Program

Global tailings production has increased steadily over recent decades, with modern mining operations capable of generating up to one million tons of tailings per day. In this context, Brazilian legislation requires that the identification and implementation of tailings management technologies follow a rigorous selection process based on the combination of economically feasible alternatives and techniques capable of reducing the risks associated with tailings management to acceptable levels throughout the entire life cycle of a facility.

Selecting both the disposal site and the disposal method is therefore a critical stage in mining project planning, as it involves not only technical and economic feasibility but also environmental, social, and regulatory constraints that may affect long-term operational sustainability. This challenge is particularly relevant in environmentally sensitive regions such as the Brazilian Amazon, where multiple competing factors must be considered within the framework.

To address this complexity, a multicriteria decision matrix for tailings disposal technology selection was developed using the Analytic Hierarchy Process (AHP). The method enables hierarchical structuring of decision criteria and supports the comparative evaluation of alternatives based on their relative importance. The weighting of criteria at each hierarchical level was established through pairwise comparisons conducted by a panel of experienced professionals from both the mining company and a partner engineering firm.

The adopted criteria encompassed technical-operational, environmental, social, legal, and economic dimensions. Technical aspects addressed constructability and expected performance of containment structures; environmental aspects focused on natural resource preservation and impact mitigation; social aspects considered surrounding communities and public acceptance; legal aspects ensured compliance with applicable regulations; and economic aspects evaluated competitiveness and financial sustainability. Each dimension was further subdivided into specific evaluation indicators to ensure consistency in the comparison process.

Qualitative judgments derived from expert assessment were complemented by rational quantitative indicators, enabling a balanced integration of subjective perception and objective technical data. This approach enhanced methodological robustness by reducing individual bias and increasing the reliability and transparency of the decision-making process.

The selection matrix was subsequently applied in an alternatives assessment for a copper mining project located in the Brazilian Amazon. The evaluated technological options included a tailings dam, a shared disposal stack integrating tailings and waste rock, and a compacted tailings stack. Based on the results obtained, the tailings dam was identified as the most suitable alternative for the evaluated operation, demonstrating the applicability of the proposed framework as a decision-support tool for tailings management in complex environmental and regulatory contexts.

GeoStable Tailings are homogenous mixtures of tailings and waste rock that are being investigated as an alternative mine waste management solution capable of improving storage efficiency and long-term environmental and closure goals. A laboratory mixing study was conducted to evaluate the influence of tailings type, solids content, mixture ratio, and placement energy on the engineering behavior of GeoStable Tailings. Whole tailings and cycloned underflow tailings were mixed with waste rock at a waste rock mass to tailings mass ratio ranging from 0.25 to 2.0. Specimens were prepared at multiple solids contents and placed using loose placement, a drop chute, and Standard Proctor compaction to represent a range of field placement energies. Testing included density, air permeability, penetration resistance, and slump evaluations. Results indicate that both whole tailings and underflow tailings can produce viable GeoStable Tailings mixtures, although performance is strongly dependent on solids content, mixture ratio, and placement method. Dry density generally increased with increasing mixture ratio and placement energy, while penetration resistance and trafficability improved with compaction. Whole tailings exhibited greater sensitivity to water content, with lower solids contents resulting in excessive slump and reduced stability. Underflow tailings displayed more consistent behavior across the range of solids contents evaluated, exhibiting limited slump and a broader range of acceptable mixture ratios. Air permeability was generally low for compacted mixtures but increased in mixtures containing higher solids contents and larger proportions of waste rock due to the development of connected macropores. Results further indicate that the engineering behavior of GeoStable Tailings is governed by the interaction between tailings characteristics, solids content, mixture ratio, and placement energy. The results provide insight into the range of mixture conditions suitable for field-scale evaluation and support the continued development of laboratory-based screening methods for GeoStable Tailings applications.

Over the past two decades, the mining industry has increasingly shifted from conventional low‑solids hydraulic tailings deposition toward dewatered tailings placement methods, including thickened and filtered tailings. This paper presents the design, construction, and performance of a thickened tailings Down‑Valley Discharge (DVD) scheme implemented in an arid and highly seismic region. The facility incorporates a 110 m‑high rockfill embankment dam with upstream filter zones that allow controlled seepage of tailings water while retaining tailings solids, with seepage captured by a smaller return water dam located approximately 500 m downstream. The scheme supports an 8 Mtpa copper operation, providing approximately 210 Mt of tailings storage over a 26‑year mine life, with staged embankment raises from initial commissioning in 2005 to the current final stage extending capacity to 2030.

The innovative integration of paste‑thickened tailings with a filter rockfill (leaky) dam has delivered significant technical, environmental, and operational benefits. These include the formation of an essentially unsaturated tailings deposit, substantially higher in situ tailings densities than achievable through conventional compaction, reduced geotechnical instability risks, improved water recovery and management, and a reduced overall environmental footprint. The paper discusses key design considerations, operational performance to date, and long‑term implications of the adopted tailings and water management strategy. The outcomes demonstrate a practical, cost‑effective alternative to filtered tailings storage facilities for high‑throughput operations and provide a framework for safer and more sustainable TSF design.

This paper presents the planning and execution approach for the detailed engineering, contracting, and construction of a dry‑stacked tailings storage facility delivered under a highly constrained nine‑month dry‑season window. The project encompassed a 50‑hectare footprint and a total storage capacity of 60 million m³, requiring disciplined information management and precise coordination across project phases.

At project outset, the reliability of existing subsurface data was reduced because core boxes from the 2019 drilling campaign were no longer available, undermining lithological confidence. To re‑establish the geotechnical baseline, the designer required a new subsurface investigation program and complementary analytical studies. In 2024, nineteen additional boreholes were drilled, instrumentation was installed, and undisturbed block samples were collected. Supporting analyses included hydrological modeling, stress–strain evaluation, refinement of the Critical State Line (CSL), and hypothetical failure assessments. These activities progressed in parallel with vegetation clearing to maintain alignment with the construction schedule.

Detailed engineering advanced concurrently with the investigation program under a deviation‑based review strategy, in which design updates were issued only when validated discrepancies emerged. The investigation was monitored daily by the designer’s field geologist, who logged samples, confirmed stratigraphy, and provided continuous feedback to the engineering team.

Procurement activities were synchronized with engineering maturity. A pre‑qualification package was released in September, followed by a competitive tender in October involving five bidders, site inspections, and two rounds of technical clarifications. The contractor was selected in late December, with kickoff in early January 2025. By March, the contractor had mobilized and initiated foundation clearing and material‑yard preparation. Construction proceeded in double shifts from Monday to Saturday. A key feature of the delivery model was the continuous presence of the designer’s geologist and geotechnical engineer on each shift, supported by a daytime CAD technician to promptly issue drawings and avoid work interruptions.

The executed scope included approximately 1.5 million m³ of excavation, full foundation cleaning, vegetation removal across ~50 hectares, and installation of the internal drainage system – 2.5 km of primary drains and 1.5 km of secondary and tertiary drains. Filtration and transition zones were constructed using controlled aggregates, including ~30,000 m³ of rockfill, ~50,000 m³ of coarse gravel, ~40,000 m³ of intermediate gravel, and ~90,000 m³ of sand. Rigorous QA/QC procedures covered quarry sourcing, stockpile management, and field placement.

The project encountered significant hydroclimatic challenges, including a record rainfall event in April – approximately 1000% above expected – which resulted in soft soil development, loss of drainage‑protection material, embankment constraints, reduced drainage‑wall stability, and restricted placement areas for excess material. Mitigation measures included iterative constructability reviews, rapid re‑planning using the Last Planner System (LPS), Six‑Weeks‑Ahead scheduling, daily coordination, and weekly safety meetings.

This case study highlights three key lessons for future dry‑stacked tailings projects: (i) detailed, assertive engineering ensures strong alignment between design intent and field conditions; (ii) placing design personnel in construction operations enhances decision agility and adaptability; and (iii) comprehensive QA/QC across all phases is critical to the performance of internal drainage systems and to minimizing rework. These practices are essential for delivering tailings infrastructure within compressed operational windows.

The safe management of tailings remains a key priority for the mining industry, particularly related to the risks associated with the long-term storage of saturated tailings and water behind containment structures within conventional tailings storage facilities. One of the industry responses has been the adoption of co‑disposal, in which dewatered tailings are stored along with waste rock or other granular spoil materials. In recent years, there has been an increase in the number of papers and keynotes published on co‑disposal design (including for two of the facilities described in this paper) (Bainbridge et al. 2021, 2022) and on trial facilities to assess feasibility for advancement to full-scale implementation. It has been highlighted, however, that there are comparatively few papers presenting case studies on the implementation of (including discussion of the challenges associated with) full-scale co-disposal facilities.

This paper will present case studies for three co-disposal facilities currently operating at the Greenhills Operations and Fording River Operations mines owned by EVR Operations Ltd. in the Elk Valley, British Columbia. These three facilities were designed based on different operating philosophies; as such, the authors will compare the technical and operational challenges and successes, as well as approaches to the coordination of multidisciplinary teams, including across multiple companies, with the intent to provide lessons learned for the wider industry.

In addition, one of the facilities discussed in this paper was developed to incorporate low-permeability zones to promote suboxic conditions within the waste rock spoil to improve long-term geochemical stability. To support this discussion, the paper will provide an example of how both geotechnical and geochemical objectives can be met when developed in tandem.

Designing a responsible and economic tailings storage facility requires confidence in tailings variability and knowledge of how this variability will impact tailings management. Obtaining answers to key tailings-related questions early in study phases with a tailings test work component during the geo-metallurgical mapping stage can support critical design decisions.

Geo-metallurgical mapping is a common practice to inform processing plant design and operation decisions based on the variability of the ore to be processed. The mapping process starts with the costly collection of drill core samples, followed by a series of laboratory tests, analyses, and modelling. The “waste stream” generated from the tests (laboratory produced tailings) are a valuable resource that can be used to characterize the variability of the anticipated tailings at minimal additional cost.

This paper presents an add-on approach to any geo-metallurgical mapping program where the laboratory generated tailings are subjected to an additional series of brief indicative tailings characterization laboratory tests. Identifying the tailings variability in terms of dewatering, transport, geotechnical or geochemical aspects can highlight root causes for potential future tailings challenges. This adds significant value to the tailings design with minimal additional cost so that owners can implement an informed and optimized tailings management plan from project initiation. The Twin Metals filtered tailings project is presented as an example to share tailings mapping testing protocols, results, and lessons learned to better future endeavors where this practice may be applicable.

Critical state soil mechanics (CSSM) is commonly used to analyse the behaviour of tailings storage facilities (TSFs). This requires determining the critical state line (CSL) which typically involves six to ten triaxial tests on reconstituted specimens. Given the strong heterogeneity of TSFs, multiple soiltypes, and therefore multiple CSLs, often exist within a single facility. Furthermore, selecting representative materials for CSL determination based on the coarsest and finest particle size distributions (PSD) does not necessarily capture the full range of CSL variability. Conversely, previous results show that the limit void ratios (emin and emax) correlate with CSL descriptors, Γ100 and λ, which are the void ratio (e) at a mean effective stress (p') of 100 kPa, and the slope of the CSL in e-log10p', respectively.

Herein, we examine the validity of previously proposed correlations between CSL descriptors and limit void ratios by expanding with 20 soils and 50 tailings an existing multi-source database of 160 nonplastic tailings and soils. We also highlight the independence from PSD of the CSL slope Mtc in deviatoric stress (q) – p' space.

The original Γ100-emin correlation remains robust across the wider range of geomaterials in the expanded database. The Γ100–emax correlation exhibits greater scatter, much of which appears attributable to particle morphology. Geomaterials dominated by hard minerals typically yield λ < 0.10, while tailings with fine platy particles produce higher values. The trend λmax = emin/3 is a practical bound on the compressibility of soils and tailings. The correlations are useful to aid the characterization of CSL variability within a TSF, make preliminary estimates of the CSL of nonplastic geomaterials, and identify spurious laboratory results.

The slope Mtc appears largely independent of PSD within individual TSFs, even where fine platy particles are present, supporting the treatment of Mtc as a well-defined property in stratified TSFs composed of thin layers with different gradations.

Understanding how water flows through tailings is critical for both dewatering system design and the long-term performance of filtered and conventionally deposited tailings. Process design is based on dynamic filtration testwork in which the time to form a cake is recorded, and a material-dependent specific cake resistance (α) can be obtained to model filtration under simplifying assumptions. Geotechnical practice typically evaluates hydraulic conductivity (K) through low hydraulic gradient permeability tests on fixed-volume specimens.

Although in both fields it is Darcy’s law that governs the fluid flow through the porous bed of solids, there is a lack of direct comparison of the results from these different set of tests and a discussion of how they can be used across disciplines.

This paper presents an analysis of filtration and geotechnical test results for a suite of iron ore tailings sourced from several operations. Samples were on the one hand subjected to constant pressure filtration tests, such as those typically used to design filtration plants; and on the other hand subjected to the geotechnically-oriented permeability tests such as the falling or constant head tests on compacted samples or the falling head test on slurry-deposited tailings.

The permeability of the materials is then calculated for all sets of tests and can be compared when considering the different void ratios and conditions at which the tests take place. Permeability during filtration has been on average 10-50% of the geotechnically derived value, even when measured at higher void ratio. The results of the study provide practical guidance to compare these typically separate datasets. When coupled with basic filtration process models, geotechnical permeability tests could be used to provide a preliminary approximation of filtration technology selection. Measured on homogeneous fixed specimens, geotechnical tests provide the upper bound of permeability and suggest options for filtration process improvement.

Geotechnical and hydrogeologic models are critical tools used to support seepage, stability, and draindown analyses of tailings storage facilities (TSFs). For many applications, a realistic representation of variably saturated behavior is required to capture changes in pore pressure, saturation, and water levels. Variably saturated flow models require definition of the saturated hydraulic conductivity and the moisture retention curve (MRC), also commonly referred to as the soil water characteristics curve. However, depending on the stage of the project, or other potential physical constraints such as limited access within an impoundment, direct measurements on intact or reconstituted tailings may not be possible, and the practitioner therefore must rely on empirically-derived pedotransfer functions and professional judgement.

Most pedotransfer function methods are based on natural soils as compared to mine tailings, and at present, the few studies that support the use of pedotransfer functions for tailings hydraulic properties are limited to a narrow range of tailing types. This study presents measured MRCs and their associated hydraulic properties spanning a broad range of textures, including low-plasticity silty and clayey silt tailings, and sand-dominated cyclone underflow tailings representative of beach and embankment construction zones. Taken together, the dataset reflects the variability encountered in contemporary tailings facilities. Results demonstrate strong texture dependence of air-entry pressure, residual water content, and pore-size distribution characteristics; finer-grained tailings exhibit higher air-entry values and greater water retention at equivalent suctions, whereas coarser-grained tailings show more rapid desaturation and reduced storage capacity. A comparison of hydraulic properties and MRCs for reconstituted and intact samples as a function of various dry bulk densities is presented. Limitations associated with predicting both the saturated and unsaturated hydraulic properties of tailings using empirically-derived pedotransfer functions are also presented. The dataset presented provides representative tailings MRCs based on tailings texture and consolidation, which can be used to support TSF design, operations, and closure planning.

This study investigates the influence of anisotropic soil behaviour on the static and dynamic response of a downstream-raised copper Tailings storage facility (TSF). Two advanced constitutive models, available in the Imperial College Finite Element Program (ICFEP), are employed in these analyses: the Bounding Surface Plasticity model (BSPM), an isotropic state parameter-based constitutive framework serving as base case; and the MIT-S1 model, which incorporates inherent and induced anisotropy. Both models are calibrated using laboratory and in-situ tailings data.

Two-dimensional plane strain hydro-mechanically coupled analyses were performed to simulate the construction and operational stages of the facility. The results show that both models predict stable dam behaviour; however, notable differences arise in predicted deformation patterns within the tailings impoundment. The MIT-S1 model produces larger settlements and a more pronounced plastic response compared with the BSPM formulation, reflecting anisotropic strength mobilisation and stress rotation effects.

The seismic response is further evaluated using one-dimensional column analyses. The BSPM successfully captures key aspects of liquefaction behaviour, including pore pressure buildup, propagation, and post-seismic settlement. In contrast, the MIT-S1 model shows limited ability to reproduce these mechanisms, indicating the need for further development to represent liquefaction processes.

Mine operators in the oil sands mining region of northern Alberta typically raise external tailings storage facilities using the upstream construction method. Constructed by hydraulic deposition of tailings streams, these facilities are often built to 40+ m height over many years. A result of the chosen deposition method, vast quantities of process water are introduced into the facility as tailings are spigotted, potentially raising the phreatic surface over time. Internal seepage management systems are required to suppress the phreatic surface and maintain low pore pressures as part of the safe operation of these facilities. Downward seepage is also required to promote the compaction of tailings to a non-liquefiable (dilative) state. Active monitoring (e.g., vibrating wire piezometers, slope inclinometers, flow gauges) informs the observational method of construction to assess performance of the structure and seepage management systems during raising.

Construction of robust internal drain systems, embedded within the tailings structure, is a preferred method of internal seepage water management to maintain a low phreatic surface. These internal drains generally consist of perforated drainpipes surrounded by a granular filter system to collect the seepage water and solid outtake pipes to discharge the collected seepage water outside of the structure. Dependent on the size of the drainpipe perforations and the grain size of the surrounding tailings, the granular filter system may require multiple layers or transition zones to mitigate the internal erosion failure mode and potential loss of tailings material that could result in piping or clogging of the drain system (reducing performance).

The project team conducted site- and material-specific Continuing Erosion Filter (CEF) testing to assess the compatibility of the granular filter system of an internal seepage management system for an operating oil sands tailings storage facility. This testing program expands on current industry knowledge of filter compatibility criteria as first developed by Dr Karl Terzaghi (1943), No Erosion Filter (NEF) testing by Sherard and Dunnigan, (1989), and CEF testing by Foster and Fell (2001), that historically focus on conventional earth dam configurations with fine-grained low-permeability cores. Filter retention compatibility ratios (D15,f/D85,b) between produced transition layer aggregates and sandy tailings streams were assessed in a laboratory environment with reconstituted test specimens, targeted gradation curves using a 300 mm permeameter cell to assess the filter compatibility of site-specific materials. Evidence-based lab results support the application of industry-accepted filter compatibility ratio limits and acceptable gradation bands where necessary, while supporting wider gradation bands where possible. Expanding the range of allowable transition aggregate gradation bands may offer flexibility in the procurement of granular filter materials and yield economic benefits in the production, crushing, material screening, and construction quality process.

Accurate characterization of water balance and seepage at tailings storage facilities (TSFs) is fundamental to long-term performance assessment, environmental risk analysis, and regulatory compliance. Water movement through tailings governs pore pressure evolution, solute transport, and the magnitude and extent of seepage to the underlying groundwater system. Despite this importance, seepage estimation in practice frequently relies on simplified approaches that inadequately capture the coupled and transient nature of these processes, introducing substantial uncertainty into risk assessments and environmental predictions.

This study presents a comprehensive, multi-scale modeling framework developed to address these limitations through explicit, process-based representation of the mechanisms governing tailings water balance, seepage generation and groundwater system response due to the TSF operation. The framework is organized across two modelling scales: (1) the deposit scale, representing high-resolution hydraulic-mechanical processes occurring within the tailings; and (2) the hydrogeological-system scale, representing regional unsaturated-saturated groundwater flow and transport.

At the deposit scale, the deposition plan is analyzed to reduce the three-dimensional problem into a set of spatially representative one‑dimensional column models. Each column simulates the transient, coupled processes associated with sequential lift placement, self-weight consolidation and settlement, rewetting from precipitation, drying driven by evaporation, and unsaturated flow through the evolving tailings profile. This integrated formulation is essential for capturing the hydraulic–mechanical interactions that govern the tailings water balance and ultimately control the magnitude and timing of seepage at the base of the deposit.

The resulting seepage rates provide a recharge boundary condition for the hydrogeological-system scale model, where they are applied to a three-dimensional transient flow and transport model representing the saturated and unsaturated groundwater system. At this scale, particular emphasis is placed on explicitly representing the vadose zone, which constitutes the principal attenuating pathway between the base of the tailings and the regional water table. The progressive advance of the seepage front through unsaturated materials is simulated, accounting for capillary storage and the mixing between natural groundwater and contact water, generating the delay and attenuation of recharge to the saturated zone. Once recharge reaches the water table, the model resolves groundwater level response and solute transport in the saturated domain, completing the hydrogeological assessment.

Sensitivity analyses at both scales are used to identify controlling parameters and bound predictive uncertainty. Collectively, this framework provides a transparent and physically grounded basis for estimating seepage and its possible environmental implications. By integrating detailed process‑based modeling across multiple spatial and temporal scales, it delivers robust and defensible predictions suited for environmental permitting, groundwater impact assessment, and long‑term water management planning.

Protection of water quality is a fundamental requirement for tailings facility management, requiring systematic approaches to source control, treatment where necessary, and comprehensive monitoring. This paper examines water quality protection for tailings facilities, including source control, treatment, and monitoring approaches.

The paper establishes the water quality context, including: sources of water quality impacts from tailings facilities; regulatory requirements for water quality protection; GISTM requirements for environmental management; and the relationship between water quality and closure. Understanding this context motivates comprehensive water quality management.

Source control approaches are examined, including: tailings characterization for water quality prediction; geochemical management of tailings; water balance management to minimize contaminant transport; and engineering controls for seepage. The paper demonstrates how source control reduces treatment requirements.

Water treatment approaches are presented, including: active treatment systems for contaminated water; passive treatment systems including constructed wetlands; treatment during operation and closure; and selection of appropriate treatment technologies. The paper addresses the range of treatment options available.

Monitoring programs are addressed, including: parameters for water quality monitoring; monitoring locations and frequency; data management and interpretation; and compliance reporting. The paper shows how monitoring verifies water quality performance.

Integration with Risk-Informed Decision Making (RIDM) is a central theme, including: water quality risks and their assessment; risk-based prioritization of protection measures; cost-effectiveness of water quality management; and communication of water quality risks.

Adaptive management is examined, including: response to monitoring results; adjustment of management measures; continuous improvement of water quality protection; and long-term stewardship considerations.

Case studies illustrate water quality protection approaches: a facility where comprehensive source control minimized treatment needs; a facility where active treatment was required for compliance; and a closure facility where passive treatment was implemented.

Recommendations are provided for water quality protection, including: source control approaches; treatment selection; monitoring programs; and adaptive management. The paper provides practitioners with tools for effective water quality management.

Spring freshet rapidly releases substantial volumes of water over a short period, generating important challenges associated with the operation and performance of tailings storage facilities, from active operations to reclamation. Snow accumulation on tailings storage facilities during winter is typically the largest component of the annual water inventory, resulting in high-flow peaks during spring snowmelt that place significant demands on tailings storage facilities, surface water management, and mine water treatment infrastructure. Despite its importance, the contribution of spring runoff to tailing storage facilities water balance is often poorly quantified, introducing uncertainty into seasonal water management strategies. In this context, developing robust snow monitoring tools approaches could ensure effective spring mine water management and reduce uncertainty.

This work aims to develop and assess the potential of a high spatial-resolution drone-based sensing approach for semi-real-time monitoring of the snow water balance in Tailings storage facilities during snowmelt. The objective of this study is to assess the capacity and accuracy of Light Detection and Ranging (LiDAR) and structure-from-motion (SfM) photogrammetry surveys conducted with an unmanned arial vehicle, to map the evolution of the snow depth of a reclaimed tailings storage facility during snowmelt in Abitibi-Témiscamingue, Québec. Although active remote sensing techniques are widely applied in watershed hydrology, their performance to map small catchments such as Tailings storage facilities in a mine water management scheme has not yet been thoroughly evaluated.

Digital elevation models were generated from both remote sensing approaches and validated using ground control points and conventional Global Navigation Satellite System surveys. Snow depth models were derived by differencing the snow-covered and snow-free digital elevation models and validated using in situ snow depth measurements. Results indicate that LiDAR is the most accurate and reliable approach with vertical errors generally within 3 cm. In contrast, SfM photogrammetry showed a limited ability to represent feature-less snow-covered surfaces, with errors reaching up to 66 cm. This is largely due to insufficient aerotriangulation tie points in such areas. The accuracy of the photogrammetry-derived digital elevation models were also more sensitive to the number and distribution of ground control points compared to LiDAR-derived models. Building on these results, a preliminary snow-water equivalent-based snowmelt model was developed using LiDAR-derived snow depths and simple temporal parameters. The developed model yielded simulated snow-water equivalent with errors generally within 3 cm. The presence of low but dense vegetation limits the accuracy of the reference snow-free digital elevation model; thus, model accuracy could still be improved. This integrated approach enables temporal snowpack monitoring and provides water volume estimation released during snowmelt, representing meaningful advancement toward effective spring-water management on tailings storage facilities.

Tailings storage facilities (TSFs) are essential engineered structures that ensure the safe containment of mine tailings and associated process water. They also represent long lasting legacy infrastructures that are required to remain in place long after mine closure (e.g., centuries to millennia) without ongoing maintenance. Their design and long-term performance depend heavily on effective water management, as elevated pond levels and phreatic surfaces can compromise dam stability, result in release of contact water, and/or increase the likelihood of overtopping during extreme flood events. Such failures can lead to severe environmental, health, and economic consequences for downstream communities and ecosystems, highlighting the critical importance of integrating robust hydrological planning into facility design and operations.

With climate variability intensifying, traditional water management approaches based solely on historical climatic records are no longer sufficient. Practitioners are increasingly adopting stochastic water balance modelling as a core design and operational tool, to enhance facility resilience and manage long-term risk. These models integrate forward‑looking climate change scenarios and anticipated future operational and post-closure conditions to quantify the probability and magnitude of extreme precipitation and runoff events. By simulating a wide and dynamic range of hydrometeorological futures, stochastic modelling provides designers and operators with a more realistic and detailed understanding of potential water management challenges and risks across the extended life cycle of a TSF.

In the context of facility design, these probabilistic insights support the development of more robust freeboard criteria, pumping capacity requirements, emergency spillway sizing, and operational storage strategies. Designers can evaluate whether current infrastructure meets future climate and operational or post-closure demands, or whether adaptive design elements – such as increased storage volumes, surge capacity, source controls, or enhanced discharge systems – are required. This approach enhances the ability of TSFs to maintain structural integrity under a broader range of plausible future conditions, improving risk management and regulatory confidence.

From an environmental and water management perspective, stochastic modelling strengthens operational decision‑making by identifying periods of heightened water risk and quantifying the effectiveness of proposed mitigation strategies. Operators can use model outputs to refine water management plans, optimize pumping schedules, and develop emergency drawdown triggers and protocols that minimize the risk of uncontrolled releases. This more dynamic and predictive approach allows mine operators to address climate‑driven water challenges proactively rather than reactively, contributing to more sustainable and environmentally responsible facility management.

This presentation outlines how forward‑looking stochastic water balance modelling is used for TSF design. It describes methodologies that integrate climate change projections, field data, evolving mine conditions, and operational guidelines, and highlights results demonstrating how probabilistic modelling can help mitigate water‑related climate risks. Building on current industry practices, the work illustrates how modern modelling tools can enhance both the design and day‑to‑day water management of tailings facilities, ultimately improving the resiliency and environmental performance of mine waste infrastructure.

Hydrogeologic models developed for legacy mine sites must rely on decades of incomplete and uncertain data. While traditional model calibration emphasizes discrete measurements targets such as hydraulic head, much of the available information at legacy sites exists as operational records, qualitative observations, and bounded estimates (for example, ranges of seepage rates or timing of facility operation). These non-traditional types of data are often underutilized despite their relevance to predictive uncertainty.

This study presents a modeling framework that systematically incorporates such information by translating historical ranges, thresholds, and operational knowledge into parameter bounds and calibration constraints. A groundwater flow and particle tracking model was developed using MODFLOW 6 and MODPATH 7 and calibrated under uncertainty using PESTPP-IES. The key application is the use of multi-scale parameterization combined with data thresholds and ranges to enforce both site-wide water balance and physical realism at multiple spatial scales. This approach allows uncertain historical data to be directly incorporated into the model parameterization and calibration process, improving the conditioning of model predictions.

The framework was applied to estimate the probability of tailings seepage migration beyond a regulatory boundary over a 200-year period, characterize likely flow pathways, and identify dominant source areas. Particle tracking was adapted to represent seepage volumes by linking particle release to simulated flux, enabling probabilistic evaluation of advective transport.

Results demonstrate that incorporating non-traditional data types can substantially reduce predictive uncertainty while producing physically consistent representations of complex historical water management practices. The approach provides a practical pathway for leveraging legacy datasets to support risk-informed decision-making for mine closure and long-term site management.

Pseudostatic slope stability analysis remains the most widely used method for seismic design of tailings dams (Saragoni and Garrido, 2022), with the seismic coefficient as its most consequential parameter. Approaches range from prescriptive values or fractions of peak ground acceleration to the displacement-compatible method of Bray and Macedo (2024), which links the coefficient to site-specific hazard and an allowable displacement threshold. The method is valid for bounded shear deformation under inertial loading but does not apply when contractive tailings undergo seismically triggered flow failure, a condition affecting a large proportion of existing facilities (Davies, 2002).

This paper consolidates scattered guidance from academic methods (e.g., Macedo et al., 2023) and regulatory standar

Pseudostatic slope stability analysis remains the most widely used method for seismic design of tailings dams (Saragoni and Garrido, 2022), with the seismic coefficient as its most consequential parameter. Approaches range from prescriptive values or fractions of peak ground acceleration to the displacement-compatible method of Bray and Macedo (2024), which links the coefficient to site-specific hazard and an allowable displacement threshold. The method is valid for bounded shear deformation under inertial loading but does not apply when contractive tailings undergo seismically triggered flow failure, a condition affecting a large proportion of existing facilities (Davies, 2002).

This paper consolidates scattered guidance from academic methods (e.g., Macedo et al., 2023) and regulatory standards (ICOLD Bulletin 194; CDA; MAC; GISTM; ANCOLD) into a structured five-step procedure for selecting the seismic coefficient in tailings dams. The procedure screens material state to identify flow-liquefaction-susceptible facilities outside the pseudostatic domain, tiers the allowable displacement by consequence category and construction method, and defines explicit escalation triggers to Nonlinear Dynamic Effective Stress Analysis (NDESA).

Four published case histories across distinct tectonic regimes illustrate the approach on both the Pass and Step 5 branches. Prescriptive methods are unconservative in subduction interface settings and unable to detect flow instability, with direct consequences when contractive materials, upstream construction, and moderate to high seismicity coincide.

ds (ICOLD Bulletin 194; CDA; MAC; GISTM; ANCOLD) into a structured five-step procedure for selecting the seismic coefficient in tailings dams. The procedure screens material state to identify flow-liquefaction-susceptible facilities outside the pseudostatic domain, tiers the allowable displacement by consequence category and construction method, and defines explicit escalation triggers to Nonlinear Dynamic Effective Stress Analysis (NDESA).

Four published case histories across distinct tectonic regimes illustrate the approach on both the Pass and Step 5 branches. Prescriptive methods are unconservative in subduction interface settings and unable to detect flow instability, with direct consequences when contractive materials, upstream construction, and moderate to high seismicity coincide.

Tailings dams serve as critical infrastructure for the storage of mine waste, often containing materials with varying levels of toxicity that may pose significant risks to human safety and environmental health in the event of structural failure. The stability of these facilities, especially under dynamic loading conditions such as seismic events, is a key concern of the mining industry and its regulatory partners. Dynamic loading can induce permanent deformation and trigger liquefaction, potentially undermining tailings dam integrity and leading to catastrophic outcomes. Nonlinear dynamic effective stress analysis is a useful tool to evaluate the stability of mine waste structures following an earthquake, as it provides estimates of deformation across the structure whereas typical empirical analyses like Limit Equilibrium provide only a Factor of Safety. This study presents a comparative evaluation of two advanced liquefaction constitutive models used in dynamic deformation modeling - P2PSand and PM4Sand - for their ability to simulate dynamic deformation and liquefaction potential in tailings storage facilities (TSFs). A two-dimensional numerical analysis was performed using FLAC2D on an idealized critical TSF cross-section. Both constitutive models were calibrated using comprehensive field and laboratory testing data, providing realistic representation of material behavior. Both models were found to be able to reproduce the general trends observed in the laboratory data and resulted in reasonable estimates of deformation and pore pressure in the numerical simulations. However, PM4Sand was found to be better suited for practical liquefaction triggering and deformation analysis due to a smoother hysteretic response and more progressive stiffness degradation. P2PSand instead can be used in scenarios where localized deformation is expected, as the model has a stiffer response.

This study evaluates the influence of constitutive models used for mine tailings on the seismic response of a tailings dam constructed using the centerline method and located in the northern Andes of Peru. For this purpose, a dynamic numerical model was developed using FLAC v9, incorporating the following three constitutive models: PM4Sand, PM4Silt, and P2PSand. Two seismic records with different seismogenetic characteristics were used and two types of tailings materials were analyzed. The results show that, while the final displacements may be comparable under proper calibration of each model, significant differences are observed in excess pore pressure generation, shear strain development, and liquefaction response. Moreover, it is observed that the frequency content of the input ground motion plays a key role in the structural response, producing greater displacements when it approaches the dam fundamental frequency. In this context, the results highlight the importance of appropriate selection and calibration of the constitutive model, as well as the consideration of the frequency content of seismic records, to achieve a reliable assessment of the dynamic behavior of tailings dams.

Recent tailings storage facility (TSF) failures have demonstrated the importance of brittle strain-weakening and static liquefaction mechanisms in controlling instability and large deformations. The 2014 Mount Polley failure involved brittle strain-weakening of a foundation layer that led to progressive failure, while the 2018 Cadia failure involved weakening of a foundation layer that triggered static liquefaction of stored tailings and subsequent displacement of the embankment. These case histories demonstrate that understanding deformation patterns associated with residual strength mobilization is essential for risk-informed TSF management.

Remediation measures intended to achieve the limit equilibrium (LE) stability factors of safety (FoS) recommended by guidelines may be constrained by downstream infrastructure or lease boundaries and may therefore be unable to fully satisfy the design criteria. In such cases, the assessment of the expected deformation response and post-trigger behaviour may provide a more appropriate framework for assessing system performance than reliance on LE FoS values alone.

This paper presents a comparative large-strain deformation assessment of an upstream TSF section with a downstream buttress under residual undrained conditions. A representative interim buttress geometry incorporating a bench, and a shear key is analyzed assuming direct mobilization of residual undrained strengths within the contractive tailings and foundation materials. Two large-strain numerical approaches available within FLAC2D are evaluated: automatic remeshing and the Material Point Method (MPM) (MPoint2D module).

The results indicate significant differences between the remeshing and MPM simulations with respect to the magnitudes of the predicted displacements. The analyses also demonstrate a significant sensitivity to mesh and material point discretization, even when simplified constitutive representations are adopted. The configuration with full shear key penetration exhibited substantially shorter runout distances than the configurations with partial or no shear key penetration.

Recent failures of tailings storage facilities (TSFs) under seismic loading have emphasized the critical role of liquefaction and cyclic softening in cyclone tailings (underflow), particularly when these materials are placed at low relative densities. Field evidence has shown that the mechanical response of cycloned tailings is strongly influenced by their placement conditions, degree of compaction, and resulting density state, which govern pore pressure generation, deformation patterns, and overall stability under seismic loading.

In this context, this study investigates the influence of relative density on the cyclic behavior of a cycloned tailings dam under realistic placement and compaction conditions, focusing on how different density states affect the seismic response of the structure. A series of nonlinear dynamic analyses was performed in FLAC2D using the advanced constitutive models PM4Sand and PM4Silt to represent coarse and fine tailings, respectively, while the model parameters were calibrated using laboratory cyclic tests and in-situ-based correlations to represent the behavior of cycloned tailings across a range of relative densities.

A calibrated numerical approach is employed to simulate the cyclic response of cyclone tailings under a crustal earthquake, considering relative densities of 60%, 65%, 70%, and 85%. The results show that lower relative densities lead to larger horizontal displacements (up to 0.9 m), greater crest settlements (up to 0.87 m, equivalent to 1.09% of dam height), and higher shear strains (up to -6%), indicating a higher susceptibility to deformation and potential instability. Conversely, increasing relative density significantly reduces displacement demand, pore pressure effects, and shear strain development, although diminishing improvements are observed beyond relative densities of approximately 70%. Additional analyses incorporating post-placement densification through a depth-dependent relative density profile produced only moderate changes in seismic response while providing a more realistic representation of in-situ conditions. The analyses also suggest that a threshold relative density between 65% and 70% provides a substantial improvement in seismic performance for the evaluated geometry.

Filtered tailings have been identified as a potentially advantageous tailings management strategy, offering benefits such as reduced water consumption and reduced consequence of failure. Despite these advantages, industry adoption remains limited, and there is no summary of existing operations. This paper presents an updated global benchmarking of filtered tailings projects, with particular emphasis on throughput capacity and climatic conditions, building upon the Study of Tailings Management Technologies commissioned over a decade ago by Mine Environment Neutral Drainage (MEND), the Mining Association of Canada (MAC), and Natural Resources Canada (NRCan).

The conclusions of the benchmarking indicate that the key trends identified in the earlier study have largely persisted: filtered tailings are predominantly applied in operations with relatively low throughput and within arid or semi-arid climates. Evidence, both documented and anecdotal, suggests that significant challenges continue to impede implementation at higher throughputs and in wetter climates. These constraints are linked to technical, operational, and economic factors associated with large-scale dewatering, material handling, and climate resilience.

The authors built upon these findings by identifying targeted areas for improvement across defined zones of the throughput and climate graph. Key considerations include stronger integration with upstream processes such as geological modeling and mineral processing, as well as resilience in dewatering technologies and approaches. Additional focus areas include material handling, management of adverse climatic conditions, and strategies for addressing out-of-specification material. The authors recommend the adoption of site-specific, fit-for-purpose tailings management solutions, emphasizing that standardized, off-the-shelf approaches are insufficient for filtered tailings. They further highlight the need for an innovative approach to research and development, alongside different ways of thinking for evaluating and establishing the business case for tailings management.

Filtered Stack Tailings (FST) facilities are commonly placed, compacted, and operated in a partially saturated state. Their performance is governed by conventional geotechnical parameters together with matric suction, soil-water retention, unsaturated hydraulic conductivity, density, fabric, wetting-drying history, and the response of the foundation system. Suction should be treated as a transient state variable rather than a permanent design property; loss of suction, localized saturation, and compression-induced pore-pressure response must be considered during design and operation.

This paper presents a scenario-based framework that links material characterization, climate and water management, numerical modelling, construction controls, monitoring, and trigger-action-response planning. Two foundation settings are considered: a filtered stack on competent foundation with a low groundwater table, and a filtered stack over previously hydraulically deposited tailings. The comparison shows that the first case is primarily a stack moisture-control problem, whereas the second requires an integrated stack-foundation assessment of consolidation, undrained response, excess pore-pressure generation and dissipation, settlement, rate of rise, and observational performance management.

Filtered tailings placed by stacking is an alternative to conventional hydraulic tailings deposition. Critical to the promise of filtered tailings is placement at an adequately dense state, which substantially mitigates liquefaction and slope instability risks associated with hydraulically deposited tailings. However, the role of placement condition on the transition from dilative to contractive behavior under applied stresses, a critical consideration for long-term stack stability, remains unclear. This study investigates the compression behavior and stress-state evolution of compacted filtered copper mine tailings subjected to increasing vertical stress. One-dimensional compression behavior was evaluated in oedometer tests on unsaturated tailings, with suction at each load increment estimated from van Genuchten soil-water retention relationships calibrated to dry density. Suction was used to compute mean effective stress in the unsaturated specimens at each level of applied stress. Critical state lines for both tailings were established from conventional triaxial compression tests on saturated specimens.

Compression paths were evaluated in void ratio–mean effective stress space relative to the saturated critical state lines. Saturated specimens remained on the dilative side of critical throughout the imposed stress range. In contrast, interpretation of unsaturated specimens using a single-valued effective stress formulation and the saturated critical state line produced apparent indications of contractive behaviour. The results demonstrate limitations associated with applying a single-valued effective stress framework and saturated critical state concepts to unsaturated filtered tailings. The findings suggest that net stress and suction as independent stress state variables are required to adequately describe the volume change behaviour and state of unsaturated tailings. Consequently, interpretation of state of unsaturated filtered tailings based solely on effective stress and saturated critical state relationships may lead to unrealistic assessments of contractive tendencies.

Hydraulic Dewatered Stacking (HDS) is an engineered co-disposal approach developed by Anglo American in which fines-free drainage sand is co-placed with tailings slurry in alternating layers, establishing three-dimensional drainage channels that accelerate consolidation, promote desaturation and enhance water recovery. The concept was demonstrated at industrial scale at the El Soldado Mine in Chile, where a purpose-built greenfield facility delivered substantial reductions in tailings degree of saturation and high process water recovery rates within weeks of deposition. A critical prerequisite underpinning that demonstration was the use of Coarse Particle Flotation (CPF) sand, which consistently delivers low fines content and high permeability. Extending HDS to operations without CPF infrastructure requires alternative drainage sand sources, typically produced through hydraulic segregation or mechanical classification of the tailings stream, over which permeability can decrease by orders of magnitude.

This paper presents preliminary results from an ongoing pilot-scale experimental program at the Universidade Federal de Ouro Preto (UFOP), Brazil, using copper tailings and hydraulically segregated beach sand from the Sossego Mine tailings storage facility (IOCG deposit). Two purpose-built consolidation cells, instrumented with tensiometers, moisture probes, and settlement monitoring, replicate the HDS geometry through interbedded lifts with sand batches prepared at progressively elevated target fines contents. Following saturated filling, lateral layer valves are opened to capture matric suction development, pore pressure dissipation, and dewatering rates. Saturated hydraulic conductivity and soil-water retention curves for the classified sand are currently under measurement. This work provides the first evaluation of non-CPF HDS performance under tropical boundary conditions to inform design thresholds and calibrate unsaturated flow models.

Filtered tailings and co-disposal practices have been discussed, debated, and researched for many years. While many operations have performed research and trials, Green’s Creek has successfully adapted this practice for 16 years, in an environment that receives more than 50 inches of precipitation a year. This paper presents a case study to share lessons learned on this proven solution of comingling for tailings and mine waste management, focusing on an operational approach.

The Greens Creek Mine in Southeast Alaska has operated a filtered tailings facility concurrently with waste rock facilities since 1989. This operation, located in both a National Forest and National Monument, has been implementing improved operational practices for tailings stack management. The location of the mine presents significant challenges including wet conditions, sensitive environments, and space constraints. Continuing to mine on the site has required Green’s Creek to implement new solutions for waste disposal to reduce risk and environmental impact. One improved practice has been to implement co-disposal methods in the existing tailings facility and the waste rock produced through mine development.

Research into tailings and waste rock co-disposal began in 2005 and initial field trials were held in 2006. This research focused on how co-disposal affected the geotechnical, geochemical, and hydrological properties of the tailings stack. With agency approval, the practice of co-disposal was implemented in 2010. Since then, practices have been modified based on challenges presented and lessons learned in operations. Benefits to the site include improved stability, reduction of ARD potential, and ability to manage the mine’s overall waste profile. These benefits come with challenges including increased permeability, increased placement complexity, and other operational considerations. Long-term benefits to the closure plan are expected such as allowing for concurrent reclamation of historic waste rock sites into the Tailings Disposal Facility during operation. This reduces the environmental and closure risk at the site associated with several facilities. Through continued engineering analyses, standard operating procedures, strategic planning, and oversight of the facility, these challenges are diminished and an accepted practice has been achieved.

This case study will show that filtered tailings and co-disposal are not only possible in high-rainfall environments, but can provide significant benefits for stability, geochemistry, closure and more. It comes with challenges, but the positive outcomes are critical to the mine’s operational success.

For decades, stability assessments of mine tailings impoundments have been largely expressed by the calculation of a global Factor of Safety (FoS) using limit equilibrium methods. While FoS provides a convenient scalar index of relative stability, its continued use as the primary determinant of safety for modern tailings storage facilities is increasingly difficult to justify. These facilities are increasingly large, staged, and constructed using materials that exhibit strain-dependent behaviour that can include significant strain-softening leading to catastrophic incidents.

Limit-equilibrium methods implicitly assume ductile response and do not account for stress redistribution, strain localization, progressive failure, or pore pressure generation associated with contractive behaviour. Further, nothing monitored directly at any facility can directly relate to a FoS value. It is also woefully inconsistent to allow comparison of “robustness” using the approach. Facilities with assigned FoS values well above unity may remain susceptible to brittle collapse while systems capable of ductile deformation may perform satisfactorily at FoS values near unity. The key determinant of stability is therefore not the magnitude of FoS, but whether the system’s constitutive behaviour permits the levels of deformation to ever be experienced occur so without causing a failure event.

This paper argues that deformation-based analysis, incorporating stress–strain behaviour, pore pressure coupling, and staged construction history, provides a more defensible basis for evaluating and communicating the stability of a given tailings facility. Modern numerical modeling tools allow explicit assessment of deformation-controlled mechanisms and can be directly integrated with monitoring data. In this context, FoS should be regarded only as a comparative indicative screening tool and never an absolute discriminator of safety. A paradigm shift is required – from equilibrium margins toward deformation capacity within a fundament understanding of constitutive behaviour as the primary metric of stability.

The mining industry has undergone a fundamental transformation in how decisions regarding tailings storage facilities (TSFs) are made, evolving from deterministic standards-based design (SBD) approaches to comprehensive risk-informed decision making (RIDM) frameworks. This evolution, accelerated by recent catastrophic failures at Brumadinho (2019), Eagle Gold (2024), and other operations, has created confusion among practitioners regarding the distinctions between various risk-related terminologies and their appropriate application across the facility lifecycle.

This paper provides critical clarification of the spectrum of decision-making approaches: Standards-Based Design (SBD), which relies on prescribed factors of safety and deterministic loading conditions without explicit consideration of failure probability or consequences; Risk-Based Design (RBD), which incorporates probabilistic analysis of failure modes but may not fully integrate broader societal and environmental considerations; Risk-Informed Design (RID), which uses risk assessment as one input among many in the design process; Risk Assessment (RA), the systematic process of identifying and analyzing potential failure modes and consequences; and Risk-Informed Decision Making (RIDM), a comprehensive framework integrating technical risk assessment with governance, stakeholder engagement, economic factors, environmental considerations, and adaptive management throughout the facility lifecycle.

Using case studies from the authors' experience across North and South American mining operations, this paper demonstrates how RIDM frameworks provide superior outcomes compared to purely deterministic approaches. Examples include tailings facilities where RIDM identified critical failure modes not captured by traditional stability analyses, leading to targeted risk reduction measures that reduced societal risk by over 80% while optimizing capital expenditure.

The paper addresses persistent industry challenges including: confusion between risk assessment as a tool versus RIDM as a process; the misconception that RIDM is only applicable to high-consequence facilities; the tendency to treat RIDM as a compliance exercise rather than a continuous decision-making framework; and the challenge of integrating quantitative risk assessment with qualitative considerations including community values and environmental justice.

Recommendations are provided for organizations implementing RIDM, including governance structures, competency requirements, stakeholder engagement protocols, and integration with existing management systems. The paper draws on the authors' leadership roles in developing CDA-USSD guidance on RIDM for tailings facilities to provide practical implementation pathways.

As climate change intensifies precipitation extremes and compound rainfall events increasingly drive failure risk, RIDM frameworks offer the adaptive capacity necessary to manage evolving risk profiles. This paper provides practitioners with clarity on terminology, methodology, and implementation to advance the state of practice in tailings facility safety.

The Global Industry Standard on Tailings Management (GISTM) has been developed with the aspirational goal of zero harm to people and the environment with zero tolerance for human fatality; to this end, many of its principles introduce a step-change in the way the industry typically approaches design, operation, governance, and closure of tailings facilities. In particular, the standard calls for the development of a robust design; however, there is limited guidance in the literature on what constitutes robustness in the design of a tailings facility.

The authors consider that a robust design is one that is practical and fit-for-purpose, while not being overly sensitive to changes and uncertainties, and is able to recover relatively easily if its performance is less than adequate. It considers the possible range of variability of the foundation and fill parameters, the construction conditions, the mine plan as well as ore and tailings processing. Further, it is a design that manages risk without being unduly conservative.

This paper explores key considerations through the various design development elements that contribute to a robust design throughout the tailings facility lifecycle including: design basis definition; option development and selection; design measures to manage integrity; material balance; constructability; performance monitoring; and costs. This paper shows that a robust design is underpinned by a thorough inquiry of each design element for its potential range of conditions, the outcomes that may occur if those conditions are realized and the associated impacts of the variability of these conditions on the risk profile, along with consideration for the challenges related to implementation of measures to enable recovery back to intended performance; this inquiry process can help designers and owners identify the critical elements that warrant additional investigation, controls and robustness to adequately manage risks.

This paper is intended to initiate discussion building towards a common understanding of what robust design looks like in practice, in support of implementing GISTM’s intent to enhance the safety of TSFs across the globe.

Reliability-based design (RBD) of tailings storage facilities (TSFs) has increased in prominence over the past decade due to various industry and societal factors. One major component of RBD is the estimation of the probability of failure (PoF) of TSF embankments due to slope instability under normal operating conditions. PoF is commonly estimated through probabilistic limit equilibrium (LE) analyses utilising Monte Carlo (MC) simulations utilising a single random variable (SRV) approach. However, a major limitation of the SRV MC approach is often overlooked; namely, that material parameter probability density functions (PDFs) are sampled and applied to entire material domains within discrete trials whereas the PDFs are intended to characterise spatial variability within material domains. Ignoring this limitation can result in significant overestimation of PoF of TSF embankments, as tail probabilities that would otherwise be averaged out, were spatial variability to be considered, exert an oversized impact on PoF estimates.

This limitation can be overcome through probabilistic methods that consider spatial variability such as the random finite element method (RFEM) and random limit equilibrium method (RLEM). This paper presents analyses of an idealised TSF embankment using both an SRV MC approach and RLEM to illustrate the impact of spatial variability on the estimated PoF. The results are found to be consistent with the literature on non-TSF slopes, with PoF significantly overestimated for TSF embankments with a deterministic factor of safety (FoS) of 1.5 or 1.3, and realistic material variability.

There is a specific requirement part of the Global Industry Standard on Tailings Management (GISTM) related to the development of a Construction versus Design Intent Verification (CDIV) report during the Tailing Storage Facility (TSF) construction and operation. This requirement is needed to ensure that the design intent is implemented and met when site conditions or construction processes vary from the design assumptions. While no further details are provided for the development of the CDIV, the document is intended to maintain a clear history and knowledge base of the quality of TSF construction, support decision-making processes and manage risks.

Quality Assurance and Quality Control processes are generally well-established in the civil and mining industries. Nevertheless, there is no guidance on how to address or develop the CDIV. The literature review demonstrates, in general, that most guidelines or standards, directly or indirectly require some type of verification and documentation of construction changes. The latter documents do not provide enough guidance on what it really means and how it needs to be addressed.

This paper assesses and describes key features on how the current engineering practice understands and addresses design intent verifications in TSF construction and operation, also identifying key requirements and technical aspects needed to develop a robust verification programme. It also brings insightful opinions of key practitioners from around the world to understand the components and processes involved in the Construction Quality Assurance and its relationship with the design intent verification. The paper concludes that the best practice for the CDIV development is to be thoroughly aligned with Design Basis Report, design drawings and integrated within the change management system. Subsequently, it must become part of the Construction Records Report (CRR) together with a robust Construction Quality Assurance and detailed as-built documentation. Finally, the CRR feeds into Deviation Accountability Report (DAR), which feeds into the risk management program, for a complete risk-informed decision-making approach.