Introduction to Tailings Dam Stability and Slurry Flow

Mining tailings dams are among the largest engineered structures on Earth, designed to store the fine-grained waste byproducts (tailings) generated during mineral extraction. These impoundments must remain stable for decades, often under harsh environmental conditions and seismic loads. Failure of a tailings dam can release millions of cubic meters of toxic slurry, causing catastrophic environmental damage and loss of life. Recent high-profile failures—such as at Brumadinho in Brazil and Mount Polley in Canada—have intensified the demand for more rigorous stability analysis methods.

Traditional geotechnical stability assessments rely on limit equilibrium methods and simplified pore‑pressure assumptions. While these approaches are useful, they often fail to capture the complex, time‑dependent behavior of the slurry inside the dam. Slurries are multiphase, non‑Newtonian fluids that evolve over time as particles settle, consolidate, and desiccate. Computational fluid dynamics (CFD) offers a way to simulate these processes with high fidelity. ANSYS Fluent, a leading CFD platform, provides engineers with the tools needed to model slurry flow, pore‑pressure evolution, and stress distributions within tailings dams. This article provides a comprehensive guide to modeling slurry flow for stability analysis using ANSYS Fluent, covering fundamental theory, workflow, and practical considerations.

Understanding Slurry Flow in Tailings Dams

Slurries are suspensions of fine solid particles (typically < 200 µm) in water, often with a high solid concentration (40%–60% by mass). The flow behavior of these suspensions is governed by particle size distribution, mineralogy, and water chemistry. At high concentrations, slurries exhibit non‑Newtonian characteristics: they can be thixotropic (time‑dependent viscosity), shear‑thinning (viscosity decreases with shear rate), or have a yield stress that must be overcome before flow begins. These properties significantly affect how the slurry moves through the dam, how it consolidates, and how pore pressures develop.

Rheological Models for Tailings Slurries

To model slurry flow accurately in ANSYS Fluent, engineers must define a suitable rheological model. Common models include:

  • Bingham plastic – a linear shear‑stress vs. shear‑rate relationship after a yield stress is exceeded. Suitable for high‑concentration slurries that behave like a viscoplastic fluid.
  • Herschel‑Bulkley – a more generalized model that accounts for shear‑thinning or shear‑thickening after yield. Widely used for tailings because it fits experimental data well over a broad range of shear rates.
  • Power law – useful when yield stress is negligible, but often inadequate for settled or highly concentrated slurries.
  • Casson – historically used for mud and mining slurries, though less common today.

Experimental characterization (e.g., using a rotational rheometer) is essential to obtain the parameters for these models. Without accurate rheological data, CFD simulations can produce misleading results.

Settling and Consolidation

Over time, particles settle under gravity, forming a dense, compacted bed at the dam floor. This sedimentation changes the slurry rheology locally and affects the evolution of pore pressure. In many tailings dams, the slurry is deposited in thin layers (beaching), allowing water to decant and solids to consolidate. ANSYS Fluent can model this transient settling using either a Eulerian‑Eulerian multiphase approach with a granular solids phase or a Eulerian‑Lagrangian approach with discrete particles. The choice depends on the solids loading and the required level of detail.

The Role of CFD in Tailings Dam Stability Analysis

Stability analysis of a tailings dam involves evaluating the balance between driving forces (gravity, seepage pressure) and resisting forces (shear strength of the dam materials and the tailings themselves). Pore‑water pressure is a critical factor because it reduces effective stress, lowering the shear strength. CFD allows engineers to simulate the evolution of pore pressure throughout the dam as slurry is deposited, consolidates, and drains. This dynamic picture cannot be obtained from static limit‑equilibrium calculations.

ANSYS Fluent’s multiphase and porous media capabilities make it possible to treat the tailings as a fluid‑solid mixture moving through a deformable porous medium (the consolidated tailings and the dam structure). Coupling CFD with geotechnical finite‑element analysis (e.g., via one‑way or two‑way data exchange) provides even more insight, but Fluent alone can already output pressure fields, velocity profiles, and free‑surface locations that feed into conventional stability calculations.

Modeling Techniques in ANSYS Fluent

Selecting the right physical model in Fluent is the cornerstone of a successful simulation. The following subsections detail the most relevant approaches for tailings dam slurry flow.

Multiphase Flow Models

Tailings slurries are inherently multiphase. ANSYS Fluent offers several options:

  • Volume of Fluid (VOF) model – best when tracking a distinct interface between the slurry and air (or water), e.g., during beaching or in the decant pond. VOF captures free‑surface dynamics but does not model distinct particle phases; the slurry is treated as a single fluid with a user‑defined rheology.
  • Eulerian‑Eulerian model – treats each phase (water, solids) as interpenetrating continua with its own velocity, temperature, and volume fraction. This is the most rigorous approach for dense slurries where particle‑particle interactions are significant. It requires closure models for drag, lift, and granular pressure (e.g., the kinetic theory of granular flow). It is computationally expensive but provides detailed information on solids distribution.
  • Mixture model – a simplified Eulerian model that solves a single momentum equation for the mixture and uses algebraic relations for slip velocities between phases. Suitable when settling velocities are well characterized and the solids fraction is moderate.
  • Discrete Phase Model (DPM) – treats particles as individual Lagrangian elements in a continuous fluid. Appropriate for low solids loading (<10% volume) but impractical for dense tailings due to computational cost.

For typical high‑solid tailings dams, the Eulerian‑Eulerian approach or the mixture model (with appropriate drag and settling closures) is recommended. The VOF model can be used in conjunction with one of these models for free‑surface tracking.

Modeling Non‑Newtonian Rheology

ANSYS Fluent allows users to define non‑Newtonian viscosity via User‑Defined Functions (UDFs) or by selecting one of the built‑in models (Bingham, Herschel‑Bulkley, etc.). For a tailings simulation:

  1. Choose the appropriate rheological model based on experimental data. Store the parameters (yield stress, consistency index, flow index) in the material database.
  2. If the solids fraction varies spatially, use a UDF to make the rheological parameters depend on the local volume fraction of solids. This is critical because the slurry near the discharge point is less concentrated than the settled bed.
  3. Enable dynamic mesh or porous media zones for regions that have consolidated into a solid‑like cake.

A common pitfall is using a Newtonian viscosity approximation, which completely misses the yield‑stress behavior. A small yield stress (e.g., 5 Pa) can dramatically affect pore‑pressure dissipation and flow front propagation.

Porous Media and Consolidation

Once tailings settle, they behave as a porous medium with decreasing permeability as consolidation proceeds. In Fluent, you can define porous zones with directional permeability and a user‑defined function to update permeability based on local solid fraction. This allows simulation of the consolidation process: initially low solid fraction (high permeability) gradually transitions to a low‑permeability cake. The Fluent porous media model then computes the flow through the cake, providing pore‑pressure distributions needed for stability analysis.

Steps in the Modeling Process

A robust workflow ensures that the simulation is both accurate and computationally feasible.

Geometry and Mesh Generation

The geometry of a tailings dam includes the dam embankment, the basin floor, the decant pond, and the slurry discharge pipeline or spigot. In many cases, the dam grows over time; a 3D model can incorporate staged construction by using multiple regions or dynamic mesh motion. The mesh must capture critical features:

  • Free surface (typically refined near the slurry‑air interface).
  • Slurry discharge zone (high velocity gradients).
  • Dam interior and foundation (where pore pressures are computed).

Hybrid meshes (tetrahedral in complex regions, hexahedral in straight sections) are common. A grid‑independence study should be performed for at least two meshes (coarse/fine) to ensure that the computed pore pressures and flow fronts are not mesh‑sensitive.

Boundary Conditions and Material Properties

Key boundary conditions include:

  • Inlet – slurry inflow rate, solid volume fraction, rheological parameters (or a UDF to vary them).
  • Outlet – decant pond surface or toe drain. Typically set as pressure outlet with a fixed hydrostatic head.
  • Walls – dam face and basin floor: no‑slip is appropriate for the fluid phases, with shear stress computed from the slurry rheology.
  • Symmetry – used if the dam geometry is symmetric.

Material properties must be defined for each phase. For the solids phase in a Eulerian model, specify density, particle diameter, and granular viscosity models. The fluid phase (water) uses standard properties. Remember to specify an appropriate time‑step size: explicit schemes may require very small time steps (e.g., 0.001 s) to maintain stability in high‑yield‑stress regions.

Solver Settings and Simulation

For transient simulations of slurry deposition and consolidation:

  • Use the pressure‑based solver with the PISO or SIMPLE algorithm for pressure‑velocity coupling.
  • Enable the volume fraction equation for the Eulerian model, or set the VOF explicit scheme if tracking the free surface.
  • Select a second‑order upwind spatial discretization for momentum and volume fraction to reduce numerical diffusion.
  • Set the under‑relaxation factors carefully—yield‑stress flows can be sensitive to relaxation factors below 0.3.
  • Initialize the domain with a small air volume above the initial tailings bed. Use patch regions to define initial slurry height.

Run the simulation for enough physical time to capture the full deposition cycle (hours to days of real time). Check convergence at each time step: residuals should drop by at least three orders of magnitude.

Post‑Processing and Validation

After the simulation, ANSYS Fluent’s post‑processing tools can visualize:

  • Contours of pore pressure and effective stress.
  • Volume fraction of solids to show the beach profile and consolidation front.
  • Vector plots of slurry velocity, indicating zones of preferential flow that may lead to piping.
  • Free‑surface location for comparison with aerial photos or drone imagery.

Validation is essential. Compare computed pore pressures at various depths with piezometer readings from the actual dam. If the simulation matches field data within acceptable tolerance (e.g., ±15%), the model can be used to predict future behavior or to test design modifications.

Practical Considerations for Stability Analysis

The ultimate goal of the CFD simulation is to provide data for a quantitative stability analysis. The following outputs are particularly useful:

  • Pore‑pressure field – directly ingested into limit‑equilibrium software (e.g., SLIDE, GeoStudio) via a pressure map. High pore pressures in the tailings mass reduce effective strength and raise the risk of static liquefaction.
  • Shear stress distribution – Fluent can compute the shear stress exerted by the slurry on the dam floor and walls. Comparing these to the shear strength of the foundation materials indicates where failure is likely.
  • Free‑surface location – the height of the phreatic surface within the dam is a classic input for stability charts. An excessively high phreatic surface suggests inadequate drainage.
  • Slurry flow pathways – velocity vectors can reveal internal erosion (piping) if high‑velocity zones are found near the dam core or foundation.

Engineers often combine the CFD results with a finite‑element stress analysis (e.g., using Ansys Mechanical) to compute the factor of safety against sliding and rotational failure. This coupled approach provides a more complete picture than either method alone.

Case Studies and Applications

Several mining operations have used ANSYS Fluent to analyze their tailings management. For example:

  • A copper mine in Chile modeled the sequential deposition of tailings in a valley‑fill dam to optimize the spigot location and reduce the risk of over‑topping. The simulation showed that moving the discharge point 20 m upstream reduced the phreatic surface height by 3 m, improving the factor of safety from 1.2 to 1.5.
  • A gold mine in West Africa used a Eulerian‑Eulerian model with Herschel‑Bulkley rheology to assess the pore‑pressure dissipation rate in a thick‑walled tailings dam. The results led to the installation of additional horizontal drainage layers, which shortened the consolidation time from 18 months to 10 months.

While details are proprietary, such examples demonstrate the practical value of CFD in tailings dam engineering.

Limitations and Challenges

No simulation is perfect. Key limitations of the approach described include:

  • Computational cost – high‑resolution Eulerian‑Eulerian simulations with DEM‑coupled models can take days on a multi‑core workstation. Scaling to full‑scale 3D transient models may require cluster computing.
  • Rheological uncertainty – tailings properties change with mineral content, particle shape, and water chemistry. The simulation is only as good as the input rheology data, which may not be representative of all deposition phases.
  • Simplified geotechnical constitutive laws – Fluent does not natively model the stress‑strain behavior of the solid dam structure. Coupling CFD with geotechnical FEA is still a specialized task requiring careful data mapping.
  • Validation data – many mines lack comprehensive instrumentation (piezometers, inclinometers) to compare against. Without validation, the simulation remains a predictive tool with unknown accuracy.

Despite these challenges, the use of CFD for tailings dam stability has grown rapidly as computational power increases and as regulatory pressure demands more quantitative risk assessments.

Conclusion

Modeling the flow of slurries in mining tailings dams using ANSYS Fluent provides a powerful basis for stability analysis. By capturing the multiphase nature of the slurry, its non‑Newtonian rheology, and the evolution of pore pressure over time, engineers can identify failure mechanisms that simpler methods miss. The workflow—geometry creation, meshing, boundary conditions, solver setup, post‑processing, and validation—demands careful attention to detail, but the payoff is a deeper understanding of dam behavior and the ability to design safer, more reliable impoundments. As the industry moves toward more rigorous risk management, CFD tools like ANSYS Fluent will play an increasingly central role in ensuring that tailings dams remain stable over their entire service life and beyond.

For further reading on the underlying physics and best practices, consult ANSYS Fluent documentation, the International Council on Mining and Metals (ICMM) tailings management guidelines, and the Geoengineer.org tailings dam resources. Additional insight into slurry rheology can be found via RheoSense mining slurry applications.