments used by mining companies are typically semi-quantitative. The framework described in this paper can be applied for both semi-quantitative and quantitative risk assessments. A quantitative estimate of likelihood may be required to better understand the probability of a failure mode and indicate the benefit of risk reduction measures.

Conclusion Risk frameworks used by mining companies are generally not suitable for TSF risk assessments due to the lower likelihoods associated with TSF failure modes. This paper presents an approach that adapts the existing frameworks used by mining companies to accommodate unique features of tailings facilities. Some mining companies have already modified their frameworks to accommodate TSF risks; this paper is intended to provide an alternative approach for mining companies that have not done so. Key modifications to adapt a typical 5 × 5 risk matrix risk framework to be suitable for TSF risk assessment include, but are not limited to: (i) extension of the risk matrix to a 5 × 7 to accommodate lower likelihood categories; (ii) adapt frequency/return periods for each likelihood category and estimate corresponding AEP for each likelihood level using Poisson distribution; (iii) including the concept of credible failure modes; (iv) supporting TARPs and ERPs; and (v) addressing the different temporal scenarios associated with the life phases of a TSF. This approach has been used for numerous TSF risk assessments and has been found to be effective.

Terminology The following terms have been used by the authors for several risk assessments: • System: a set of interrelated and interacting elements or components acting to achieve defined goals or functions.

1357

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Component: part of a system; can also be referred to as a sub-system. There can be multiple levels of sub-systems, the degree of breakdown being dependent on the context of the risk assessment. A component can be composed of two or more elements. • Element: a singular part of an integrated system, component, or process. There are a finite number of elements in any system. • Hazards: according to the USACE (2014), these are sources of harm to valued assets (lives, environment, economy, social, business). In tailings facilities risk assessment, a hazard is a circumstance or condition that could pose a risk to the dam safety e.g., earthquakes, floods, poor design, poor Construction Quality Control (CQC), etc. • Initiating event: is the loading mechanism, physical condition, combination of circumstances, or operational occurrence that initiates the failure process. This is sometimes referred to as the “trigger.” • Failure mechanism: is the underlying the process of failure, beginning with the initiator and ending at the system failure, the sequence/pathway or combination of causes/events/conditions that links the different levels of the failure modes and leads to a failure of either a component/element or of an overall system. • Failure mode: is the state or condition of loss of function, a description of the element /component/system in its’ failed state, or a deviation from its intended state. There are a finite number of failure modes for each element/component of the system. An example assuming a flood conditions: the initiating event would be the flood; the failure mechanism would be the rise in water level, overtopping the dam for a long enough period to cause erosion of the crest of the dam, and starting the breach process; and the failure mode would be breach by overtopping. • Failure Scenario: is the combination of the failure mode, the result failure that results and the associated consequences. • Preventative controls: are controls that are intended to prevent the occurrence of failure: o

Design controls: are part of the dam design (e.g., spillway, dam slopes, filter, riprap, etc.)

o

Operational controls: are related to the dam operation. For example, water management, beaching procedures, and loading rates.

• Surveillance: are actions related to monitoring (inspections, observations, etc.), instrumentation readings and interpretation. Note that surveillance is not a control. Surveillance can identify where a concern may be developing, assess the effectiveness of a preventative design control and support the implementation of measures to improve the design control. However, surveillance does not prevent a failure scenario directly and therefore is not considered as a preventative control.

1358

EXTENDING A MINING COMPANY’S RISK ASSESSMENT FRAMEWORK TO TAILINGS FACILITIES • Maintenance: area actions related to maintaining the structures and equipment so that they function as intended. • Mitigative controls: are intended to limit/reduce the consequences once the dam has failed. • Critical controls: are site-specific risk controls that are crucial to preventing a high consequence event or mitigating the consequences of such an event. The absence or failure of a critical control would significantly increase the risk despite the existence of other controls (MAC, 2021). • Incremental consequences: are consequences of a TSF failure above any damage or loss caused by the same initiating event if the TSF had not failed (ICOLD, 2005). • Residual risk: refers to a risk at any time frame, it could be assessed before, during or after a control is implemented (USACE, 2014). However, the authors recommend the use of the term “current risk” for the risks after existing preventative and mitigative controls are in place, and the term “residual risk” for risks remaining after further reduction/mitigation measures have been considered.

References Federal Energy Regulatory Commission (FERC). 2021a. Chapter 17 – Potential failure mode analysis. In Engineering Guidelines for the Evaluation of Hydropower Projects. Washington. Federal Energy Regulatory Commission (FERC). 2021b. Chapter 18 – Level 2 risk analysis. In Engineering Guidelines for the Evaluation of Hydropower Projects. Washington, pp. 26–27, 45–46. Global Tailings Review (GTR). 2020. Global Industry Standard on Tailings Management (GISTM). International Commission on Large Dams (ICOLD). 2005. A reconnaissance of benefits, methods, and current applications. In Risk Assessment in Dam Safety Management –- Bulletin 130. International Council on Mining and Metals (ICMM). 2021. Tailings Management, Good Practice Guide. Mining Association of Canada (MAC). 2021. Chapter 4 – Planning. In A Guide to the Management of Tailings Facilities. Version 3.2, pp. 19–20. Reagan, R.T., Mosteller, F., and Youtz, C. 1989. Quantitative meanings of verbal probability expressions. Journal of Applied Psychology 74(3): 433–442. Small, A., Küpper , A., Johndrow, T., Al-Mamun, M. 2023. Credible Failure Modes – Summary of 2021 and 2023 Workshops. Submitted in August 2023 for publication as part of the Proceedings of Tailings and Mine Waste Conference, 2023. U.S. Army Corps of Engineers (USACE). 2014. Chapter 2 – Dam Safety Program Framework. In Safety of Dams – Policy and Procedures. Washington, pp. 2–8.

1359

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

1360

Chapter Twelve

Site Investigation

1361

1362

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Balabag Tailings Storage Facility – Successful Construction Monitoring and Supervision Farzad Daliri, GHD, Vancouver, Canada Rob Longey, GHD, Hobart, Australia David Brett, GHD, Hobart, Australia Cliff James, TVI Pacific Inc., Calgary, Canada

Abstract The Balabag Tailings Storage Facility (TSF) is a 80 m high zoned earth and rockfill embankment with a central clay core transitioning to upstream for Stage 2 with filter drains. The tailings dam has been raised in three stages, first with a starter dam to crest RL442.5 (stage 1), and then to stage 2 RL465 and stage 3 RL482 by the downstream construction method. Several construction monitoring technologies and instrumentation methods have been used to provide better supervision and monitor the construction stages, including vibrating wire piezometers, settlement monitoring points, bathymetric surveys, and drone videos. Drone video has been used successfully for weekly surveillance of the TSF as part of an overall technical review role for the project. The construction monitoring role included full-time technical support by in-country personnel, with an annual in-person field inspection by senior personnel from Australia and Canada. The bathymetric survey included the impoundment survey presenting the tailings level, pond water level, tailings volume and longitudinal cross-sections providing the tailings pond profile and tailings beach slopes. This paper reviews the construction supervision, monitoring technologies, and the organization used to monitor the Balabag TSF construction stages, as an example of a successful supervision of a challenging Extreme consequence category tailings dam in a remote area.

Introduction The Balabag Gold-Silver Project is 100% owned by TVI Resource Development Philippines Inc. (TVIRD). The project is located at Sitio Balabag, Barangay Depore, Municipality of Bayog, Province of Zamboanga del Sur in the Philippines. The Balabag Tailings Storage Facility (TSF) is an 80 m high zoned earth and rockfill embankment with a central clay core transitioning to upstream for Stage 2 and future Stage 3, with filter drains. The dam has been raised in two stages, first with starter dam to crest RL 442.5 (Stage 1), and then to Stage 2 crest RL 465 by the downstream construction method with plans for further raising. The Stage 1 embankment

1363

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA has been completed and the TSFs were commissioned in July 2021. TVIRD is currently constructing the Balabag TSF to the approved Stage 2 crest level of RL 465. Upon completion of Stage 2 construction, the TSF is estimated to have a storage capacity of 1.3 Mm3, which is equivalent to approximately two years of storage, assuming the tailings density is 1.1 t/m3 and considering sediment generation of 1,200 m3/week. The Stage 2 Balabag TSF was previously assessed to be a “High A” Dam Failure Consequence Category (DFCC) dam in accordance with ANCOLD Guidelines on Consequence Categories for Dams (ANCOLD, 2012). This was due to the population at risk (PAR) of “10–100” based on a high-level dam break analysis and the environmental consequences in the event of failure when the dam is full of tailings (GHD, 2021). The dam break assessment has been conducted for Stage 3 and the results indicated that the DFCC will increase to Extreme. This paper aims to review the construction supervision methods and monitoring technologies of the Balabag TSF construction stages and operation.

Project details Geology The depth of the residual soils encountered under the TSF was approximately 2 m to 5 m. The surficial bedrock is extremely to highly weathered, and very weak to weak, becoming moderately to slightly weathered and medium strong to very strong at greater depth. Clayey residual soils consisting primarily of silty clay with some sand and gravel and with indications of boulders being present were encountered over the bedrock in the vicinity of the spillway. The materials were soft to firm, had a low plasticity, and were mostly covered with vegetation. Limestone and/or andesite clasts less than a maximum size of 4 cm are also present. Based on the borehole logs, the area is underlain by loose to medium dense silty to sandy soil from surface down to 4.55 m below the ground level. This was followed by highly to moderately weathered moderately jointed lithic tuff down to 14.0 m.

Seismicity The Philippine Islands are within a complex and rapidly deforming area between the two opposing subduction zones. In central Luzon, the Philippine fault zone (PFZ) appears to diverge into several northsouth trending splays. The Philippines is a seismically and volcanically active region where damaging earthquakes have struck numerous times within the 400-year historical period. The Balabag project area is in a relatively active seismic zone. A site-specific seismic hazard assessment (SHA) has been undertaken for Balabag TSF. This assessment is based on the historical seismicity in the region, taking into consideration the regional and local tectonic settings, as well as the effects these parameters have on the seismic hazard of

1364

BALABAG TAILINGS STORAGE FACILITY – SUCCESSFUL CONSTRUCTION MONITORING AND SUPERVISION the site. Peak Ground Acceleration (PGA) for Balabag TSF is estimated as being 0.460 g to 0.431 g for 1:475 and 1.127 g to 1.104 g for 1:10,000 AEP.

Hydrology The Probable Maximum Precipitation (PMP) for the Balabag site was developed by GHD during the Stage 2 design review works using the Generalized Short Duration Method (GSDM) and Generalized Tropical Storm Method (GTSM). These methods are adopted by the Australian Bureau of Meteorology (BOM). Table 1 presents the Probable Maximum Precipitation (PMP) depth values. Table 1: Probable maximum precipitation (PMP) depth Duration (hr)

0.5

1

2

3

4

5

6

24

72

PMP (mm)

335

489

736

898

1019

1126

1189

1,380

2,085

Cross-section The Stage 3A downstream batter slope was designed to be 2.5H:1V up to RL465 where the slope changes to 1.5H:1V up to the RL475 crest level. The downstream batter slope for Stage 3B was designed to be 2H:1V up to RL475; where the slope changes to 1.5H:1V up the RL482 crest level. The upstream batter slope for Stage 3A and Stage 3B are 2H:1V. The slope of clay core and filter zones are 2H:1V. The width of zone 1 clay core, thick finer zone 3A and thick zone 2A are 10 m, 2 m, and 2 m respectively. Figure 1 presents a general cross-section of Balabag tailings dam, including a possible Stage 4 raise to RL500 m.

Figure 1: Cross-section of Balabag tailings dam

Water management Water management includes designing emergency spillways for a Stage 2 and Stage 3 raise. The Stage 2 spillway included a gabion spillway on the crest and the batter slope connecting to a rock based downstream

1365

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA spillway. Stage 3 included two emergency spillways: Stage 3A spillway and Stage 3B spillway, excavated in rock in the left abutment, were designed for 1H:1V overall cut slope with 0.3 m bench width.

TSF Supervision and monitoring organization chart Figure presents the design, construction, and supervision/monitoring organization chart that has been used to date for the Balabag project. The design lead was based in Vancouver, organizing the weekly client meeting with TVI management, weekly drone footage meetings with the GHD/TVI site team and GHD senior reviewers, and a weekly meeting with design team in Chile and Philippines. The design lead also provided weekly progress reports for construction supervision. Employing this organization chart and with good collaboration between GHD and TVI, the construction monitoring of the Balabag TSF has been conducted successfully, resulting in the design and construction of an Extreme Consequence Category TSF in a remote area with a high level of QA and monitoring.

Figure 2: Balabag TSF design and construction monitoring supervision site organization chart During the COVID-19 lockdown period, and even in normal times, it has not been practical for a high level of detailed field inspection of construction projects in remote areas by experienced technical engineering staff. This was the case for the Balabag project. This could equally apply in developed countries, where the travel time to site takes more than a few hours. At the Philippine site, this issue was overcome using on-site, in-country technical engineering staff working on a rolling shift basis, supported by senior engineering personnel in Australia and Canada undertaking weekly Teams conferencing, and an annual site inspection by the ex-pat team. The weekly Teams conference reviewed drone video footage of the site. The videos have been found to be very useful and more informative than still images. The drone route typically follows an established pattern that gives an overview of the site, while more detailed close-up views can be requested by the reviewers if a particular issue arises.

1366

BALABAG TAILINGS STORAGE FACILITY – SUCCESSFUL CONSTRUCTION MONITORING AND SUPERVISION

Summary of instrumentation and monitoring Piezometers A total of 18 vibrating wire piezometers were installed within the TSF embankment to monitor the pore pressure during and post-construction. There were three arrays of piezometers, with six piezometers for each array to be installed in different zones. Upon installation, readings were initially taken daily to monitor construction pore pressures, particularly in Zone 1, and then on a weekly basis. The data is plotted by site personnel to detect any abnormal trend. Figure 3 and Figure 4 present the instrumentation plan and piezometer location section for Stage 3 Balabag TSF.

Figure 3: Stage 3 instrumentation plan

Figure 4: Proposed piezometer location section for Stage 3

1367

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Settlement monitoring points Four settlement monitoring points were installed on the downstream edge of the Stage 2 dam crest at 60 m intervals. At least two survey monuments should be installed in natural ground on a good foundation where settlement is not expected and there is minimal traffic from construction or disturbance from the community.

Bathymetric survey Bathymetric surveys measure the depth of a water body and also map the underwater features of a water body. The bathymetric surveys results can be used to estimate the remaining storage of the TSF and tailings settled density. Multiple methods can be used for bathymetric surveys, including multi-beam and singlebeam surveys. Bathymetric surveys for Balabag TSF pond are conducted monthly by TVI. Figure 3 presents the bathymetric survey results of the TSF pond conducted in June 2023. The settled density of the tailings was estimated to be 1.15 t/m3 based on the bathymetric survey results.

Figure 5: Examples of bathymetric survey results

Seepage collection pond A seepage collection pond and V-notch weir were installed in the downstream toe of Stage 2 to monitor the seepage coming from the drains, as shown in Figure 6. The photo shows the V-notch weir at the seepage collection pond with measured head at 75 mm (17.7L/s). The GHD design team inspects the seepage collection pond on a weekly basis. Flowing water has been consistently clear/colourless at times of inspection. The seepage collection pond/weir was be relocated to the Stage 3 toe line during Stage 3 construction.

1368

BALABAG TAILINGS STORAGE FACILITY – SUCCESSFUL CONSTRUCTION MONITORING AND SUPERVISION

Figure 6: Seepage collection pond and V-notch weir installed at the downstream toe of Stage 3 Figure 7 presents the monitoring results of the Stage 2 and Stage 3 seepage collection pond from October 2021 and compared them with the rainfall data. The average value of the seepage from the V-notch weir was estimated to be 17.7 L/s for Stage 2 and 24.9 L/s for Stage 3.

Figure 7: Results of monitoring the seepage collection pond for Stage 2 and Stage 3

1369

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Drone videos A drone has been used to collect video footage of the site for weekly inspection reviews by off-site personnel. A flight plan is prepared by the site engineer using way-points plotted on a drawing plan, taking into account the key areas that the site engineer aims to capture on video. The flight plan is then discussed with the drone operator, along with instructions such as having to zoom in on specific areas, to ensure that all the necessary details are clearly shown to the viewers and to allow for a more effective discussion. A copy of the drone footage is then obtained and sent out prior to the meeting to allow the reviewers to view the footage and note all the concerns that need to be raised to the site engineer and addressed by the site team. The drone videos were typically taken at the start of each week (Monday), unless weather conditions or other issues prevented this, in which case the flight would be undertaken at the earliest time thereafter. The videos were then emailed to Canadian and Australian staff for review the following day in a meeting between the onsite engineer and the offsite senior engineers. The meetings using Teams were typically of 30 minutes duration. The findings made during the review were then discussed by the site engineer with the client at a weekly meeting to ensure that the client construction supervisors understood the issues prior to giving out instructions to personnel in the field. The schedule was designed to capture the construction progress for the past week and give ample time for the site team to address all the concerns that were raised during the drone review meeting. The availability of video also simplified the client discussions. An important aspect of the system of video capture and review meetings has been the responsiveness of the client and construction team in addressing the concerns of the review team. Figure 8 shows an example of a drone photo for the Stage 2 completion.

Figure 8: Drone photo of Balabag embankment at Stage 2 completion (upstream side and pond)

1370

BALABAG TAILINGS STORAGE FACILITY – SUCCESSFUL CONSTRUCTION MONITORING AND SUPERVISION

Conclusion Balabag gold and silver TSF is an Extreme Consequence Category tailings dam in a remote area of the Philippines, designed by GHD. An organization chart involving several parties from Australia, Canada, the Philippines, and Chile resulted in collaborative work between TVI and GHD to successfully monitor and supervise the dam with a high level of quality assurance. In addition, several construction monitoring/instrumentation technologies were employed at the Balabag TSF to provide better supervision and monitoring of the construction stages including vibrating wire piezometers, bathymetric surveys, and drone videos. Drones have been proven capable of capturing a clear overview of site progress for technical and nontechnical viewers, providing flexibility for inspection covering specific areas of concern, particularly on matters that require immediate resolution and updates, and serving as a tool to identify potential dam hazards and risks that could lead to harm, damage, and failures. Importantly, they have proven to be an excellent tool for communication between designers and constructors, so as to allow timely correction of issues noted by external reviewers.

References ANCOLD. 2012. Guidelines on the Consequence Categories for Dams.

1371

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

1372

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Monitoring of Iron Tailings Saturation Using Water Moisture Sensors Luciano Souza, Klohn Crippen Berger, Brazil Thatyane Gonçalves, VALE S.A., Brazil Evelyn Santos, Klohn Crippen Berger, Brazil José Ccotohuanca, Klohn Crippen Berger, Brazil João Paulo Silva, VALE S.A., Brazil Andy Small, Klohn Crippen Berger, Canada Frank Pereira, VALE S.A., Brazil

Abstract This research explores the importance of considering the unsaturated nature of tailings in the Brazilian mining industry’s tailings characterization efforts. It presents a case study involving two upstream-raised dams in Brazil, discussing the geotechnical campaign and material characterization conducted to understand the facilities’ saturation conditions. The paper further discusses the installation of water content sensors to indirectly measure the degree of saturation, along with a brief overview of the calibration process implemented to enhance the precision of water content measurements in these sensors. The research demonstrates the significance of considering the calibration process when using such sensors in tailings engineering. The obtained results demonstrate that the sensor could accurately represent the expected saturation profile, with saturations close to 100% below the water table and gradually decreasing towards the surface. The case study and findings emphasize the importance of characterizing the unsaturated portion of tailings reservoirs to support further studies. Without adequate monitoring, the application of unsaturated soil mechanics can be approximate, particularly in climatic conditions similar to those in Brazil.

Introduction An old topic has recently resurfaced with the industry’s advancement of tailings characterization efforts in Brazil and the academic community. Often, especially with the traditional method of subaqueous deposition using slurry tailings, the concerns related to tailings characterization are focused on the saturated nature of these materials. However, the new methods of disposal, especially the filtered tailings deposits and the closure process of legacy facilities, have led many practitioners and academics to revisit the importance of

1373

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA considering unsaturated tailings. The unsaturated nature of the deposited tailings will influence not only the geomechanical behaviour of the deposited material (Reid and Smith 2021, Świdziński and Smyczyński 2022, Russell et al. 2022) but will also change its mobility in a dam breach scenario (Fontaine and Martin 2015, Chen, J.H., B. Chin and R. Friedel 2019), especially for tailings with a high solids content. In this way, to consider the application of a new unsaturated framework for tailings, one needs to consider the correct characterization using field and laboratory data, as well as instrumentation monitoring to validate the criteria used in the design and to understand how the saturation varies with time. This work aimed to present a campaign composed of field and laboratory tests and instrumentation installation in two upstream-raised dams in Brazil. The campaign’s main objective was understanding the facilities’ current and future saturation conditions. Combining CPTu testing, index testing, soil water characteristic curves, piezometric data, calibration of instruments – using the methodology presented by (Landim et al. 2023), and monitoring data, it was possible to define the saturation profile of multiple points in the facility. Only the data related to the water content sensors will be presented in this paper.

Case study: tailings dikes Dikes I and II are upstream-raised iron tailings storage structures in Brazil. Each structure has a compacted embankment starter dike, over saturated tailings disposed above residual/saprolite natural soils. Both structures are approximately 15 m high, with downstream slopes of 2H:1V. Figure 1 shows the typical cross-sections, the geotechnical units that comprise each section, and the phreatic condition. The location of each cross-section is shown in Figure 3.

Figure 1: Main section of Dikes I and II

Geotechnical campaign and material characterization Multiple CPTu/SCPTu and sample collections were performed over the last three years as part of the geotechnical characterization design of the two dams. The main objective of these tests was to feed the geotechnical and hydrogeological models. The location of the samples is illustrated in Figure 3a.

1374

MONITORING OF IRON TAILINGS SATURATION USING WATER MOISTURE SENSORS Complemented by instrumentation data, it was possible to verify that a significant part of the current tailings’ reservoirs is in an unsaturated condition, with the current water level close to the foundation, similar to what is observed in Figure 1. Between 2022 and 2023, a geotechnical campaign was performed, and disturbed and undisturbed samples were collected in both dikes at the same points where the instruments were installed – Figure 3a. The results from index testing showed similar conditions for both structures, with a predominance of finegrain sand with no plasticity. Figure 2 and Table 1 present the grain size distributions for the tailing samples collected in Dikes I and II. 100%

90%

80%

PASSING (%)

70%

60%

50%

40%

30%

DI-WC-01 DI-WC-02 DI-WC-03

20%

DI-WC-04 DII-WC-01

10%

DII-WC-02 DII-WC-03

0% 0,001

0,01

0,1

1

10

PARTICLE SIZE (mm)

Figure 2: Grain size distribution – Dikes I and II Table 1: Summary of laboratory tests results Sample

Depth (m)

Gs

Bulk unit weight (g/m³)

Dry unit weight (g/m³)

Water content (%)

DI-WC-01

1.20–2.00

3.20

2.42

2.13

13.4

DI-WC-02

1.80–2.20

3.39

2.52

2.24

12.8

DI-WC-03

1.80–2.00

3.45

2.37

2.21

7.1

DI-WC-04

1.80–2.00

3.27

2.45

2.18

12.2

DII-WC-01

1.80–2.00

3.37

2.03

1.92

5.7

DII-WC-01

1.80–2.00

3.46

2.11

2.01

5.3

DII-WC-01

1.80–2.00

3.57

2.45

2.05

19.3

Due to the phreatic level of Dikes I and II, to assess the punctual and temporal changes in volumetric water content (q) and estimate the corresponding saturation (%Sr) profile, 22 water content sensors were strategically installed at various depths within both structures. Figure 3 provides a visual representation of the sensor locations.

1375

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Instrumentation campaign Two types of instrumentation were considered to measure the saturation degree using instrumentation indirectly. First, using water content sensors and measurements of tailings’ physical properties (Table 1) and second, through tensiometers in conjunction with soil-water characteristics curves (SWCC) defined in the laboratory, which this paper will not present. The instrumentation campaign included the two approaches indicated above. For such, several tensiometers were installed at the exact location of the water content sensors to have a direct instrumentation comparison. Given that the data obtained from the tensiometers is currently undergoing evaluation, this paper only presents the results derived from the water content sensors. The water content sensor’s location and depth of the sensor’s installation are shown in Figure 3.

Dam

Dike I

(a)

Dike II

(b)

Instrument ID

Depth (m)

DI-WC-01A

1.0

DI-WC-01B

2.0

DI-WC-01C

6.0

DI-WC-02A

1.5

DI-WC-02B

3.5

DI-WC-02C

7.5

DI-WC-03A

3.0

DI-WC-03B

8.0

DI-WC-03C

12.0

DI-WC-04A

2.0

DI-WC-04B

7.0

DI-WC-04C

11.0

DII-WC-01A

3.0

DII-WC-01B

8.0

DII-WC-01C

12.0

DII-WC-02A

3.0

DII-WC-02B

7.0

DII-WC-02C

13.0

DII-WC-02D

17.0

DII-WC-03A

3.0

DII-WC-03B

8.0

DII-WC-03C

12.0

(c)

Figure 3: Spatial distribution of water content sensors (a) Sensor’s location (b) Water content sensor specification (c) Depth of sensor installation The water content sensor instrument chosen to monitor the water content variation was Rika RK51001. It operates based on the principle of electromagnetic pulse, specifically Frequency Domain Reflectometry (FDR), to measure the soil’s dielectric constant. The instrument has a stainless-steel probe

1376

MONITORING OF IRON TAILINGS SATURATION USING WATER MOISTURE SENSORS that can be inserted into the soil surface or profile, allowing real-time measurement of the soil’s volumetric moisture content. The instrument’s dimensions, cross-section, and image can be seen in Figure 3b, with the technical specifications in Table 2. Table 2: Technical specification of the Rika RK510-01 water content sensor Item

Technical specification

Range

0 – 100%

Accuracy

± 2% (0 – 50 %)

Output signal

4-20mA, 0-5V, 0-2V, RS485 optional

Response time

0.60

Figure 3 shows the results of the brittleness assessment for the normal and quick probes at TP14. Thin lines trace the IB values with depth, thick lines show the inferred level of the water table, and dashed lines delineate the extent of the main liquefaction zone. Figure 3 indicates that the data from Contractor A (red) consistently classifies the brittleness of the tailings as low to moderate, whereas data from Contractor B (black) classifies the brittleness as high. The differences in brittleness classification fall back to differences in the measurement of the primary CPT parameters by the two contractors (see Figure 4). This figure, pertaining to position TP14, confirms lower tip resistances (qc), higher sleeve friction (fs), and lower dynamic pore pressures (u2) measured by Contractor A. Considering that equilibrium pore pressures at TP14 were higher during the investigation by Contractor A, it could be expected that Contractor A should have measured lower tip resistances. It is noted that the differences were generally more pronounced in areas where high dynamic PWPs were encountered and at the lower ranges of cone resistance. Cone size (10 cm2 versus 15 cm2) and pore fluid (silicon oil versus glycerin) were investigated as possible causes of the differences in brittleness classification. Cone size did not have an effect over and above natural expected variations. The impacts of the pore fluid on the measured u2 (desaturation) and inferred u0 (PPD deviations) were however significant, the former more so than the latter. Finally, verification of the bespoke design and construction aspects of the respective CPTs will require sophisticated calibration studies.

1465

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 4: TP14 showing differences in primary parameters (qc, fs, u2, and u0) as measured by the Contractor A versus Contractor B Regardless of the remaining uncertainties, the following recommendations are offered: • Use of silicon oil as pore fluid for the measurement of primary parameters during the first quick probe, and then converting to glycerin for the second probe to get the best PPD data. • Awareness of the fact that apparent small differences in the measurement of primary parameters could result in large differences in liquefaction potential and brittleness classifications. In critical cases, it may be worthwhile to verify results from one contractor by using a second contractor. • The influence of equilibrium pore pressures, especially changes in the pore pressure regime between CPT campaigns, must always be considered. Thus, accurate PPD data are required that is backed-up by reliable piezometer data. • Researchers are encouraged to consider studies in laboratory calibration chambers where fully assembled cone systems can be tested. A key consideration would be calibration of the entire penetrometer system, with all its modules, where both total stress and pore pressure can be controlled and measured. One objective of the calibration chamber experiments is to confirm whether there are

1466

BRITTLENESS OF IRON ORE TAILINGS – FACT OR ARTIFACT, A CASE STUDY possible deviations of the sleeve measurements in response to pore pressures, similar to the tip endare-ratio effect. Another is the performance of various pore fluids (silicon oils and glycerin) during penetration and PPD.

Conclusions A case study of two CPT investigations at an iron ore TSF was presented to highlight the influence of the pore fluid used to saturate piezocones and its apparent impact on the brittleness classification of tailings. The pore fluids considered in this case study include silicon oil and glycerin. The fact that different brittleness classifications can potentially result from only using different saturation fluids is concerning. The differences may also be related to the designs of the respective bespoke CPT systems, including measurement of tip resistance and sleeve friction, interference between the different measurement components, and possible influences of different tailings materials with different characteristics. To resolve these design issues will require careful and controlled laboratory experimentation. For CPT investigations on critical facilities and where liquefaction and brittleness are concerns, it is strongly recommended to allow for dual CPT probes at each test location. The first probe is a quick semicontinuous probe using silicone oil as saturation fluid for the measurement of the primary CPT parameters: qc, fs and especially u2. The second probe, using glycerin, allows for more reliable pore pressure dissipation tests, as well as stoppage time for the measurement of other secondary parameters such as seismic velocities.

References Bishop, A.W. 1967. Progressive failure – with special reference to the mechanism causing it. In Proceedings, Geotechnical Conference on Shear Strength Properties of Natural Soils and Rocks. Oslo. 2: 142–150. Bishop, A.W. 1973. The stability of tips and spoil heaps. Quarterly Journal of Engineering Geology and Hydrogeology 6(3–4): 335–376. Jefferies, M. and Been, K. 2016. Soil Liquefaction: A Critical State Approach. 2nd edition. CRC Press. Lunne, T., Powell, J.J.M. and Robertson, P.K. 2002. Cone Penetration Testing in Geotechnical Practice. CRC Press. Lunne, T., Strandvik, S., Kåsin, K., L’Heureux, J-S., Haugen, E., Uruci, E., Veldhuijzen, A., Carlson, M. and Kassner, M. 2018. Effect of cone penetrometer type on CPTU results at a soft clay test site in Norway. Cone Penetration Testing 2018: 417–422. CRC Press. Macedo, J. and Vergaray, L. 2021. Properties of mine tailings for static liquefaction assessment. Canadian Geotechnical Journal 59(5): 667–687.

1467

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Paniagua, P., Lunne, T., Gundersen, A., L’Heureux, J-S. and Kåsin, K. 2021. CPTU results at a silt test site in Norway: effect of cone penetrometer type. IOP Conference Series: Earth and Environmental Science 710. Reid, D. 2012. Update on the Plewes method for liquefaction screening. In Proceedings of Tailings and Mine Waste, 2012. Keystone, CO. 337–345. Robertson, P.K. 2009. Interpretation of cone penetration tests – a unified approach. Canadian Geotechnical Journal 46(11): 1337–1355. Teh, C.I. and Houlsby, G.T. 1991. An analytical study of the cone penetration test in clay. Géotechnique 41(1): 17–34.

1468

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Characterization of In-Situ State Parameter through Tube Measurement and CPTu Methods in Partially Saturated Tailings Ezra Coyle, SLR Consulting, USA Bobby Otieno, Klohn Crippen Berger, Australia David Reid, University of Western Australia, Australia

Abstract Current in-situ techniques for measuring state parameter (y) are commonly based on CPTu empirical correlation, with CPTu numerical methods calibrated to critical state laboratory testing becoming more common. For tailings that are near but not fully saturated, these empirical methods are not reliable. Understanding how loose the tailings are in situ through the y becomes critical to determine their liquefaction potential and subsequent strengths. This paper presents the assessment of state parameter through an extensive site investigation and established empirical methods. The methods and results are presented for the measurement of y through the collection of piston tube samples to the full depth of the tailings column at an upstream gold mine tailings storage facility (TSF). With the aid of two corresponding critical state lines (CSL), these samples were used to determine void ratio, saturation, density and hence y. Adjacent to where the samples were collected, cone penetration tests (CPTu) were completed and used to calculate y through the Plewes and cavity expansion methods for comparison to the tube y measurements. The results were compared using two CSLs for the respective tailings, to improve confidence in either technique or highlight factors for consideration when utilizing both methods. The validity of the CPTu methods in near saturated tailings is discussed for agreement and apparent limitations.

Introduction The evaluation of state parameter (y), particularly in an upstream raised Tailings Storage Facility’s (TSF’s) structural zone is an important consideration when assessing liquefaction potential and post liquefied strengths, and predicting subsequent stability outcomes. State parameter represents the distance between the void ratio of a material and the critical state line (CSL) at the same mean effective stress (p’) (Been and

1469

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Jeffries, 1985). The current industry standard for evaluation of y is based on CPTu empirical methods, with CPTu numerical methods calibrated to critical state laboratory testing becoming more common. In instances where the tailings are near but not fully saturated (partially saturated), matric suction alters the measured cone penetration resistance (qc), and these methods cannot be relied upon. Emerging techniques for the evaluation of y using cone penetration testing (CPTu) in partially saturated tailings require a programme of CPTu calibration chamber work, at a laboratory bench scale (Russell et al., 2022). This paper presents a site investigation performed at an upstream gold mine tailings storage facility (TSF), that aimed to improve confidence in the estimates of saturation level (Sr), in-situ void ratio (e0) and y through the collection of 82 piston tubes and subsequent laboratory work.

Field program A site investigation involving the successful collection of 82 piston tube samples was performed at the upstream raised TSF in predominantly nearly (but not fully) saturated tailings. The objectives of the investigation were the direct measurement of Sr and e0 for the tailings column, to determine its susceptibility to liquefaction should a potential trigger occur. Complimentary piezocone penetration tests with seismic measurement (SCPTu) were performed adjacent to the sampling locations to compare direct laboratory measurements against seismic measurements and CPTu empirical correlations. The samples were then transported to a soils laboratory where a program comprising index testing and advanced strength tests was completed. The scope of advanced testing is not covered in this paper in detail but included measuring the CSL for two batches of the tailings based on current state of practice methods (Reid et al., 2021). The sections below detail the procedures followed to collect the field tube samples, and handle, transport, and complete measurements in the laboratory.

Field tube sampling procedure A total of 82 thin-walled 86 mm ID, 600 mm long tube samples were retrieved from eight locations along the internal beach perimeter of the TSF using a custom piston sampler in general accordance with the procedures outlined in ASTM D1587 (2016). The samples were retrieved at depths of up to 43 m. The sampler was deployed from a CPTu rig, to ensure no introduction of water associated with traditional drilling. The custom piston tube sampling apparatus shown in Figure 1 uses a CO2 gas injection system that “cuts” the retrieved sample just above the level of the tube’s bottom cutting edge and breaks the suction force generated at the base of the sample. Previous attempts to recover samples of smaller diameter without gas cutting had generally been unsuccessful.

1470

CHARACTERIZATION OF IN-SITU STATE PARAMETER THROUGH TUBE MEASUREMENT AND CPTU METHODS IN PARTIALLY SATURATED TAILINGS

CO2 INJECTION HOLE

Figure 1: Tube sampler showing gas cylinder (left) and tube cutting face (right) showing co2 injection hole and squared off sample In total 97 samples were attempted, giving a retrieval rate of approximately 85 per cent. After tube retrieval, the samples were squared-off at both ends until a uniform, planar surface was achieved. This allowed for a neater fit of the sealing plug and greater accuracy when estimating sample length and associated in-situ sample volume. Once a uniform, planar surface was achieved, the distance from the sample surface to the top of the tube was measured at three points using a vernier (at both ends). This measurement, along with the tube’s total length and ID, was accepted as the best estimate of the sample’s total volume (Vt), representing the closest approximation to in-situ void ratio. The samples were then weighed and packed in boxes with padding and transported vertically to the soil laboratory. Despite efforts made to reduce transportation-induced densification, some densification of low Plasticity Index (PI) samples was expected and observed. This was noted and considered for corrections when measurements were taken in the laboratory. At four of the eight sampling locations, complimentary shear wave measurements were taken from the CPTu rig at downhole depth intervals of 1 m, prior to performing the piston tube sampling. The other four sampling locations only had traditional CPTu soundings performed (no seismic).

Laboratory procedure Laboratory measurements taken on the recovered tubes were carried out as follows:

1471

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA • Upon arrival in the laboratory the specimens were re-weighed, with the masses measured in the laboratory consistently matching the site measurements, confirming that the sample sealing procedures had been successful at avoiding moisture loss. • Specimen dimensions were taken again in the laboratory in two ways: o

Distance to the J-plug seals at each end, along with thickness of each J-plug.

o

Direct measurement to the sample after removal of the J-plug. This measurement consistently indicated greater distances between the specimen and the end of the tubes (i.e., a smaller sample length) than that taken in the field, consistent with transport-induced densification.

• The entire specimen was then extruded into a large drying tray, then allowed to dry at room temperature for about one week. The mass of tray, wet specimen and “dry” specimen after air drying were recorded in this process. • The specimen was then transferred to a sample bag and mixed thoroughly. A subsample was taken for Gravimetric Water Content (GWC) measurement in a hot (~105°C oven). Only a small portion of the sample was exposed to hot oven drying to avoid potentially affecting the mechanical behaviour and mineralogy of the majority of the recovered material which was needed for subsequent reconstituted triaxial and direct simple shear testing. The process of measuring initial moisture loss of the entire sample through air drying, and the GWC of a sub sample, enabled the original mass of water and solids in the piston sample to be calculated. • In some cases, the process of subsampling and hot oven drying was carried out in a beaker, facilitating measurement of the salt concentration of the pore fluid. This additional step involved: o

Addition of a known mass of deionized water to the specimen in the beaker, after drying.

o

Thoroughly mixing the specimen and deionized water, then leaving the soil to settle.

o

Removing clear water from the top of the beaker after settling, then measuring the salt concentration of the mixture. This enabled calculation of the original salt concentration of the water within the piston sample. If this salt concentration was sufficiently high, it can affect the density of the fluid. This can result in either underestimating or overestimating the sample and/or fluid density.

Tailings properties The properties of the piston tube samples were characterized through index tests that included particle size distributions (PSDs) through hydrometer tests, water contents and salt content tests. The results of the PSDs are shown in Figure 2, indicating a predominantly fine-grained low plasticity silt. Specific gravity of solids ranged from 2.73 to 2.85.

1472

CHARACTERIZATION OF IN-SITU STATE PARAMETER THROUGH TUBE MEASUREMENT AND CPTU METHODS IN PARTIALLY SATURATED TAILINGS

Figure 2: Particle size distribution test results The salt content tests were carried out to determine the effect of dissolved salts on the saturation level. The salt content test results presented a slight to negligible difference on the overall saturation level in comparison to the zero-salt case, and the initial base case assumption of 40 g/L used as a sensitivity. The results and comparisons of the salt content tests are presented in Table 1. Table 1: Salt content results Tube sample

Salt content (g/L)

Saturation (actual)

Saturation (base case)

Saturation (0 g/L)

Difference

PT01 01

11.3

78%

80%

78%

-2%

PT01 02

15.3

90%

91%

89%

-1%

PT01 06

27.3

79%

79%

77%

0%

PT01 07

19.2

130%

131%

129%

-1%

PT01 08

15.5

79%

81%

78%

-2%

PT01 09

6.2

78%

79%

77%

-1%

PT01 010

14.2

87%

88%

86%

-1%

PT01 012

11.8

97%

99%

97%

-2%

1473

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Two composite CSLs were performed for the facility prior to the tube and CPTu campaign (a 2018 campaign). One was completed for a tailings batch composition to represent the upper tailings (0 to 33 m) with less fines content, and the second, a composition to represent the lower tailings (below 33 m) with higher fines content. The two CSLs are presented in Table 2 with their respective fines content.

Table 2: Critical state line test results Material composite

Mtc1

l e2

G3

Fines content (%)

Upper tailings CSL (0 to ~33 m depth)

0.80

0.043

0.80

64

Lower tailings CSL (below ~33 m)

0.83

0.043

0.83

75

1. Mtc – Critical friction ratio 2. le – Slope of CSL, for semi-log idealization (base e) 3. G - Reference void ratio on CSL

State parameter outcomes CPTu based y interpretations of the eight soundings were processed using current techniques (Plewes et al., 1992; Shuttle et al., 2016) and are plotted alongside the tube-derived y interpretations for three representative soundings in Figure 3. The CPTu campaign was performed at the standard penetration rate (2 cm/sec) and was interpreted as predominantly drained penetration (-0.02SD) materials. This is attributed to the contribution of the negative pore-water pressures during penetration that are generated in the dilative material.

Pore-water pressure dissipation results

Figure 3: Three typical PPD test results including both (u2) and (u1) readings versus time As previously indicated, several PPD tests were conducted during the investigation. Figure 3 shows three typical PPD test results including both (u2) and (u1) readings versus time at three different depths. The PPD results at a depth of 11.90 m show almost immediate excess pore-water dissipation when penetration stopped. On the other hand, the PPD results at a depth of 16.15 m show an initial pore-water pressure of approximately 280 kPa (u1) and 344 kPa (u2), respectively. Surprisingly, u2 started larger than u1, which can be attributed to the effect of clamping at the beginning of the PPD test. The (u1) slightly increase followed by a rapid sharp decrease in pore-water pressure to the in-situ value of 103 kPa (u0). The u2 decreases almost

1538

USE OF CPT WITH DUAL PORE-WATER PRESSURE FILTER ELEMENTS IN CHARACTERIZATION OF MINE TAILINGS immediately after the penetration stopped. Finally, at a depth of 19.02 m and after the penetration stops, the pore-water pressure first shows a slight increase in magnitude followed by a subsequent decrease of porewater pressure in time towards the in-situ value. In all three cases shown in Figure 3, the elapsed time is relatively small and typically less than 300 seconds. The dissipation at 11.90 m is almost immediate, while at 16.15 m it is a little bit delayed, by up to about 60 seconds. The dissipation at 19.02 m was stopped before the in-situ value was reached.

Olson (2009) susceptibility-compressibility and (u1-u0) trends Figure 4a shows the results of the measured pore-water pressure at filter location (u1) during CPT penetration and the in-situ (u0) pore-water pressure distribution measured with vibratory wire piezometers. It can be seen that except for the upper 2.6 m where predrill occurred and (u1) @ (u0), in general u1 is greater than u0. Figure 4b shows the difference (u1-u0) versus depth of the same sounding.

Figure 4: (u1), (u0) and (u1)-(u0) versus time and corrected normalized tip resistance qt1 versus effective stress sʹvo Figure 4c shows the corrected normalized tip resistance qt1 versus effective stress sʹvo, of the same CPT sounding as Figures 4a and 4b. Also included in Figure 4c are the boundary lines separating dilative from contractive behaviour based on compressibility proposed by Olson (2009). The compressibility boundary lines are based on the slope of the critical state line, l10, and were developed based on estimated compressibility values from case histories. The three compressibilities are associated with a l10 = 0.03 for

1539

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA low compressibility materials, l10 = 0.06 for medium compressibility materials, and l10 = 0.17 for high compressibility materials. The mine tailings associated with the facility reflected in the CPT in Figure 4 are classified as high compressibility with l10 ranging within 0.10 to 0.20. Most of the soundings show the corrected normalized tip resistance qt1 is on the contractive side of the High Compressibility Boundary line. It is interesting to note that there are some corrected normalized tip resistance qt1 values that are on the dilative side of the boundary line. Furthermore, the zones where qt1 falls on the dilative side correspond to depths (i.e., effective stresses) in Figure 4b in which the difference (u1-u0) is less than approximately 200 kPa. On a preliminary basis, the line for the difference (u1-u0) of 200 kPa appears to be related to the boundary line separating dilative from contractive behaviour of high compressible mine tailings. Further assessment is warranted to provide firm evidence regarding this observation.

Summary and conclusions The results of a comprehensive field investigation program conducted across three different TSFs using CPT with dual pore-water pressures filter element locations (u1 and u2) and presented and discussed. Typical CPT results with dual filter elements are presented and relevant features are highlighted in relation to tailings characterization, state, aging, and stress history, as well as other behaviour factors. The following presents a summary and the conclusions of the work: 1. Figure 2a summarizes the results of the 6 CPTs with dual filter elements at the three TSFs using the Robertson (2016) behaviour classification chart. 2. Figure 2b through 2d summarize the general trends for (u2-u0)/(u1-u0) ratio, and Table 1 summarizes the numerical values. In general, the (u2-u0)/(u1-u0) ratio is highest for the claylike materials and lowest for the sand-like materials, with the transitional materials showing intermediate values. 3. Within each material-like the contractive materials display a higher (u2-u0)/(u1-u0) ratio than its dilative counterpart. This is attributed to the contribution of the negative pore-water pressures during penetration that are generated in the dilative material. 4. The overall dissipation behaviour in the CPTs with dual filter elements is similar to that observed in a regular CPT piezocone. 5. Figure 4c summarizes the results from one CPT in terms of the corrected normalized tip resistance qt1 versus the effective stress sʹvo, including the boundary lines separating dilative from contractive behaviour based on compressibility proposed by Olson (2009).

1540

USE OF CPT WITH DUAL PORE-WATER PRESSURE FILTER ELEMENTS IN CHARACTERIZATION OF MINE TAILINGS 6.

Figure 4b suggests that the difference (u1-u0) of 200 kPa appears to be related to the boundary line separating dilative from contractive behaviour of highly compressible mine tailings. Additional assessment of this aspect is needed.

7. Pending additional analysis, other pore-water differences (u1-u0) could be related to state parameters. The authors continue performing additional assessments of the data with the purpose of establishing other possible correlations. Those results will be presented in future publications.

References Baligh, M.M. 1986. Undrained deep penetration, II: pore pressures. Géotechnique 36(4): 487–501. Burns, S.E. and Mayne, P.W. 1998. Monotonic and dilatory pore pressure decay during piezocone tests. Canadian Geotechnical Journal 35(6): 1063–1073. Campanella, R.G., Robertson, P.K., and Gillespie, D. 1986. Factors affecting the pore water pressure and its measurement around a penetrating cone. In Proceedings of the 39th Canadian Geotechnical Conference, Ottawa, pp. 291–299. Campanella, R.G. and Robertson, P.K. 1988. Current status of the piezocone test. Penetration Testing 1988 (1), Balkema, 93–116. Contreras, I.A. and Grosser, A.T. 2009. Evaluation of CPT response under fast penetration rate in silty soils. In Proceedings of the 57th Annual Geotechnical Engineering Conference. Minneapolis, Minnesota, February 2009. Contreras, I.A. and Harvey, J.W. 2021. The role of the vane shear test in mine tailings. In Proceedings of Tailings and Mine Waste 2022. Jamiolkowski, M., Ladd, C.C., Germaine, J.T. and Lancellotta, R. 1985. New developments in field and laboratory testing of soils. In Proceedings of the 11th International Conference on Soil Mechanics and Foundation Engineering. San Francisco, California, August 1985, vol. 1: pp. 57–153. Lunne, T., Powell, J. and Robertson, P. 1996. Use of piezocone tests in non-textbook materials. In Advances in Site Investigation Practice. Proceedings of the international conference held in London on 30–31 March 1995. Mayne, P.W., Kulhawy, F. H. and Kay, J.N. 1990. Observations on the development of pore-water stresses during piezocone penetration in clays. Canadian Geotechnical Journal 27(4): 418–428. Olson, S.M. 2009. Strength ratio approach for liquefaction analysis in tailings dams. In Proceedings of the 57th Annual Geotechnical Engineering Conference. Minneapolis, Minnesota, February 2009.

1541

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Robertson, P.K. 2016. Cone penetration test (CPT)-based soil behaviour type (SBT) classification system – an update. Canadian Geotechnical Journal 53(12): 1910–1927. doi:10.1139/cgj-2016-0044 Terzaghi, K., Peck, R.B. and Mesri, G. 1996. Soil Mechanics in Engineering Practice. John Wiley & Sons. Wroth, C.P. 1984. The interpretation of in-situ soil tests: 24th Rankine Lecture. Géotechnique 34(4): 449–489.

1542

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Analysis of Vibrational Waves Induced by Machinery in Tailings Dams Ricardo Cabette Ramos, Vale S.A., Brazil Thatyane Martins Gonçalves, Vale S.A., Brazil Gino Calderon Vizcarra, Vale S.A., Brazil Frank M.S. Pereira, Vale S.A., Brazil

Abstract The liquefaction phenomenon is a significant cause of failure of dams. Forensic investigations showed that liquefaction was the main cause of the failure of tailings dams in Brazil that occurred recently. Liquefaction has received special attention in decommissioning designs that have the goal of increasing the physical integrity and operational safety of tailings dams. Induced vibrations have been studied in order to characterize the main sources of vibrations during the operation of construction machinery for dam reinforcement works. The methodology consisted of the utilization of engineering seismographs to characterize induced vibration sources that may trigger the liquefaction of a tailing dam. This paper presents the results and interpretation of ground vibrations developed on the surface and subsurface and generated by different mechanical sources such as a bulldozer, a hydraulic excavator, and a dump truck. These were used in test areas: a trial embankment and an experimental excavation in a tailings storage facility. The analysis showed that the waves measured at depth had a lower amplitude than the waves measured at the surface. This behaviour was systematically similar for all equipment studied. The values were approximately a third in the case of the excavation and a half for the embankment. Different curves and equations were obtained showing the damping of vibrations versus distance, on a daily basis, and by equipment and activity during a determined period. It was verified that the vibrations generated by the dump trucks had a very low intensity. Based on this study, it is possible to determine vibration thresholds that permit safe conditions during activities for the reinforcement of tailings dams.

Introduction After recent tailings dam failures in Brazil, seismic monitoring and the study of induced vibrations have become essential for the elimination of upstream raised dams. Zhang et al. (2022) reported that induced vibrations can potentially trigger liquefaction of tailings. Due to this fact, mining companies have been

1543

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA increasingly investing in technology to ensure the physical integrity and operational safety of geotechnical structures during the closure process, aiming to assess wave transmission and attenuation over distance (Sousa et al., 2021). Thus, this study aims to verify the levels of vibration induced during operations, analysing the characteristics of their intensities and the attenuation patterns as a function of the distance (Dowding, 1985; Nateghi, 2011). Field tests were conducted using engineering seismographs installed at different locations and depths, based on the geometric characteristics of the structures and the vibration source, to monitor the behaviour and attenuation of vibrations generated by three types of equipment: a bulldozer, a hydraulic excavator, and a dump truck. This work allows the evaluation of thresholds and triggered action response plans (TARPs) to be used in excavations, during the closure process of the area.

Case study: instrumentation campaign The instrumentation campaign aimed to achieve continuous monitoring during experimental excavations and complementary works, including the construction of embankments in the excavated area, as well as conducting vibration attenuation tests of equipment at varying distances. This campaign was carried out was performed by installing instruments in two locations for each test area. The first location named Position 1 was into the tailings area with predominantly sandy silt granulometry, the second location named Position 2 was into the natural soil area which material is classified as clayey sandy silt. Twelve engineering seismographs were utilized, with five equipped with depth geophones installed in 8-meter boreholes, and the remaining on the surface. During the excavation and embankment construction process, five of the surface seismographs were positioned near the depth seismographs to enable comparisons of vibration intensities during the execution of these activities. It should be noted that during the development of the short-duration attenuation tests, five surface seismographs were used, repositioned along the excavation axis at distances of 3 m, 6 m, 12 m, 25 m, and 50 m, and arranged perpendicular to the test track where the selected equipment was operated. The depth seismographs GFP-04, GFP-05, and GFP-02 were also aligned with the surface seismographs at fixed distances, while GFP-01 and GFP-03 were positioned laterally at slightly variable distances depending on the equipment movement. The locations of the instrumentation were predetermined to generate vibration attenuation curves for the different excavation and transport equipment operations. The arrangement of the seismographs was similar for Position 1 and 2, as shown in Figure 1.

1544

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS

Figure 1: Representation for Positions 01 and 02 of the arrangement of surface and depth seismographs (source: Vale S.A.)

It is important to emphasize that the behaviour and attenuation of these induced vibrations were evaluated for all four types of equipment used in the construction, namely the Komatsu D61EX Bulldozer, Hyundai 220LC and Komatsu 350LC Excavators, and Scania P 450XT Dump Truck. Some of the key characteristics of this equipment are listed in Table 1.

1545

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Table 1: Characteristics of the monitored equipment Dimension

Type of equipment

Weight

Power

Hydraulic Excavator Hyundai 220LC

22,200 kg

155 HP (2.000 rpm)

9,65 x 2,99 x 2,92

Hydraulic Excavator Komatsu 350LC

35,000 kg

260 HP (194kW)

11,31 x 3,00 x 3,13

Dump Truck Scania P450 XT

33,100 kg

480 HP (353kW) 1900rpm

7,08 x 2,6 x 3,545

Bulldozer Komatsu D61EX

19,000 kg

170 HP (127 kW)

4,80 x 3,35 x 3,02

Bulldozer CAT D 6N

17,997 kg

148 HP

5,82 x 2,95 x 2,97

(length x width x height)

During the vibration testing stages, successive passes of the earthmoving equipment used in the operations were evaluated. In these tests, the equipment would travel through the tracks that composed the Experimental Excavation area one by one, typically with three passes for each equipment. Additionally, specific activities of each equipment were performed, such as: • Dump trucks: moving loaded or unloaded (Figure 2).

Figure 2: Damping tests operations of the dump truck: loaded dump truck (Source: Vale S.A.)

1546

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS • Bulldozers: moving or operating with blade (Figure 3).

Figure 3: Damping tests operations of the bulldozer: bulldozer operating with the blade (source: Vale S.A.) • Hydraulic excavator: moving, excavating, or striking the ground with the bucket (Figure 4). It is important to emphasize that the operation of the excavator striking the ground with the bucket is not a normal operation of the equipment and is considered an operational error. However, it needs to be properly studied and analyzed.

Figure 4: Damping tests operations of the hydraulic excavator: (a) excavator moving, (b) bucket strikes on the soil, (c) excavator digging and (d) detail of the excavation start (source: Vale S. A.)

1547

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Results The results of the damping tests and monitoring during excavation and embankment construction, allow the understanding of the behaviour of induced vibrations. The attenuation of induced vibrations in this type of study is achieved through potential regression models with the following formulation: PPV/Vpi = K1 x D -a1 Where: PPV/Vpi = Peak particle velocity, measured as the intensity of induced vibration at a point, in mm/s; D = Distance between the source/equipment and the measurement point, in meters; K1 e a1 = Constants defined based on the source/equipment of induced vibration and the characteristics of the medium through which they propagate.

Damping test of Scania dump truck The monitoring of the dump trucks was carried out in two operations: unloaded and loaded. Figure 5 shows the attenuation curves for different operating modes and on different days. It is noticeable that the highest vibrations in terms of PPV always occurred for the unloaded truck, and all values were very low. The recorded values have shown very low intensity when comparing with the bulldozer and hydraulic excavator. Table 2 presents only the maximum values from the two equations for Positions 01 and 02.

Figure 5: Attenuation curves for the dump truck for (a) Position 01 and (b) Position 02 Table 2: Maximum equations obtained under the condition of unloaded dump truck traffic for Positions 01 and 02 Position

Description

Equation (PPV/Vpi)

R²

01

Empty dump truck 16/11/22

2,4659 x D^-0,430

0,269

02

Empty dump truck 14/09/22

0,3758 x D^-0,213

0,628

1548

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS

Hyundai 220 LC and Komatsu 350 LC Hydraulic Excavators Figure 7 compares the equations of the highest intensity obtained for the three tested activities for positions 01 and 02 related to the use of the hydraulic excavator.

Figure 6: Attenuation curves for hydraulic excavator in different activities for (a) Position 01 and (b) Position 02 Upon analysing the graphs in Figure 6, it can be observed that the activity of excavating with the excavator, specifically in the first meter, indicated the generation of the most intense vibrations among the three equations. The excavator moving (first meter) showed lower intensities compared to the other two operations. It is also worth noting that there was no systematic relationship between higher intensities and depth for the different excavator tests. Table 3 presents the results of the three maximum equations mentioned in the above study for positions 1 and 2. Table 3: Equations obtained in the activity comparisons of the hydraulic excavator for Positions 01 and 02 Position

Location

Description

Equation (PPV/Vpi)

R²

01

Third Meter

Bucket Strikes – Third Meter

572,230 x D^-1,614

0,9891

01

First Meter

Excavation – First Meter

725,45 x D^-1,608

0,7946

01

First Meter

Moving – First Meter

18,60 x D^-0,804

0,9165

02

Second Meter

Excavation – Second Meter

72,228 x D^-1,030

0,6072

02

Third Meter

Moving – Third Meter

29,271 x D^-0,891

0,7899

02

Second Meter

Bucket Strikes – Second Meter

178,130 x D^-1,256

0,5659

It is noteworthy that the determination coefficients R² for Position 01 of the three equations ranged from good to excellent. However, for Position 02, the values of the determination coefficients R² ranged from fair to good.

1549

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Komatsu D61EX bulldozer Figure 7 shows a comparison of the equations with the highest intensity obtained in the two tested activities of the bulldozer.

Figure 7: Attenuation curves for bulldozer in different activities for (a) Position 01 and (b) Position 02 Table 4 presents the results of the two maximum equations obtained in the study, for positions 1 and 2. In this comparison, it was observed that the maximum equations for the bulldozer moving and operating with a blade are very close, with the exception of the equation in position 2 for the bulldozer moving.

1550

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS Table 4: Equations obtained in the activity comparisons of the bulldozer for Positions 01 and 02 Position

Location

Description

Equation (PPV/Vpi)

R2

01

Second meter

Bulldozer operating with the blade – second meter

22,145 x D^-0,640

0,845

01

First meter

Bulldozer moving – first meter

22,261 x D^-0,642

0,96

02

First meter

Bulldozer operating with the blade – first meter

25,084 x D^-0,890

0,96

02

Second meter

Bulldozer moving – second meter

36,869 x D^-0,849

0,87

Vibrations generated during excavation advancement Tables 5 and 6 provide a comprehensive summary of the highest vibrations observed during the “Excavation” activities, with a focus on the Surface/Depth relation. For Position 1, there was a considerable variation, with a maximum of 18.88, a minimum of 1.96, and an average of 5.17. As for Position 2, the maximum was 13.61, the minimum was 1.45, and the average was 3.45. Table 5: Comparisons of the highest vibration intensities Surface/Depth for pairs of seismographs and as a function of depth in the excavation phase for Position 1 Surface seismographs

Depth seismographs Point

Distance (m)

PPV/Vpi max (mm/s)

Surface / Depth

21.565

GFP-04

9

3.444

6.26

48

50.52

GFP-04

55

15.173

3.33

3

23.953

GFP-04

3

1.269

18.88

Point

Distance (m)

PPV/Vpi max (mm/s)

GFS-06 / GFP-04

GFS-06

7

Second Meter

GFS-06 / GFP-04

GFS-06

Third Meter

GFS-06 / GFP-04

GFS-06

Phase

Relation

First meter

Table 6: Comparisons of the highest vibration intensities Surface/Depth for pairs of seismographs and as a function of depth in the Excavation Phase for Position 2 Surface seismographs

Depth seismographs Point

Distance (m)

PPV/Vpi max (mm/s)

Surface / Depth

7.346

GFP-04

21

2.183

3.37

30

12.611

GFP-04

23

1.734

6.79

GFS-03

3

13.305

GFP-02

8

2.168

6.14

GFS-06

7

49.443

GFP-04

4

3.634

13.61

Point

Distance (m)

PPV/Vpi max (mm/s)

GFS-06 / GFP-04

GFS-06

23

Second Meter

GFS-06 / GFP-04

GFS-06

Third Meter

GFS-03 / GFP-02

Fourth Meter

GFS-06 / GFP-04

Phase

Relation

First meter

1551

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Vibrations generated during the construction of the embankment Tables 7 and 8 present a general summary of the highest vibrations during the “Construction of the Conquest Embankment” activities, relating Surface/Depth. For Position 1, a variation was recorded during the period with a maximum of 10.35, a minimum of 1.38, and an average of 2.95. It can be observed that the maximum value of this relation is lower than the one recorded during the excavation phases. Regarding Position 2, a maximum of 17.59, a minimum of 1.07, and an average of 2.86 were obtained. Table 7: Comparisons of the highest vibration intensities Surface/Depth for pairs of seismographs and as a function of depth in the Construction Phase of the Conquista Embankment for Position 1. Surface seismographs

Depth seismographs Point

Distance (m)

PPV/Vpi max (mm/s)

Surface / Depth

7.771

GFP-04

4

1.308

5.94

40

3.728

GFP-04

3

1.269

2.94

22

12.65

GFP-05

22

1.222

10.35

Point

Distance (m)

PPV/Vpi max (mm/s)

GFS-06 / GFP-04

GFS-06

6

Second meter

GFS-06 / GFP-04

GFS-06

Third meter

GFS-07 / GFP-05

GFS-07

Phase

Relation

First meter

Table 8: Comparisons of the highest vibration intensities Surface/Depth for pairs of seismographs and as a function of depth in the Construction Phase of the Conquista Embankment for Position 2 Surface seismographs Phase

Relation

Depth seismographs

Point

Distance (m)

PPV/Vpi max (mm/s)

Point

Distance (m)

PPV/Vpi max (mm/s)

Surface / Depth

First meter

GFS-06 / GFP-04

GFS-06

23

2.238

GFP-04

22

1.048

2.14

Second Meter

GFS-06 / GFP-04

GFS-06

10

3.239

GFP-04

38

1.324

2.45

Third Meter

GFS-03 / GFP-02

GFS-03

28

2.010

GFP-02

27

1,151

1,75

Fourth Meter

GFS-06 / GFP-04

GFS-06

7

22.321

GFP-04

5

1,269

17,59

1552

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS

Discussion From the analyses conducted for Positions 01 and 02, it was observed that: • The vibrations generated by the dump trucks were of very low intensity when comparing with the bulldozer and hydraulic excavator; • The depth vibrations were significantly lower than the surface vibrations: based on the results obtained in the “Excavations,” an attenuation of approximately 1/3 can be expected, while for the “Embankment Construction” an attenuation of approximately 1/2 can be expected for depth vibrations. • The set of field tests allowed for recording and identifying relationships between vibration levels and distances from sources in various studied cases, both for each instrumentation stage and for each equipment and specific activity related to some of them. • No attenuation or amplification of vibrations was identified as the excavations progressed “meter by meter” up to the third meter of excavation. •

In the monitoring of services at position 1, the highest recorded value on the surface seismographs was PPV/Vpi = 143.52 mm/s during a “damping test” of the Komatsu excavator at the third meter from bucket strikes, and the highest recorded value on the depth seismographs was PPV/Vpi = 15.173 mm/s during the “Excavation” of the second meter.

• In the monitoring of services at position 2, the highest recorded value on the surface seismographs was PPV/Vpi = 62.01 mm/s during a “damping test” of the Komatsu Excavator at the second meter, and the highest recorded value on the depth seismographs was PPV/Vpi = 5.801 mm/s during the “Excavation” of the second meter. • For the “Excavations” at position 01, the highest recorded value was PPV/Vpi = 50.52 mm/s on sensor GFS-06 during the excavation of the second meter, and for the “Embankment Construction,” the highest recorded value was PPV/Vpi = 12.65 mm/s, also on GFS-07 at the Third Meter. Nevertheless, no amplification or attenuation of vibrations with depth was observed. The same occurred in the “Equipment Vibration Damping Tests” with distance. • For the “Excavations” at position 02, the highest recorded value was PPV/Vpi = 49.443 mm/s on sensor GFS-06 during the excavation of the fourth meter, and for the “Embankment Construction,” the highest recorded value was PPV/Vpi = 22.321 mm/s, also on the same sensor and depth. Nevertheless, no amplification or attenuation of vibrations with depth was observed. The same occurred in the “Equipment Vibration Damping Tests” with distance. • Table 9 presents a comparison of the highest values obtained on the surface seismographs in the two areas, Position 1 and Position 2. In this comparison, it is observed that the maximum vibrations

1553

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA at Position 1 were higher than the equivalent values at Position 2 in three out of five comparisons/activities, with a maximum 3.15 times higher for the damping tests of the dump Trucks, which, however, generated low-intensity vibration levels. Table 9: Overall summary and comparisons of vibration intensities of surface seismographs at Positions 1 and 2

Position

1

Activity

Damping test

Equipment

Operation

Excavation depth

PPV/Vpi (mm/s)

Relation P1/P2

Dump truck

Unloaded displacement – GFS-02

Surface

1,466

3,15

Bulldozer

Operating with the blade – GFS-01

Third meter

8,276

0,60

Hydraulic excavator

Komatsu – Bucket strikes – GFS-01

Third meter

143,52

2,31

1

Excavation

–

GFS-06

Second meter

50,52

1,02

1

Construction of embankment

–

GFS-07

Third meter

12,65

0,57

Dump truck

Unloaded displacement – GFS-01

First meter

0,465

–

Bulldozer

Displacement CAT D6 – GFS-01

First meter

13,848

–

Hydraulic excavator

Komatsu - Bucket strikes – GFS-01

Second meter

62,101

–

2

Damping test

2

Excavation

–

GFS-06

Fourth meter

49,443

–

2

Construction of embankment

–

GFS-06

Fourth meter

22,321

–

•

Table 10 provides a comparison of the highest analogous values obtained from the depth seismographs in the two areas/campaigns, position 1 and 2. In this comparison of depth records, in

1554

ANALYSIS OF VIBRATIONAL WAVES INDUCED BY MACHINERY IN TAILINGS DAMS four occasions/activities, the maximum records from position 1 were higher in four out of five comparisons, with a maximum of 2.62 times for the excavation activities. Table 10: Overall summary and comparisons of vibration intensities of depth seismographs at Positions 1 and 2 Position

1

Activity

Damping test

Equipment

Operation

Excavation depth

PPV/Vpi (mm/s)

Relation P1/P2

Dump truck

Unloaded displacement – GFP-04

Surface

0,441

2,44

Bulldozer

Displacement – GFP-04

Surface

2,388

1,37

Hydraulic excavator

Hyundai – Bucket strikes – GFP-04

Surface

5,312

1,15

1

Excavation

–

GFP-04

Second meter

15,173

2,62

1

Construction of embankment

–

GFP-04

Third meter

1,726

0,89

Dump truck

Unloaded displacement – GFP-04

Surface

0,181

–

Bulldozer

CAT D6 – Operating with the blade – GFP-04

Third meter

1,742

–

Hydraulic excavator

Hyundai – Bucket strikes – GFP-04

Surface

4,635

–

2

Damping test

2

Excavation

–

GFP-04

Second meter

5,801

–

2

Construction of embankment

–

GFP-04

Third meter

1,947

–

•

The maximum values from position 1, both for surface and depth measurements, were significantly higher than those from position 2. This could be attributed to water, partly due to differences in piezometric levels and partly due to intense rainfall during the testing period at position 1.

Conclusion This study aimed to analyze vibrations induced by different mechanical sources during the construction of the “Experimental Excavation,” Field services were carried out using engineering seismographs, primarily involving vibration monitoring and equipment vibration attenuation testing. The presented studies are essential for planning future activities involving the use of these equipment/sources for the decharacterization works of this structure.

1555

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA The results showed that the vibrations generated by the dump trucks were of very low intensity compared to the other equipment. Depth vibrations were significantly lower than surface vibrations, with an expected attenuation of approximately 1/3 for excavations and 1/2 for embankment construction. Overall, this study provides valuable insights into the behaviour and attenuation of induced vibrations during excavation and construction activities. The findings contribute to understanding the potential impact of such activities on the surrounding environment, and can guide future engineering practices to mitigate vibration-related issues. However, it should be emphasized that the interpretations and results of this study are valid only for these induced vibration tests produced by mobile equipment/sources and the specific instrumentation carried out during the predefined activities/operations in the execution of the “Experimental Excavation” under the existing conditions at the site.

References Dowding, C.H. 1985. Blast Vibration Monitoring and Control. Englewood Cliffs: Prentice Hall Inc. Nateghi, R. 2011. Prediction of ground vibration level induced by blasting at different rock units. International Journal of Rock Mechanics and Mining Sciences 48(6): 899–908. Petley, D. and Monticelli, J.J. 2020. Brumadinho: relatório do Expert Painel sobre o rompimento da barragem de rejeitos de Feijão. Terrae Didatica, 16, e020008. Accessed 01/07/2023 at: https://periodicos.sbu.unicamp.br/ojs/index.php/td/article/view/8659109/22252 Sousa, G.M., Ferreira, S.A. and Gomes, R.C. 2021. Methodology for automated monitoring of induced vibrations in tailings dams built upstream. Geotechnical and Geological Engineering 1–10. DOI:10.21203/rs.3.rs162752/v1. Zhang, J., Zhang, J., Tian, H., Guo, W. and Huo, W. 2022. Liquefaction risk assessment of tailings dams considering seismic and dynamic properties of tailings. Journal of Earthquake Engineering 26(1): 64–86.

1556

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Evaluation of Liquefied Strength of Uncompacted Tailings Sand Using Cone Penetration Test at an Oil Sands Tailings Facility Ying Zhang, WSP, Canada Ayman H. Abusaid, WSP, Canada Gordon W. Pollock, WSP, Canada Yvonne (Yirao) Qiu, WSP, Canada Ryan Moore, Suncor Energy Inc., Canada Jason Rhee, Suncor Energy Inc., Canada

Abstract Tailings sand placed hydraulically and uncompacted in a beach above water (BAW) setting can be potentially liquefiable when the structure is raised at a high rate. Experience has shown that flow liquefaction often takes place with small trigger events and no warning and can have serious safety, environmental, and infrastructure consequences. Although dozer compaction is commonly used to construct a non-liquefiable shell, flow liquefaction of the uncompacted tailings sand upstream of the shell remains a key design element. An upstream tailings sand structure (Sand Dump) at Suncor has been constructed with 5 m lifts at a rate between 10 m and 20 m per year since 2012. The final height of the sand structure is approximately 130 m and it is expected to be completed by 2024. The Cone Penetration Test (CPT) is commonly used as a field investigation tool to evaluate the potential of flow liquefaction and has been used extensively at the Sand Dump for this purpose for the compacted and uncompacted BAW zones. This paper focuses on the liquefaction assessment for the uncompacted tailings sand. CPTs have been carried out across the project site at the same locations in successive years to evaluate the impact of subsequent sand loading on the liquefaction potential of the uncompacted tailings sand. This paper provides the estimated liquefied strength of the uncompacted tailings sand and its correlation with the distance from the discharge location. The findings are used to verify the initial design assumptions and could provide valuable information for the design of similar future structures.

1557

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction The Sand Dump is a tailings sand structure at Suncor’s Millennium mine site, north of Fort McMurray, Alberta, Canada, as shown in Figure 1. An overview of the Sand Dump design was provided by Pollock et al. (2014). The sand structure has been constructed with 5 m lifts at a rate between 10 m and 20 m per year since 2012 and it is expected to reach its final height of approximately 130 m in 2024. Tailings sand placed hydraulically in a BAW setting can be potentially liquefiable when the structure is raised at a high rate. Continuous dozer compaction has been implemented during tailings placement to form a nonliquefiable shell as part of the upstream tailings dam. The assessment and management of the liquefaction potential of compacted tailings sand at the Sand Dump were discussed by Zhang et al. (2020).

Figure 1: Location and plan view of the Sand Dump at Suncor

The tailings sand between the compacted shell and the supernatant tailings pond, as shown in Figure 2, is uncompacted and assumed potentially liquefiable in the design. The uncompacted tailings sand, if liquefied, forms a significant portion of the failure mass, as indicated in Figure 3, and therefore is an important design element of the structure. This paper focusses on the liquefaction assessment for the uncompacted BAW zone using CPT data. The impact of subsequent sand loading on the liquefaction potential of the uncompacted tailings sand is also discussed through evaluating CPT data from successive years at the same locations. An estimation of the post liquefied strength of the tailings sand based on published CPT correlations is discussed, as well as the correlation with the distance from the tailings discharge location.

1558

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY

Figure 2: Satellite image of the Sand Dump in June 2023

Figure 3: Typical profile of the east containment dam

1559

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Tailings placement and tailings sand properties At Suncor, whole tailings are hydraulically transported from the extraction plants through pipelines to the perimeter of the Sand Dump. The compacted zone is constructed by spigotting tailings into a cell that is surrounded by several dry dykes, with one end of the cell open to facilitate flow towards the pond. Occasionally, due to operational constraints, the end of cell has a dry dyke with a spill pipe installed to facilitate drainage out of the cell. Upon deposition, the sand particles settle down near the discharge location to form the sand beach. Some fines are trapped in the beach while the rest flows to the fluid pond. There are also occasions of discharging tailings with high fines content directly into the uncompacted beach, which could create high fines layers within the tailings beach. Continuous dozer compaction has been implemented in the cells, and tailings outside the cells are left uncompacted. The fluid fine tailings and water in the fluid pond are pumped to another tailings facility to keep a minimal pond size at the Sand Dump. Tailings sand at the Sand Dump is fine-grained sand with a trace (< 5%) of medium-grained sand and a trace of silt and clay. The fines content (particle size < 45 µm) is typically in the range of 3% to 5% in the compacted zone and 3% to 7% in the uncompacted sand zone (Abusaid et al, 2020).

Liquefaction susceptibility assessment method Zhang et al. (2020) provided an overview of flow liquefaction susceptibility analysis methods, among which the Qtn,cs method by Robertson (2010) was used as the primary approach at the Sand Dump. The soil at a specific depth is considered to be susceptible to liquefaction if Qtn,cs ≤ 70. Robertson (2016) indicated that using Qtn,cs = 70 is slightly conservative but it is considered as an appropriate screening tool for tailings structures with very high consequence. Robertson (2022) updated the fines content correction coefficient (Kc), mainly for clay-like soils. The Robertson (2022) update considers the soils as contractive when Qtn,cs ≤ 70, dilative when Qtn,cs ≥ 80, and partially contractive for soils with Qtn,cs between 70 and 80. Although the Robertson method is considered to be the state of the art for interpreting CPT results with respect to liquefaction, one uncertainty in using Qtn,cs in this regard is that it is calculated using vertical stress even though lateral effective stress strongly influences the cone resistance (Houlsby and Hitchman, 1988, 1989). It is recognized that it is challenging to determine the in-situ lateral stress at each CPT location.

Liquefaction potential of uncompacted tailings sand at the sand dump Thirteen CPT programs with about 230 CPTs at 120 locations in the uncompacted tailings sand at the Sand Dump were conducted between 2012 and 2022. The CPT data from the locations that might be influenced by dozer compaction were excluded from the database used in this analysis. Figure 2 shows 20

1560

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY locations (black circle symbols) with only one CPT conducted at those locations and 25 locations (blue square symbols) with multiple CPTs conducted at those locations. The Qtn,cs

method by

Robertson (2010, 2022) was used to assess the liquefaction potential of the uncompacted tailings sand. Figure 4 shows the distribution of the Qtn,cs values from the CPT testing in the uncompacted tailings sand at the Sand Dump. About 45% of Qtn,cs values are lower than 70, indicating the sand is contractive and could experience significant strength loss at large strain. Approximately 40% of the Qtn,cs values are higher than 80, indicating the sand is dilative and not liquefiable. Slightly over 15% of the Qtn,cs values are between 70 and 80, indicating the sand is contractive but not expected to have significant strength loss during undrained shear.

Figure 4: Distribution of Qtn,cs values from CPTs in the uncompacted tailings sand

Impact of subsequent loading on the liquefaction potential Hydraulically placed tailings sand is expected to consolidate and become denser upon the loading by subsequent sand placement especially for loose uncompacted sand. However, it is unclear if the increase in density can change the sand state from being contractive to dilative. A review of the Qtn,cs values from CPTs tested at the same locations in successive programs at the Sand Dump was carried out to examine the improvement in the sand state.

1561

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA There are 20 locations along the east containment dam with CPTs from multiple programs, as shown in Figure 2. The thickness of the sand that was placed between two CPT tests at the same location ranged from 8 m to 40 m. Figure 5 presents examples of the Qtn,cs profiles from different programs at two CPT locations. The distance between the as-built locations of the CPTs from the different programs are less than 5 m. The thickness of the tailings sand placed between adjacent CPT programs (about 20 m for Location A) are labelled on the left side of Figure 5. The water tables at the time of CPT testing are also provided in the figure and are typically at or very close to the ground surface. A review of all the Qtn,cs comparison plots indicated that subsequent sand loading at the Sand Dump can improve the state of tailings sand when Qtn,cs is lower than 50. Subsequent loading can sometimes improve the state of tailings sand when Qtn,cs is up to 60, but there are cases with Qtn,cs between 50 and 60 and no improvement is observed. In general, subsequent sand loading is expected to increase the Qtn,cs for uncompacted tailings sand to at least 50 and sometimes up to 60, and occasionally to 70 and 80. To further evaluate the state of the uncompacted tailings sand under loading, the Qtn,cs values from all 45 CPT locations in Figure 2 are plotted against the vertical effective stress at the time of the CPT testing in Figure 6. Since the water table was very close to the ground surface at the time of the CPT testing, the vertical effective stress corresponds to the depth below ground surface. For example, the vertical effective stress at a depth of 10 m is about 100 kPa. Along with the individual CPT data, the 10th, 20th, 30th percentile and the average Qtn,cs values are determined for an interval of 10 kPa when the vertical effective stress is less than 200 kPa and an interval of 50 kPa when vertical effective stress is higher than 200 kPa, as shown by the dashed lines. Solid lines are then used to represent the trends of the dash lines. Figure 6 shows that Qtn,cs is low near the ground surface and increases significantly with vertical effective stress up to approximately 100 kPa after which there is little to no increase. Although the Qtn,cs values showed a general trend of increase with vertical effective stress (depth), the CPT database at the Sand Dump showed that Qtn,cs can be lower than 70 up to a vertical effective stress of 600 kPa (or a depth of 60 m) for the uncompacted tailings sand, which is consistent with the design assumption for the uncompacted tailings sand of this structure. The state of the uncompacted tailings sand is unclear at greater depth since CPT typically encounters refusal before reaching a depth of 60 m in the uncompacted zone at the Sand Dump.

1562

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY

a) CPT data from Aug 2012, Jan 2014, and Jul 2015 programs at Location A

b) CPT data from Aug 2013, Jan 2014, Jul 2015, and Jun 2020 programs at Location B

Figure 5: Impact of subsequent sand loading on the Qtn,cs values for uncompacted tailings sand

1563

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 6: Correlations between the Qtn,cs values and the vertical effective stress

Impact of distance from the tailings discharge location The impact of distance from the tailings discharge location on the state of the uncompacted sand was evaluated for the Sand Dump. Figure 7 shows the typical profile of the east containment dam with the CPT locations plotted in profile. The tailings discharge location was assumed to be the upstream edge of the compacted zone for this study; however the actual discharge location varies throughout the compacted zone for this particular structure. The data was also subdivided into zones above El. 320 m and below El. 320 m to take into account the impact of depth. Given the large amount of CPT data, the 10th, 20th, and average Qtn,cs values were determined for each 100 m distance zone for comparison with the representative Qtn,cs values from other zones. Figure 8 shows some examples of the comparison of the 10th and 20th percentile Qtn,cs values below El. 320 m. A review of all the data comparison revealed a general trend of decrease in Qtn,cs values further away from the discharge location for both data sets, above El. 320 m and below El. 320 m, which could be attributed by the decreased deposition energy and change in tailings composition.

1564

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY

Liquefied shear strength ratio (Su(liq)/σv0’) Significant strength loss (strain softening) is expected when loose sand liquefies. Flow liquefaction often takes place with no warning and can result in significant loss of life, environment, and infrastructure. Therefore, the Sand Dump design assumed the uncompacted tailings sand is potentially liquefiable. Based on case histories, the designers’ experience, and a field trial, a liquefied shear strength ratio (Su(liq)/σv0’) of 0.05 was adopted in the design for the case with flow liquefaction. Su(liq) is the undrained shear strength at large strain which corresponds to the liquefied shear strength for loose sand. σv0’ is the in-situ vertical effective stress prior to the liquefaction triggering event. Researchers (Olson and Stark, 2003; Robertson, 2010; Jefferies and Been, 2016; and Robertson, 2022) have proposed various correlations to estimate the liquefied shear strength ratio using different corrected CPT tip resistances. The Robertson (2022) proposed Su(liq)/σv0’ vs Qtn,cs relationship, as shown in Figure 9 and discussed below, was used in the current study: •

The proposed correlation covers sand-like soil (Ic ≤ 3) and clay-like soil (Ic > 3) and it involves different equations to estimate the undrained shear strength of both types of soils. Ic is a soil behaviour type index used to categorize soils based on CPT data (Robertson and Cabal, 2022).

•

The correlation is based on a database incorporating recent experience and case histories. The Class A and B case history data for both sand- and clay-like soils are used in establishing the correlation. The case histories only cover vertical effective stress less than 300 kPa, and most cases are less than 200 kPa. Robertson (2022) stated that the estimated undrained shear strength ratio could be conservatively low for soils under higher stresses and advanced laboratory testing is required to evaluate the curvature of the critical state line at those higher stresses so a higher undrained shear strength ratio can be used.

•

The correlation uses the lower bound of the estimated undrained shear strength ratios for loose and very loose soils (35 < Qtn,cs < 60) from the Class A and B case histories. The undrained shear strength ratios were estimated using the mean CPT values in the zones believed to be involved in the failure. For dense to very dense soils (Qtn,cs > 80), the correlation uses the drained shear strength. For soils without case histories (Qtn,cs less than 35 or between 60 and 80), the correlation uses interpolation and extrapolation. There are some sources of uncertainty in the back analysis of the case histories used in establishing

the Robertson (2022) Su(liq)/σv0’ vs Qtn,cs relationship. For example, Robertson (2010) notes that the back analyses may not represent the actual failure mechanism including retrogressive failure and inertia effects, and the CPT dataset used may not be entirely within the failure mass.

1565

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 7: Distance categories for the east containment dam

a) 10th percentile Qtn,cs values

b) 20th percentile Qtn,cs values

Figure 8: Comparison of some representative Qtn,cs values for tailings sand with different distances to the discharge location

1566

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY

Figure 9: Proposed relationship between liquefied strength ratio and Qtn,cs (Robertson, 2022) Figure 10 presents the estimated undrained shear strength ratio using the Robertson (2022) method for the Qtn,cs values for the uncompacted sand at the Sand Dump. Jefferies and Been (2016) suggested design values close to 10th percentile for cyclic loading cases and towards 20th percentile for static design problems, while recommending the design engineers to consider the influence of construction method and the scale of variation. Robertson (2022) indicated that 30th percentile (approximately mean minus one standard deviation) of the CPT values can be used to represent the weaker zones that control stability according to case histories. For the Sand Dump or similar structures, one undrained shear strength ratio could be selected for the weak plane (sliding along a potential loose layer) and a different one for the backscarp (to reflect cross bedding) of the failure surfaces. Also, the liquefied shear strength ratio is expected to be higher near the discharge location and lower further away from the discharge location, based on the assessment of the impact of the distance from the discharge location. The undrained shear strength ratios estimated using all data from the uncompacted zone should be conservative since the critical slip surfaces are typically close, at least for the Sand Dump, to the compaction boundary for the liquefaction scenario. Therefore, the analysis of the data and the preceding commentary indicate that the design value of Su(liq)/σv0’ = 0.05 is conservative based on the Robertson (2022) relationship.

1567

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Figure 10: Estimated undrained shear strength ratio using Robertson (2022) method

Summary and conclusions The uncompacted tailings sand beach at the Suncor Sand Dump was designed to be potentially liquefiable, and a post-liquefaction shear strength ratio of 0.05 was used in the design. The sand structure has been under construction since 2012 and reached a height of approximately 120 m as of June 2023. Multiple CPT programs were carried out over the course of the Sand Dump construction. Results of CPTs pushed in the uncompacted tailings sand were evaluated using the Robertson (2010, 2022) method and the following was concluded: •

A significant portion (~45%) of the uncompacted tailings sand is potentially liquefiable, confirming the original design assumption.

•

The Qtn,cs values versus effective stress profile indicated that the uncompacted tailings sand is very loose near the ground surface, and the Qtn,cs increases significantly up to a stress level of 100 kPa, after which the increase in Qtn,cs becomes less prominent.

•

Subsequent sand loading can improve the state of the uncompacted tailings sand when Qtn,cs is lower than 50. Subsequent loading can sometimes improve the state of tailings sand when Qtn,cs is up to 60. In general, subsequent sand loading is expected to increase Qtn,cs at uncompacted tailings sand to at least 50 and occasionally up to 60.

1568

EVALUATION OF LIQUEFIED STRENGTH OF UNCOMPACTED TAILINGS SAND USING CONE PENETRATION TEST AT AN OIL SANDS TAILINGS FACILITY •

The Qtn,cs values generally decrease further away from the tailings discharge location, and this is also expected to be true for the liquefied shear strength ratio.

•

The liquefied shear strength ratio of the uncompacted tailings sand at the Sand Dump was estimated using the Robertson (2022) method. The majority of the estimated liquefied shear strength ratios are higher than the design value of 0.05, especially if a cross-bedding liquefied shear strength is used as part of the design. For other tailings sand structures, the liquefied shear strength ratio could be impacted by the tailings sand properties and the tailings placement methods and rate, which should be considered when selecting the design liquefied strength ratio.

Acknowledgment The authors would like to recognize Dr. E.C. McRoberts of WSP for his technical guidance throughout the project. Disclaimer Suncor Energy Inc. and its affiliates (collectively "Suncor") do not make any express or implied representations or warranties as to the accuracy, timeliness or completeness of the statements, information, data and content contained in this paper and any materials or information (written or otherwise) provided in conjunction with this paper (collectively, the "Information"). The Information has been prepared solely for informational purposes only and should not be relied upon. Suncor is not responsible for and is hereby released from any liabilities whatsoever for any errors or omissions in the Information and/or arising out of a person’s use of, or reliance on, the Information. Suncor does not endorse any of the companies, organizations, products and/or services mentioned or described in this paper.

References Abusaid, A.H., Zhang, Y., Pollock, G.W., Moore, R. and Rhee, J. 2020. CPT dynamic pore water pressure and liquefaction potential in tailings sand at Suncor. In Proceedings of Tailings and Mine Waste 2020. Houlsby, G.T. and Hitchman, R. 1988. Calibration chamber tests of a cone penetrometer in sand. Geotechnique 38(1): 39–44. Houlsby, G.T. and Hitchman, R. 1989. Calibration chamber tests of a cone penetrometer in sand: discussion. Geotechnique 39(4): 727–731. Jefferies, M.G. and Been, K. 2016. Soil Liquefaction: A Critical State Approach. 2nd edition. CRC Press. Olson, S.M., and Stark, T.D. 2003. Yield strength ratio and liquefaction analysis of slopes and embankments. Journal of Geotechnical and Geoenvironmental Engineering 129(8): 727–737.

1569

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Pollock, G.W., Mettananda, D.C.A. and MacGowan, T. 2014. Design of Suncor’s first tailings sand structure under tailings reduction operations – sand dump 8. In Proceedings of Tailings and Mine Waste 2014. Keystone, CO, USA. 449–462. Robertson, P.K. 2010. Evaluation of flow liquefaction and liquefied strength using the cone penetration test. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 136(6): 842–853. Robertson, P.K. 2016. Cone penetration test (CPT)-based soil behaviour type (SBT) classification system – an update. Canadian Geotechnical Journal 53(12): 1910–1927. Robertson, P.K. 2022. Evaluation of flow liquefaction and liquefied strength using the cone penetration test: an update. Canadian Geotechnical Journal 59(8): 620–624. Robertson, P.K. and Cabal, K.L. 2022. Guide to Cone Penetration Testing for Geotechnical Engineering, 7th edition. Zhang, Y., Abusaid, A.H., Pollock, G.W., Moore, R. and Rhee, J. 2020. Managing the liquefaction potential of compacted tailings sand at Suncor. In Proceedings of Tailings and Mine Waste 2020.

1570

Chapter Thirteen

Supportive Technologies

1571

1572

Proceedings of Tailings and Mine Waste 2023 November 5–8, 2023, Vancouver, Canada

Hydraulic Dewatered Stacking – Developing Strategies for Brownfield Applications at Mogalakwena, South Africa Murray McGregor, SRK Consulting, UK Phil Newman, American Technical and Sustainability Services (UK) Ltd, UK Andrea López, Anglo American Technical and Sustainability Services, Chile

Abstract Anglo American has patented a new technology system for tailings deposition called hydraulic dewatered stacking (HDS). The premise of HDS is to hydraulically co-dispose fines-free sand with tailings to improve the overall drainage of the facility and deliver an unsaturated tailings facility. The objective of this novel approach is to improve water recovery, speed up reclamation, and enhance the safety of the storage facility (by reducing pore pressures and supernatant water). As part of the HDS technology development pathway, Anglo American is currently implementing two industrial-scale pilots to validate the science behind the HDS concept, improve the knowledge of the expected benefits, and generate learning and experience that will prove valuable in full-scale operation. This paper presents an HDS demonstration at the Mogalakwena mine in South Africa, referencing the successful greenfield trial at El Soldado, Chile, and speaks to the unique constraints at Mogalakwena, where the sand availability is reduced. The field trial at the Mogalakwena mine uniquely focuses on the application of HDS within a largescale active tailings facility. The primary objective is to determine design criteria for a proposed full-scale HDS implementation at Blinkwater 2, a tailings expansion project. A trial sand channel was constructed in 2022, and lessons learned were used to develop the trial that began in quarter 2 of 2023. Since the trial is constructed on top of an active tailings facility, the sand channels will be dewatered, with excess water reporting to the central tailings pond. This will allow the surrounding tailings to drain and consolidate as though the sand channels were connected to a drainage network. The final design and initial results for the trial will be presented with preliminary field observations. Extensive instrumentation has been proposed within the footprint of the Mogalakwena trial area, with additional instrumentation outside the footprint so that a comparison can be made between the augmented drainage from HDS and traditional slurry deposition that is ongoing adjacent to the trial area.

1573

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Introduction Major failures of tailings storage facilities continue to occur within the mining industry and have led to significant economic, environmental, and human losses. It is important to note that these failures are not restricted to regions with poor governance, but continue to happen across all jurisdictions around the globe. Scrutiny is increasing within the mining industry, and safe tailings management is becoming increasingly important to maintain social license to operate. There is clearly a need for continued innovation to improve practice across the globe leading to tailings safety, such as the hydraulic dewatered stacking (HDS) method described herein. This paper describes the HDS method and highlights some of the recent field-scale work undertaken or planned at the Mogalakwena platinum project in South Africa, on top of the current Blinkwater 1 (BW1) tailings facility.

Hydraulic dewatered stacking The HDS method uses engineered co-disposal of fines-free sand (i.e., free draining) and tailings placed in discrete features to promote internal drainage within the tailings facility. The goal is to have the fines-free sand connect with an internal drainage system that allows pore water to rapidly drain under gravity to a collection point, where it can be extracted from the TSF, rather than remaining as a supernatant pond or porewater within the facility. The HDS method was developed from Anglo American’s experience with coarse particle recovery (CPR). A new approach to flotation that has previously been widely used in some minerals is now starting to gain traction in base metals hydrometallurgical circuits (Filmer and Alexander, 2016 and 2017; Filmer et al., 2020). Through early rejection of barren sand, additional production capacity in the downstream process plant is made available, delivering value opportunities (Arburo et al., 2022). It should be noted that the HDS method is not restricted to projects that are producing CPR sand and can be applied to assets that operate with a medium to coarse grind (i.e., D80 of ~150µu). The main potential advantages of the HDS method are listed below: • Improved drainage and consolidation of tailings (quicker access to tailings beach, faster reclamation during closure phase). • Improved safety of the TSF; reduced pore water within the tailings and external embankments (increased stability). • Reduced potential for strength loss through higher consolidation and the delivery of unsaturated tailings facility (enhanced liquefaction resistance). • Reduced risk of piping (due to a lower hydraulic gradient within the dam).

1574

HYDRAULIC DEWATERED STACKING – DEVELOPING STRATEGIES FOR BROWNFIELD APPLICATIONS AT MOGALAKWENA, SOUTH AFRICA • Improved water recovery (higher quantity, with likely lower suspended solids). • Reduced cost relative to filtered dewatering technologies for similar project conditions. A proof of concept study was undertaken to assess the potential for CPR sand to provide effective drainage for tailings facilities, using the HDS method. This study indicated that further evaluation was needed to test the effectiveness of the HDS method. Laboratory testing and numerical assessment from the proof of concept study are summarized in Newman et al. (2022). The first large-scale trial is underway at El Soldado mine, Chile (Newman et al., 2022) and a second trial is assessing HDS within an active tailings facility at the Mogalakwena mine, South Africa – the subject of this paper. A preliminary sand placement trial was completed in 2022 at the BW1 facility and several lessons were learned, which were then applied to the El Soldado trial. For the 2023 full-scale demonstration, the experience from El Soldado (commissioned in quarter 3 of 2022) enabled rapid commissioning and fast progress on sand channel construction at BW1. The design, instrumentation design, and early stages of channel placement are described in this paper – building on, and contributing to, the rapidly growing experience of this new approach to tailings management.

Mogalakwena BW1 trial Trial objectives The Mogalakwena demonstration has two primary objectives: to demonstrate the feasibility of HDS implementation in an operating TSF; and to determine the zone of influence from the drainage channels installed – a key aspect of the HDS technology The outcomes of the trial will be incorporated into the design of the Blinkwater 2 tailings facility, as well as other platinum tailings operations.

Operational objectives • To determine whether controlled placement of fines-free sand can be achieved using the proposed deposition unit • To determine whether the supply pipeline can be effectively managed, so that the deposition unit can reach all of the trial area. • To determine whether deposition can be completed with limited disruption to operations (i.e., can co-disposal of sand and thickened tailings be achieved)? • To determine how fines-free sand can be deposited at significant distances from the external embankment. • To determine whether the deposition unit can safely traffic across the tailings surface.

1575

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

Technical objectives • To discover whether the deposited sand can effectively drain the surrounding tailings mass. • To determine whether this method can deliver improved water return (rapid drainage with minimal suspended solids). • To determine what level of accelerated consolidation and subsequent strength development will occur in zones where fines-free sand is co-disposed with thickened tailings. • To determine whether the deposited drainage channels will be resistant to clogging. • To determine what spacing will adequately achieve the above objectives – i.e., what is the zone of influence? The zone of influence of a drainage channel within the tailings mass is a key design parameter when assessing implementation options, particularly in platinum, where the sand available is likely to be limited.

BW1 constraints The HDS trial at the Mogalakwena mine has been designed based on a HydrofloatTM plant upgrade that will produce a fines-free CPR sand as a reject. The current plant redesign indicates that the CPR sand will make up between 13% to 18% of the 1,300 tph of material from the plant (approximate ratio of between 1:5 and 1:6 CPR sand to tailings). This is unlike CPR sand from copper tailings, where a ratio of up to 1:3 CPR sand to tailings can be available. The lack of sand availability is a key constraint when designing a trial to prove the HDS method for platinum tailings. Process plant upgrades for the Mogalakwena site are underway at the time of writing, and full-scale production of CPR sand is not yet available. Generating enough CPR sand for the proposed HDS trial was not possible in the short term; therefore, an imported sand will be used for first stages of the BW1 trial. The expected CPR sand is poorly graded fines-free sand, while the imported sand is more well graded, with up to 15% fines (90% of solids in effluent 15 mm/yr and 200,000 tonnes per day) tailings production. For tailings stored in legacy, inactive, or operating TSFs, the issue of dewatering cannot be addressed by filtration systems. The application of large-scale in-situ mass dewatering of tailings by electrokinetics is a method that must be demonstrated in the near future. With tailings dewatered, there is a much greater likelihood of successful closure with the creation of a stable geologic landform that supports community-desired land use. Note, however, that a long-term concern for chemical stability is not addressed by dewatering, an observation that often pits the geotechnical “best” against the geochemical “best” design. In-pit disposal is a possibility that several mines are considering or implementing, as is the use of backfill, which can be placed in underground mines to increase the stability of underground openings.

Remining and reprocessing The topic of metal recovery from mine tailings is of growing interest for a range of reasons, including removal of environmental and safety liabilities, waste reduction and circularity, and recovery of additional resources such as critical minerals. ICMM (2023) highlights tailings reprocessing as one solution to reduce the potential for disasters from current and legacy mine wastes. The impending transition to net-zero carbon energy systems will require a major influx of minerals and metals, some of which were not previously exploited on a large scale (Sovacool et al., 2020). Reprocessing tailings is one potential avenue to recover needed resources in a socially and environmentally responsible manner. Remining and reprocessing can produce value, reduce the waste volume to be disposed and provide social benefits and sustainable industry for the local workforce, thus reducing environmental impacts. The most common methods of remining have been hydraulic or monitor mining (e.g., Anglo-Gold Ashanti (2011, 2016), BlueGold, Kinross Gold (2015), and Sibanye-Stillwater Ltd (2021).

1877

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Companies that are required to file reports under Section 13 or 15(d) of the US Securities Exchange Act of 1934 provide annual and quarterly technical and financial statements (Securities Exchange Act of 1934, 1934). A search of SEC filings from 2001 to 2023 produced few documents describing projects for which tailings reprocessing has been conducted or considered (Holley et al., 2023). This search did not identify any filings that indicate the use of continuous mechanical mining methods. None of the reports included details of potential geotechnical stability risks that might arise during the excavation of TSF material. A survey of persons designated as QP (Qualified Person) in MMSA (Mining and Metallurgical Society of America) identified only two people who had served as the official QPs signing off for a tailings remining project. It is clear that a distinct set of challenges emerges with respect to the characteristics, potential recoverability, and quantification of the resource, and there was a clear lack of agreement about the relative difficulty of assessing grades and recoverable content. In fact, we know that value exists in mine tailings, but we do not know what, where, how much, and what their value is. There is no defined state-of-practice for characterization of TSFs for remining projects, and no real understanding of the issues involved with the “de-characterization” of TSFs that must be done for remining projects. Entrepreneurs are considering recovery from historic wastes in addition to from new main product operations. However, in the USA right now, permitting a new mining project, whether it is a legacy or existing or new mining operations, even without considering the issues embedded in ESG and SDG, the question is becoming “Who will have to pay the social own environmental costs required to meet the global critical metal demand?” Is it going to be developed countries, or countries that are actually the main producers of these materials? In any case, issues arise concerning environmental and social justice and equity, because those impacts and costs may well be differentially distributed across cultures, communities, and countries. Any new project has been proposed with a holistic perspective that incorporates research, efficiency, sustainable development, and environmental justice.

New materials Many research projects have been focused on recycling and reuse of mine tailings for new materials globally (Araujo et al., 2022). We can modify the tailings waste streams to actually produce high-value products. For example, the world is running out of sand (Golev et al., 2022; Ikotun et al., 2022). This includes good quality sand for concrete manufacture, and for other uses as well, including sand for high purity silicon glass. Vale is developing their own source from their tailings of silica sand, which is being marketed now (Vale, 2022). Many researchers have pursued tailings as substitutes for materials in concrete, which is the secondmost consumed substance on earth after water. Our global consumption of concrete stands at ~30 billion

1878

NEW DIRECTIONS FOR TAILINGS MANAGEMENT tonnes each year, relying upon the annual production of over four billion tonnes of cement, and almost none of it is recycled. The greatest focus has been on cement substitution, because the production of cement produces significant volumes of global emissions of CO2. We can also make alkali-activated materials, generically termed “geopolymers.” Alkali activation transforms alumino-silicate minerals into a paste which can be used to make products primarily for civil construction. If we know the mineral characterization of the tailings, we can control the reactions to produce a stronger and higher valued mineral polymer material. The need for insulation is extremely important for all the new construction, and tailings could be manufactured to produce geofoams that are low density and fast setting (Tumarkin, 2021). 3D printing is possible with newly developed geofoam mixes, and there are early indications that the product of this process results in immobilization of heavy metals. With sulfides in the tailings dealt with separately, the silicates, carbonates, and alumino-silicates can be used for all sorts of materials, including aggregates, clinker, ceramics, zeolites (Campoverde and Guaya, 2023), and glass. A readily available example of a mineral glass is fiberglass, made from melted silica. Melted basalt is also being produced, with glass fibers used for insulation, fiber reinforcement, and composites including rebar that may be substituted for steel in reinforced concrete (with higher strength, low weight, and no corrosion). Using this process, tailings could be melted to produce similar products. Atagic et al. (2022) report on many patents involving tailings and nanotechnology, and there has been much research addressing use of tailings to produce silica and metal nanoparticles (Cruz et al., 2021; Brar et al., 2022; Maass et al., 2019; Wong-Pinto et al., 2021; Shirmehenji et al., 2021; Vasconcellos Brandao et al., 2023; Banerjee et al., 2023). Nanoparticles are widely used, including for catalysis, medication delivery, coatings, and fabrics. Nanomining and bionanomining techniques have been used to produce copper or other metal nanoparticles from mine waste. Nanoparticles will be a key application for the future, and also may be an important technology for passive and active AMD management. The components of tailings, however, are complex in their mineral composition and structural character. Tailings vary greatly for different geologies, ores, and processing, so we have to characterize the tailings. If we establish a tailings resource database with a comprehensive analysis of tailings, types, chemical composition, and physical properties for various mining areas, we can determine the optimal scheme for the utilization of different tailings resources.

Carbonation – CO2 sequestration A very active area of current research is in the use of tailings for carbon sequestration, the process of capture and long-term storage of atmospheric carbon dioxide by mineral carbonation (Back, 2022). This is actually accelerated weathering, currently mostly concerned with ultramafic minerals; hence, the focus has been on Canada and Australia for the production. Mafic silicate minerals weather very quickly with silica

1879

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA dissolution and precipitation of mineral carbonates when the solution is combined with hydrated CO2. The reactions may be speeded by microbial processes. Concern for industry’s carbon footprint is driving sequestration research. FPX Nickel (Canada) has established CO2 Lock Corp. to pursue large-scale, low-cost and permanent carbon capture and storage (FPX Nickel Corp., 2022). Tesla has approached Giga Metals to use carbonation of tailings to produce “green” nickel. Will other consumers pay a premium for “green” metals? Testing is underway to understand how to accelerate the process, and how the process will then scale up. This “green” metal concept involves a premium on the price, and it is not yet known whether consumers will pay that premium for “green” metals. Carbonation research needs to extend beyond ultramafics, e.g., for accelerated weathering (likely requiring microbial enhancement) of more common porphyritic tailings.

Other comments Management of huge tailings volumes focuses on the presence of water and the low strength of the material. We need to develop demonstrations of technologies capable of massive and in-situ dewatering and consolidation of tailings. The use of electrokinetics and electroosmosis deserves to be tried. A living digital model (“digital twin”) should be created for every legacy, inactive, and active TSF. This model should be integrated with all monitoring/sensing data, accessible at all times, and incorporating excellent visualization capabilities. For a TSF, such a model serves as the basis to understand tailings material behaviour and TSF performance that can be validated during construction. Such a model will also directly address risk and increase trust in the risk assessment.

Conclusion Tailings management in the future will be expected by stakeholders to transparently approach zero waste and zero harm. We cannot just consider tailings as waste; they must be considered as a resource. In the USA and other developed countries, new mine operations will likely not be permitted without the explicit and equitable incorporation of environmental and social benefits in the project. These goals will become mandatory for any successful mining proposal, including mine waste and mine water from new and existing mines, by-product recovery projects and remining recovery projects.

References AngloGold Ashanti. 2011. Anglogold Ashanti to acquire interest in first uranium from village main reef. Retrieved from https://www.sec.gov/Archives/edgar/data/1067428/000120561311000108/aga_interest.htm AngloGold Ashanti. 2016. Mineral Resource and Ore Reserve Report.

1880

NEW DIRECTIONS FOR TAILINGS MANAGEMENT Araujo, F.S.M. et al. 2022. Recycling and reuse of mine tailings: A review of advancements and their implications. Geosciences 12(9): 319. https://doi.org/10.3390/geosciences12090319 Atlagic, S.G. et al. 2020. Recent patents in reuse of metal mining tailings and emerging potential in nanotechnology applications. Recent Patents on Nanotechnology, Bentham Science 15(2): 256–269. https://doi:10.2174/1872210514666201224104555 Back, J., Zevenhoven, R., Fagerlund, J. and Sorjonen-Ward, P. 2022. Mineral carbonation using mine tailings – a strategic overview of potential and opportunities. 16th International conference on Greenhouse Gas Control Technologies, GHGT-16. Retrieved from https://ssrn.com/abstract=4285256 Banerjee, A. et al. 2023. Development of nanomedicine from copper mine tailing waste: a pavement towards circular economy with advanced redox nanotechnology. Catalystes 13(2). https://doi.org/10.3390/catal13020369 Brar, K.K., et al. 2022. Green route for recycling of low-cost waste resources for the biosynthesis of nanoparticles (NPs) and nanomaterials (NMs) – A review. Environmental Research 207(112202). https://doi.org/10.1016/j.envres.2021.112202 Buus, C. 2021. Golden Sunlight Mine: A Case Study for Tailing Reprocessing as a Closure Strategy. Accessed July 1, 2023 at https://bc-mlard.ca/files/presentations/2021-5-BUUS-golden-sunlight-mine-reprocessing.pdf Campoverde, J. and Guaya, D. 2023. From waste to added-value product: synthesis of highly crystalline LTA zeolite from ore mining tailings. Nanomaterials 13(8) . https://doi.org/10.3390/nano13081295 Comex. 2023. Advanced Sorting Technologies for the Mining Industry. Accessed July 1, 2023 at https://www.comex-group.com/wp-content/uploads/2023/03/Mining-industry.pdf Cruz, D.R.S., et al. 2021. Recycling of mine waste in the synthesis of magnetic nanomaterials for removal of nitrophenol and polycyclic aromatic hydrocarbons. Chemical Physics Letters 771(138482). https://doi.org/10.1016/j.cplett.2021.138482 Dance, A. 2019. Metallurgical Testing Methods to Evaluate Pre-concentration. Presentation at the meeting of the Canadian Mineral Processors (CIM). Accessed July 1, 2023 at https://www.srk.com/en/publications/metallurgical-testing-methods-pre-concentration Dance, A. 2022. Opportunities for Pre-Concentration: Development of a Lab-Scale Evaluation Test. Presentation at the SME Mine Exchange. Accessed July 1, 2023 https://www.srk.com/en/videos/opportunities_for_preconcentration Evans, J. 2014. Microbial Miners. Royal Society of Chemistry: Chemistry World. Accessed July 1, 2023 at https://www.chemistryworld.com/features/microbial-miners/7879.article FPX Nickel Corp. 2022. FPX Nickel establishes subsidiary company CO2 Lock Corp. to pursue large-scale, lowcost and permanent carbon capture and storage. Retrieved from https://fpxnickel.com/2022/03/fpx-nickel-

1881

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA establishes-subsidiary-company-co2-lock-corp-to-pursue-large-scale-low-cost-and-permanent-carboncapture-and-storage/ Golev, A., et al. 2022. Ore-sand: A potential new solution to the mine tailings and global sand sustainability crises. Report Version 1.4. The University of Queensland and University of Geneva. Harvey, W. 1628. Exercitatio Anatomica de Motu Cordis et Sanguinis in Animalibus. Frankfurt: William Fitzer. Descriptive information accessed June 25, 2023 at https://www.cabinet.ox.ac.uk/de-motu-cordis-1628 Holley, E., Fahle, L., Malone, A., Zaronikola, N., Nelson, P. P. and Spiller, D.E. 2023. Reprocessing mine tailings to recover metals: assessment methods and state of practice. In review. ICMM. 2022. Tailings Reduction Roadmap. Retrieved from https://www.icmm.com/en-gb/news/2022/icmmpublishes-tailings-reduction-roadmap ICMM. 2023. Circular Economy. Retrieved from https://www.icmm.com/en-gb/our-work/innovation-forsustainability/circular-economy Ikotun, J. Adeyeyey, R. and Otieno, M. 2022. Application of mine tailings sand as a construction material. MATEC Web of Conferences 364. https://doi.org/10.1051/matecconf/202236405008 Kinross Gold Corporation. 2015. Kinross reports 2015 third-quarter results. Retrieved from https://www.sec.gov/Archives/edgar/data/701818/000157104915009097/t83541_ex99-1.htm LKAB. 2021. LKAB and Boliden collaborate to recycle mining waste and create circular products. Accessed July 1, 2023 at https://lkab.com/en/press/lkab-and-boliden-collaborate-to-recycle-mining-waste-and-create-circularproducts/ Maass, D., et al. 2019. Biomining of iron-containing nanoparticles from coal tailings. Applied Microbiology and Biotechnology 103: 7231–7240. https://doi.org/10.1007/s00253-019-10001-2 Machiels, L., Dinu, G. and Onisel, L. 2022. NEMO Closing workshop to present key results to wide group of stakeholders. Retrieved from https://h2020-nemo.eu/wp-content/uploads/2022/08/NEMO-DeliverableReport-8.8.pdf Martens, E. et al. 2021. Toward a more sustainable mining future with electrokinetic in situ leaching. Sci. Adv. 7. https://www.science.org/doi/10.1126/sciadv.abf9971 Nelson, P. P. 2022. Future of tailings management. In Tailings Management Handbook: A Life-Cycle Approach. SME, Chapter 17, pp. 277–314. Nwaila, G.T., et al. 2022. The minerals industry in the area of digital transition: an energy-efficient and environmentally conscious approach. Resources Policy 78(102851). Retrieved from https://doi.org/10.1016/j.resourpol.2022.102851 Okada, N., et al. 2020. Automated identification of mineral types and grain size using hyperspectral imaging and deep learning for mineral processing. Minerals 10(9). https://doi.org/10.3390/min10090809

1882

NEW DIRECTIONS FOR TAILINGS MANAGEMENT Robertson, A.M. 2012. FMEA risk analysis: failure modes and effects analysis. Presentation. Accessed June 25, 2023 https://docplayer.net/32185788-Fmea-risk-analysis-failure-modes-and-effects-analysis-andrew-mrobertson-robertson-geoconsultants-inc.html Shirmehenji, R., Javanshir, S. and Honarmand, M. 2021. A green approach to the bio-based synthesis of Selenium nanoparticles from mining waste. Journal of Cluster Science 32: 1311–1323. https://doi.org/10.1007/s10876020-01892-7 Sibanye Stillwater Limited. 2021. Sibanye-Stillwater invests further in the circular economy as it expands its tailings retreatment exposure through a 19.99% investment in New Century Resources. Retrieved from https://www.sec.gov/Archives/edgar/data/1786909/000120561321000135/sibanye_release.htm Sovacool, B.K., Ali, S.H., Bazilian, M., Radley, B., Nemery, B., Okatz, J. and Mulvaney, D. 2020. Sustainable minerals and metals for a low-carbon future. Science 367(6473): 30–33. https://doi.org/10.1126/science.aaz6003 Tumarkin, Paul. 2021. Startup licenses sustainability technology for mining industry. Retrieved from https://news.arizona.edu/story/startup-licenses-sustainability-technology-mining-industry Vale. 2022. Sand produced by Vale is a solution to sand sustainability and mine tailings reduction, according to universities. Accessed July 1, 2023 at https://www.vale.com/w/sand-produced-by-vale-is-a-solution-to-sandsustainability-and-mine-tailings-reduction-according-to-universities Vasconcellos Brandao, I.Y. das Neves, et al. 2023. Bionanomining of copper-based nanoparticles using preprocessed mine tailings as the precursor. Journal of Environmental Management 338. https://doi.org/10.1016/j.jenvman.2023.117804 Wong-Pinto, L.-S., Mercado, A., Chong, G., Salazar, P. and Ordonez, J.I. 2021. Biosynthesis of copper nanoparticles from copper tailings ore – An approach to the “bionanomining.” Journal of Cleaner Production 315. https://doi.org/10.1016/j.jclepro.2021.128107

1883

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA

1884

Chapter Sixteen

Author Index

1885

1886

Proceedings of Tailings and Mine Waste, 2023 5–8 November, 2023, Vancouver, Canada

Author Index Abbaszadeh, Sam ...........................................................................................................................................................................587 Abusaid, Ayman H. ...................................................................................................................................................................... 1557 Acharya, Prabin ............................................................................................................................................................................ 1817 Adria, Daniel ....................................................................................................................................................................................251 Afriyie, George ............................................................................................................................................................................ 1055 Al-Mamun, Mohammad (Mamun) .............................................................................................................................................. 1245 Ang, Eduardo ...................................................................................................................................................................................... 29 Ang, Leila ....................................................................................................................................................................................... 1831 Ansah-Sam, Monica .........................................................................................................................................................................399 Ansere-Bioh, Florence .................................................................................................................................................................. 1845 Archer, A. ....................................................................................................................................................................................... 1457 Ardıç, Ömer ....................................................................................................................................................................................... 43 Ayderman, Aykut ............................................................................................................................................................................... 43 Aynaya, Pepe ..................................................................................................................................................................................451 Bacalzo, Myra ..................................................................................................................................................................................339 Baldwin, Susan Anne .................................................................................................................................................................... 1625 Barbaran, Johanna ...................................................................................................................................................................... 1347 Barham, Thomas .................................................................................................................................................................. 933, 1141 Barish, Mike ................................................................................................................................................................................... 1671 Barker, Malcolm ............................................................................................................................................................................ 1301 Barlow, Kayleigh .............................................................................................................................................................................239 Barría, Esteban ................................................................................................................................................................................387 Bastos Ferreira, Daniel ...................................................................................................................................................................891 Beatty, Karsyn ..................................................................................................................................................................................745

1887

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Bechard, Karen ............................................................................................................................................................................. 1613 Beier, Nicholas A. ................................................................................................................................................................. 297, 527 Bjelkevik, Annika ..............................................................................................................................................................................349 Bleasdale-Pollowy, Aaron .............................................................................................................................................................819 Blin, Régis ....................................................................................................................................................................................... 1779 Bocking, Ken .....................................................................................................................................................................................399 Bohlin, Thomas ..................................................................................................................................................................................427 Boily, Sylvain ....................................................................................................................................................................................413 Borja, Raquel ..................................................................................................................................................................... 1117, 1317 Boshoff, Johan ................................................................................................... 189, 387, 461, 653, 1081, 1091, 1755, 1845 Boswell, Jeremy ............................................................................................................................................................................ 1031 Botto, Tancredi .................................................................................................................................................................................973 Boyd, Timothy ................................................................................................................................................................................ 1419 Brash, Jennifer ..................................................................................................................................................................................181 Bray, Jonathan ................................................................................................................................................................................... 15 Bretas, Victor ....................................................................................................................................................................................993 Brett, David ................................................................................................................................................................................... 1363 Brink, Nicholas R. ............................................................................................................................................................................689 Brown, Tenaya .................................................................................................................................................................................781 Bruton, Mark .................................................................................................................................................................................. 1587 Bundrock, Steve ...............................................................................................................................................................................539 Burgos, Jose ................................................................................................................................................................................... 1587 Burnett, Anthony ...............................................................................................................................................................................727 Burton, Bill ...................................................................................................................................................................................... 1625 Butt, Stephen ....................................................................................................................................................................................853 Cabal, Kelly .................................................................................................................................................................................. 1419 Cabette Ramos, Ricardo ....................................................................................................................................... 375, 1543, 1805 Calais, Gabriel Henrique ..............................................................................................................................................................361

1888

AUTHOR INDEX Calderon Vizcarra, Gino .................................................................................................................................................. 375, 1543 Caldwell, Sam ..................................................................................................................................................................................703 Calvo, Alejandro .......................................................................................................................................................................... 1337 Carvalho Nunes, Alfredo João .................................................................................................................................................. 1793 Castilho, Breno .................................................................................................................................................................................993 Castro, Jeanne .............................................................................................................................................................................. 1399 Cavalieri, Francesco ..................................................................................................................................................................... 1793 Ccotohuanca, José .............................................................................................................................................................. 959, 1373 Celestino, Tarcísio B. .................................................................................................................................................................... 1805 Cerna-Diaz, Alfonso ......................................................................................................................................... 131, 467, 933, 1141 Cervantes, Manuel ....................................................................................................................................................................... 1337 Cezar Rissoli, Ana Luisa .................................................................................................................................................................. 891 Chapman, Peter ............................................................................................................................................................................. 1199 Charca, Osmar .................................................................................................................................................................................933 Chen, Jiarui .......................................................................................................................................................................................131 Cirillo, Fabiana ............................................................................................................................................................................. 1081 Clark, Stephen .................................................................................................................................................................................769 Cobos, Diego ................................................................................................................................................................................. 1337 Coelho Mendes, Anselmo José ......................................................................................................................................................891 Coffey, Jarrad .................................................................................................................................................................. 1017, 1257 Coffin, Jeff ..................................................................................................................................................................................... 1231 Contreras, Iván A. ............................................................................................................................................................. 1493, 1531 Cordero, Alfonso .......................................................................................................................................................................... 1479 Corradi Coelho, Leonardo .............................................................................................................................................................361 Correa, Adolfo .................................................................................................................................................................................807 Costa, Filipe .......................................................................................................................................................................... 993, 1399 Coughenour, Nathan .................................................................................................................................................................... 1687 Coyle, Ezra .................................................................................................................................................................................... 1469

1889

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Cueto, Ignacio A. ....................................................................................................................................................................... 57, 93 da Cunha, Zandra Almeira ...........................................................................................................................................................361 Daliri, Farzad ................................................................................................................................................................................ 1363 Davidson, Richard ........................................................................................................................................................ 143, 467, 985 Davidson, Scott ................................................................................................................................................................................793 Davies, Michael ............................................................................................................................................................................. 1003 Dávila, Rafael ..................................................................................................................................................................................679 Davis, Michael ............................................................................................................................................................................... 1767 Dawson, Richard ..............................................................................................................................................................................327 de Assis, Luciano M. ..................................................................................................................................................................... 1739 de Caralho Thá, Pedro .................................................................................................................................................................. 361 de Faria, F.S. ................................................................................................................................................................................. 1859 de Oliveira Dourado, Daniel ........................................................................................................................................................361 de Oliveira Faria, André ................................................................................................................................................................. 69 Debastiani, Willyan Giorgio .........................................................................................................................................................361 DeMars, Shelby ...............................................................................................................................................................................601 Dennis, Philip ....................................................................................................................................................................................841 Dickinson, Simon ...............................................................................................................................................................................117 Dina, Evelyn ......................................................................................................................................................................................251 Dinh, Olivier ................................................................................................................................................................................... 1779 Dixon, Tyler ......................................................................................................................................................................................627 Dlamini, Sifiso ...................................................................................................................................................................................653 Dueñas, Jorge ................................................................................................................................................................... 1117, 1755 Dufault, Daryl ......................................................................................................................................................................... 601, 881 Dulmage, Adam ............................................................................................................................................................................ 1699 Duncan, Glenn ..................................................................................................................................................................................339 Dzemua, Gideon Lambiv ...............................................................................................................................................................745 Eckhardt, Bridget .............................................................................................................................................................................781

1890

AUTHOR INDEX El Takch, Ali ......................................................................................................................................................................................399 Engman, Jamie .............................................................................................................................................................................. 1277 Error, Braden ....................................................................................................................................................................................985 Ertürk, Hülya Salihoğlu ...................................................................................................................................................................213 Esfandiari, Zahra .............................................................................................................................................................................297 Etezad, Michael .................................................................................................................................................................. 399, 1445 Evans, Dale .......................................................................................................................................................................................561 Evans, Stephen G. ....................................................................................................................................................... 239, 251, 275 Ezama, I. ............................................................................................................................................................................................105 Falck, Hendrik ..................................................................................................................................................................................745 Fanni, Riccardo .................................................................................................................................................................................117 Farina, Paolo ....................................................................................................................................................................................973 Fernandes, Marcos Túlio ................................................................................................................................................................361 Fernández, Fabricio ........................................................................................................................................................................451 Ferreira, Juliano ...............................................................................................................................................................................575 Fiestas, Jimmy ................................................................................................................................................................................ 1755 Finocchiaro, Nicola Mirko ........................................................................................................................................................... 1129 Fisher II, James C. ........................................................................................................................................................................ 1671 Fonseca, Rodrigo .......................................................................................................................................................................... 1399 Fortier, Christopher .........................................................................................................................................................................727 Fourie, Andy ............................................................................................................................................................ 117, 1069, 1505 Franklin, Kathryn ..............................................................................................................................................................................881 García Schmidt, Ignacio .............................................................................................................................................................. 1663 Garfias, Juan.....................................................................................................................................................................................117 Gavioli, Fernanda ...........................................................................................................................................................................993 Gazzarrini, Paolo ............................................................................................................................................................................327 Ghafghazi, Mason ............................................................................................................................................................. 117, 1445 Ghahramani, Negar ............................................................................................................................................................. 251, 275

1891

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Gheibi, Amin .......................................................................................................................................................................................... 3 Gibson, Charlotte ............................................................................................................................................................................745 Godley, Darryl ................................................................................................................................................................. 1155, 1663 Gonçalves, Thatyane ................................................................................................................................. 375, 1543, 1373, 1805 Gore, Matthew S. ............................................................................................................................................................................169 Gounder, Raguvind ...................................................................................................................................................................... 1141 Gover, Scott .................................................................................................................................................................................. 1199 Gregianin Testa, Francisco ......................................................................................................................................................... 1651 Grosso, Andrea ............................................................................................................................................................................. 1129 Gu, Frank ..........................................................................................................................................................................................819 Guay, Francis ...................................................................................................................................................................................413 Guerra Acosta, Noe ......................................................................................................................................................................... 57 Guerzoni, Henrique .........................................................................................................................................................................575 Guillén-Guillén, Jorge Bricio ........................................................................................................................................................... 29 Guimont, Hubert ........................................................................................................................................................................... 1687 Gunnteg, Marcus .............................................................................................................................................................................117 Gupta, Ranjiv ...................................................................................................................................................................... 933, 1601 Gutierrez, Javier .............................................................................................................................................................................189 Hall, Cassandra ...............................................................................................................................................................................587 Hammett, Ross ..................................................................................................................................................................................793 Han, Chao ...................................................................................................................................................................................... 1613 Harris, Simon ....................................................................................................................................................................................503 Harvey, Jason W. ............................................................................................................................................................. 1493, 1531 Hatton, Christopher N. .................................................................................................................................................................... 213 Hawn, Julia .................................................................................................................................................................................... 1601 Haynes, Andy ...................................................................................................................................................................................181 Herweynen, Wes .......................................................................................................................................................................... 1105 Herza, Jiri ...................................................................................................................................................................................... 1257

1892

AUTHOR INDEX Hickman, Ron ......................................................................................................................................................................................... 3 Hilgers, Jason ...................................................................................................................................................................... 933, 1601 Hinostroza, Jose ...............................................................................................................................................................................287 Hockley, Daryl .................................................................................................................................................................................263 Hogan, Arielle A. ......................................................................................................................................................................... 1493 Hojka, Kris .........................................................................................................................................................................................539 Holley, Rachel ............................................................................................................................................................................... 1755 Holmes, Andrew ...............................................................................................................................................................................841 Holmström, Tobias ...........................................................................................................................................................................349 Hone, Dale ........................................................................................................................................................................................715 Honores, Diana ................................................................................................................................................................................189 Huamán, Miguel ...............................................................................................................................................................................287 Huamanyauri, Jeymy ......................................................................................................................................................................287 Huanchi, Jhoel ..................................................................................................................................................................................287 Hussien, Mahmoud N. ................................................................................................................................................................... 1517 Huza, Jessica ....................................................................................................................................................................... 413, 1289 Ileme, Ogechi Mary ..................................................................................................................................................................... 1347 Imran, Md Al .................................................................................................................................................................................. 1625 Inaudi, Daniele .............................................................................................................................................................................. 1779 Ingabire, Edouardine-Pascale ......................................................................................................................................................413 Innis, Sally .........................................................................................................................................................................................251 James, Cliff .................................................................................................................................................................................... 1363 James, Michael .............................................................................................................................................................................. 1289 James, Rachel ...................................................................................................................................................................... 841, 1613 Jamieson, Heather ...........................................................................................................................................................................745 Jamil, Mohammad Shafaet ...........................................................................................................................................................853 Jeldes, Isaac A. ...............................................................................................................................................................................689 Johndrow, Tamara ............................................................................................................................................................. 587, 1245

1893

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Jónsson Menzies, Tristan ................................................................................................................................................................. 641 Joseph, Stephen ............................................................................................................................................................................ 1091 Julien, Michel ....................................................................................................................................................................... 413, 1289 Kafash, Masood H. ........................................................................................................................................................... 3, 143, 467 Karimi, Zana ............................................................................................................................................................................... 3, 143 Kaswalder, Francesco .................................................................................................................................................................. 1129 Katapa, Kanyembo ............................................................................................................................................................ 143, 1141 Keizer, Jonathan ..............................................................................................................................................................................759 Kelly, Shane A. .............................................................................................................................................................................. 1493 Ketilson, Erik .................................................................................................................................................................................. 1663 Knutsson, Roger ................................................................................................................................................................................427 Kormann, Alessander ................................................................................................................................................................... 1409 Koshima, Akira .................................................................................................................................................................................327 Kozikowski, Michal ..........................................................................................................................................................................263 Kratochvil, David .............................................................................................................................................................................867 Kunz, Nadja ......................................................................................................................................................................................251 Küpper, Angela ............................................................................................................................................................................ 1245 Lage de Castro, Alessandro ..........................................................................................................................................................375 Lameiras, F.S. ................................................................................................................................................................................ 1859 Laporte, Patrick ...............................................................................................................................................................................413 Larkin, Hayley ............................................................................................................................................................................... 1755 Laxman, Manoj ............................................................................................................................................................................. 1301 Le Borgne, Vincent ........................................................................................................................................................................ 1699 Leal, Ana ........................................................................................................................................................................................ 1409 Ledesma, Osvaldo N. ....................................................................................................................................................................... 57 Lemos Júnior, Marcos Antônio .................................................................................................................................................... 1651 Lepine, Thomas .............................................................................................................................................................................. 1289 LePoudre, Chad ...............................................................................................................................................................................263

1894

AUTHOR INDEX Leshuk, Tim ........................................................................................................................................................................................819 Létourneau, Yanick ..........................................................................................................................................................................413 Li, Weidong ......................................................................................................................................................................................727 Li, Yuan ................................................................................................................................................................................. 155, 1479 Liantono, Josephine ...................................................................................................................................................................... 1069 Lima, Carlos ......................................................................................................................................................................................575 Limoges Shaigetz, Marielle ...........................................................................................................................................................413 Linton, Nick ..................................................................................................................................................................................... 1767 Liu, Chenying ...................................................................................................................................................................................... 15 Liu, Fangzhou ................................................................................................................................................................................. 1817 Liu, Shielan ..................................................................................................................................................................................... 1385 Liu, Wei .............................................................................................................................................................................. 1069, 1445 Longey, Rob ............................................................................................................................................................. 339, 1105, 1363 Longhi, M.A. ................................................................................................................................................................................... 1859 Lopes dos Santos Lopes, Henrique ............................................................................................................................................ 1793 Lopez Rivarola, F. ..........................................................................................................................................................................105 López, Andrea .................................................................................................................................................................. 1573, 1587 Lotter, Li-Bonné ............................................................................................................................................................................. 1091 Lundell, Dan ......................................................................................................................................................................................427 Lupo, John ...................................................................................................................................................................................... 1003 Macciotta, Renato ...........................................................................................................................................................................527 Macedo, Jorge ......................................................................................................................................................................... 15, 947 Magalhães, Gabriel G. .............................................................................................................................................................. 1739 Mancini, Silvia ..................................................................................................................................................................................841 Marino, Nicholas A.C. ................................................................................................................................................................. 1739 Marsh, Michael .................................................................................................................................................................................853 Martinelli, Mario ..............................................................................................................................................................................297 Martins, P.F.F. ................................................................................................................................................................................ 1859

1895

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Masengo, Edouard ............................................................................................................................................................. 413, 1289 McDougall, Scott ................................................................................................................................................ 227, 239, 251, 275 McGreevy, Julian ......................................................................................................................................................................... 1625 McGregor, M. ..................................................................................................................................................................... 105, 1573 McKellar, Megan .............................................................................................................................................................................239 McLeod, Harvey ........................................................................................................................................................................... 1385 McNab, Louise ......................................................................................................................... 189, 461, 653, 1081, 1091, 1845 Medina, Miguel ................................................................................................................................................................................309 Melendez, Stefhany ........................................................................................................................................................................481 Mendes, Mardon .............................................................................................................................................................................807 Mensa, Nathaniel Asifu ............................................................................................................................... 189, 653, 1091, 1845 Methiwala, Jayamini .................................................................................................................................................................... 1301 Meunier, Daniel ................................................................................................................................................................................503 Meuzelaar, Tom ...............................................................................................................................................................................793 Miles, Sarah M. ...............................................................................................................................................................................829 Millar, Robert ...................................................................................................................................................................................881 Miller, Elizabeth ...............................................................................................................................................................................793 Miranda, Jessica ..............................................................................................................................................................................327 Mitchell, Andrew ..............................................................................................................................................................................227 Mohammadian, Abdolvahid ..........................................................................................................................................................309 Monroy, Manuel ............................................................................................................................................................................ 1479 Monteiro Furtado, Renata .......................................................................................................................................................... 1651 Moon, Nigel ................................................................................................................................................................................... 1831 Moore, Caitlin ..................................................................................................................................................................................539 Moore, Ryan .................................................................................................................................................................................. 1557 Morales, Camilo ...............................................................................................................................................................................387 Moreno, Laura ............................................................................................................................................................................... 1337 Morrison, Kimberly Finke ............................................................................................................................................................ 1055

1896

AUTHOR INDEX Motta, Barbara ............................................................................................................................................................................. 1409 Mulligan, Ryan P. .............................................................................................................................................................................239 Murphy, Brent ...................................................................................................................................................................................793 Murray, Len ................................................................................................................................................................................... 1277 Mutsaerts, Rachael ..........................................................................................................................................................................867 Myers, Kenneth ................................................................................................................................................................................551 Nasseri-Moghaddam, Ali ...............................................................................................................................................................169 Nelson, Priscilla P. ......................................................................................................................................................................... 1871 Newman, Phil ..................................................................................................................................................................... 1573, 1587 Ngan, Aldrich ...................................................................................................................................................................................819 Nishiyama, Matt ........................................................................................................................................................................... 1711 Noël, M. .............................................................................................................................................................................................105 Norambuena Mardones, Raul .................................................................................................................................................. 57, 93 Novikov, Anton ................................................................................................................................................................................... 93 Nugent, Laura ............................................................................................................................................................................... 1601 O’Sullivan, Kaitlyn ............................................................................................................................................................................. 81 Oats, Bill ......................................................................................................................................................................................... 1105 Obeidat, Dafar N. ............................................................................................................................................................ 1493, 1531 Oboni, Cesar ................................................................................................................................................................................. 1187 Oboni, Franco ............................................................................................................................................................................... 1187 Oboudi, Marjan ...............................................................................................................................................................................679 Oester, Tyler ....................................................................................................................................................................................781 Ofori, Maud Ofosua .................................................................................................................................................................... 1845 Olofsson, Dennis ...............................................................................................................................................................................349 Olson, Scott M. .................................................................................................................................................................................131 Ordonez, Richard ............................................................................................................................................................. 1117, 1755 Osborne, Dave .............................................................................................................................................................................. 1671 Otieno, Bobby ............................................................................................................................................................................... 1469

1897

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Paihua, Solange ...............................................................................................................................................................................947 Pascoe, Marnie ............................................................................................................................................................................. 1041 Pastor, Brandon ............................................................................................................................................................................ 1755 Patterson, Kate ............................................................................................................................................................................. 1277 Pauly, Matthew J. ...........................................................................................................................................................................689 Penman, James .............................................................................................................................................................................. 1277 Pereira Lima, André ..................................................................................................................................................................... 1793 Pereira, Frank ................................................................................................................................ 375, 1373, 1409, 1543, 1805 Pereira, Karina ............................................................................................................................................................................. 1739 Perotti, Deborah A. ...................................................................................................................................................................... 1805 Pezzi, Andrea ................................................................................................................................................................................ 1129 Piciacchia, Luciano ...........................................................................................................................................................................667 Pinheiro, Mauricio ............................................................................................................................................................................327 Pinto Chaves, Arthur ..................................................................................................................................................................... 1651 Pollock, Gordon W. ..................................................................................................................................................................... 1557 Porter, Michael .............................................................................................................................................................................. 1217 Portocarrero, Julio ...........................................................................................................................................................................881 Preciado, Humberto F. .................................................................................................................................................................... 29 Prince, Mitchell .............................................................................................................................................................................. 1031 Printzell, Lena ...................................................................................................................................................................................349 Priscu, Caius ....................................................................................................................................................................... 1003, 1169 Priscu, Doina .................................................................................................................................................................................. 1169 Purrington, James ......................................................................................................................................................................... 1587 Qiu, Yvonne (Yirao) .................................................................................................................................................................... 1557 Quaglia, Gastón ........................................................................................................................................................................... 1155 Quintero, Sebastian ........................................................................................................................................................................917 Raabe, Kevin ....................................................................................................................................................................................515 Rahman, Asma ..................................................................................................................................................................................841

1898

AUTHOR INDEX Ramírez-Chávez, Rigoberto ............................................................................................................................................................ 29 Rana, Nahyan M. .................................................................................................................................................................. 251, 275 Re, Giacomo .................................................................................................................................................................................. 1805 Redmond, John .................................................................................................................................................................................143 Reid, David .............................................................................................................................................................. 117, 1469, 1505 Repenning, Ricardo .........................................................................................................................................................................973 Revelo Mendez, Nathalia ..............................................................................................................................................................503 Rezende, Larissa ........................................................................................................................................................................... 1409 Rhee, Jason .................................................................................................................................................................................... 1557 Ribbons, Jordan ............................................................................................................................................................................ 1199 Rivet, Sam ...................................................................................................................................................................................... 1711 Roach, Paul .................................................................................................................................................................................... 1409 Roberts, Jeff .....................................................................................................................................................................................841 Rocha, Claudio .................................................................................................................................................................................973 Rocha, Sabrina .............................................................................................................................................................................. 1399 Rodrigues, João Paulo .................................................................................................................................................................... 575 Rodriguez, Arturo ............................................................................................................................................................................917 Rodriguez, Caleb ......................................................................................................................................................................... 1505 Rodriguez, Fernando ......................................................................................................................................................................189 Rogers, Josh ............................................................................................................................................................. 189, 1117, 1755 Roman, Brahian ................................................................................................................................................................................481 Romero, Jesus E. ...............................................................................................................................................................................201 Rondinel, Efrain A. ...........................................................................................................................................................................689 Rothrock, Tara ..................................................................................................................................................................................439 Rudy, Andy .................................................................................................................................................................................... 1231 Ruffing, Daniel .............................................................................................................................................................................. 1687 Russell, Brad .................................................................................................................................................................................. 1217 Rust, E. ............................................................................................................................................................................................. 1429

1899

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Rust, M. ........................................................................................................................................................................................... 1429 Sadrekarimi, Abouzar ............................................................................................................................................................ 81, 905 Saeedi, Armin ...................................................................................................................................................................................297 Sage, Ed ......................................................................................................................................................................................... 1755 Salzsauler, Kristin ............................................................................................................................................................... 793, 1613 Santander, Ricardo .........................................................................................................................................................................703 Santos Júnior, Mauro Pio dos .......................................................................................................................................................... 69 Santos Rodrigues, Raphael ......................................................................................................................................................... 1793 Santos, Evelyn ............................................................................................................................................................................... 1373 Saunders, Sam ...................................................................................................................................................................................... 3 Sauvé, Madeleine ...........................................................................................................................................................................601 Scampoli, Lucas ............................................................................................................................................................................. 1081 Schafer, Haley L. .............................................................................................................................................................................527 Schmall, Paul......................................................................................................................................................................................561 Schuettpelz, Craig ...........................................................................................................................................................................613 Seda, Jesus H. ..................................................................................................................................................................................689 Sena, Nathalia .................................................................................................................................................................................993 Servigna, Daniel .................................................................................................................................................................... 29, 1419 Sgaoula, Jamel ............................................................................................................................................................................. 1517 Shandro, Janis ............................................................................................................................................................................... 1169 Sharp, James ................................................................................................................................................................................. 1445 Shaw, Brienna ..................................................................................................................................................................................601 Shelton, Ashleigh ........................................................................................................................................................................... 1845 Sheykhloo, Pooya A. .................................................................................................................................................................. 3, 985 Siaw, Robert .................................................................................................................................................................................. 1845 Siemoneit, Steven ............................................................................................................................................................................587 Silva Dia, Enzo .................................................................................................................................................................................959 Silva, João Paulo .......................................................................................................................................................................... 1373

1900

AUTHOR INDEX Silva, R.K.R. .................................................................................................................................................................................... 1859 Simjee, Yusuf ....................................................................................................................................................................................641 Simms, Paul .......................................................................................................................................................................................297 Singh, Ryan .................................................................................................................................................................................... 1257 Small, Andy ................................................................................................................................................. 275, 1245, 1347, 1373 Smith, Kyle ........................................................................................................................................................................................117 Smith, Larissa ....................................................................................................................................................................................841 Snow, Robert ................................................................................................................................................................................. 1601 Soares Pereira, Géssica ................................................................................................................................................................. 891 Sousa, Simone ..................................................................................................................................................................................575 Souza, Deni ........................................................................................................................................................................................807 Souza, Luciano .............................................................................................................................................................................. 1373 Sova, Madeline R. ...........................................................................................................................................................................213 Sportsman, Emily ..............................................................................................................................................................................793 Springer, Darren ..............................................................................................................................................................................503 Stenberg, Mikael .............................................................................................................................................................................349 Strachan, Clint ........................................................................................................................................................................ 515, 587 Sucupira, Vinícius .............................................................................................................................................................................807 Surrette, Allison ................................................................................................................................................................................745 Take, W. Andy ............................................................................................................................................................. 239, 251, 275 Teixeira, Felipe Jorge .................................................................................................................................................................... 361 Tellez, Holman ..................................................................................................................................................................................309 Terlisky, A.G. ....................................................................................................................................................................................105 Thiaga, Milan ...................................................................................................................................................................................627 Tiwari, Bandana ........................................................................................................................................................................... 1505 Toney, Tony ................................................................................................................................................................................... 1671 Torres, Paola ....................................................................................................................................................................................947 Torres, Rogerio ................................................................................................................................................................................327

1901

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Töyrä, Sara ......................................................................................................................................................................................427 Trejo, Engels .....................................................................................................................................................................................917 Trottier, Marc Olivier ......................................................................................................................................................................881 Tyler, Hadleigh ................................................................................................................................................................................143 Vachon, Melody ...............................................................................................................................................................................841 van Zyl, Dirk .....................................................................................................................................................................................155 Vander Vis, Kimberly .....................................................................................................................................................................613 Vargas-Moreno, Carlos Omar ........................................................................................................................................................ 29 Varia, Manthan ................................................................................................................................................................................155 Vasconcelos, Anelisa .......................................................................................................................................................................327 Vasquez, Jose ..................................................................................................................................................................................807 Vega, I.J. ...........................................................................................................................................................................................105 Velloso, Raquel Quadros ...............................................................................................................................................................451 Vera, Francisco ................................................................................................................................................................................387 Vermeulen, N.J. ............................................................................................................................................................................. 1457 Vestman, Marcus ..............................................................................................................................................................................349 Vides, Antonio ..................................................................................................................................................................................667 Vieira Carneiro, Jessé Joabe .......................................................................................................................................................... 69 Villalobos, Miguel ............................................................................................................................................................................993 Villalta, Geinfranco ..................................................................................................................................................................... 1317 Viola, Raphael .................................................................................................................................................................................959 Vitorugo, Lirielly ........................................................................................................................................................................... 1739 Vulpe, Cristina ............................................................................................................................................................................... 1069 Vummidi, Narayanee .................................................................................................................................................................... 1755 Walder, Roberto .......................................................................................................................................................................... 1779 Waldron, Mike .................................................................................................................................................................................933 Walkenbach, Tanya .......................................................................................................................................................................613 Walsh, Andrea .................................................................................................................................................................................239

1902

AUTHOR INDEX Walsh, Maguire ............................................................................................................................................................................ 1081 Wang, Calvin ...................................................................................................................................................................................461 Wanninayake, Ajitha ......................................................................................................................................................................627 Warren, Greg ..................................................................................................................................................................................793 Watts, Bryan ................................................................................................................................................................................. 1217 Wedage, Pathma ...........................................................................................................................................................................727 Welch, Taylor ................................................................................................................................................................................ 1017 Whatnall, Oliver ...............................................................................................................................................................................703 White, Renee ....................................................................................................................................................................................399 White, Trevor ...................................................................................................................................................................................227 Whiter, Bria ......................................................................................................................................................................................503 Whitmore, Nicole ............................................................................................................................................................................769 Whittall, John ...................................................................................................................................................................................227 Wiklund, Viktor ................................................................................................................................................................................349 Wilkins, Desiree ...............................................................................................................................................................................439 Willan, Martyn Bryan .....................................................................................................................................................................181 Williams, David ...............................................................................................................................................................................917 Williams, Holly .................................................................................................................................................................................263 Wilson, Stuart ...................................................................................................................................................................................703 Winter, Christopher R. ...................................................................................................................................................................489 Wishart, Laura .............................................................................................................................................................................. 1277 Wood, Ray .................................................................................................................................................................................... 1409 Xia, Xuexin ........................................................................................................................................................................................297 Yenne, Lisa ............................................................................................................................................................... 3, 467, 933, 985 Zabolotnii, Elena ..............................................................................................................................................................................297 Zambrano, Gonzalo .......................................................................................................................................................................297 Zanoni, Luca ................................................................................................................................................................................... 1129 Zegarra, Edwin ................................................................................................................................................................................189

1903

TAILINGS AND MINE WASTE 2023 ● VANCOUVER, CANADA Zehforoosh, Farshad .......................................................................................................................................................................905 Zhang, Ying ................................................................................................................................................................................... 1557 Zhu, Yi ............................................................................................................................................................................................. 1041

1904