Tailings Model
Tailings dam failures are among the most catastrophic events in the mining industry. When a dam fails, a flood of hyper-concentrated slurry of water and mine waste can flow with tremendous force across large distances, causing loss of life, infrastructure damage, and severe environmental harm. Understanding the runout, including how fast, how deep, and how far the inundation will reach. Understanding the flow behavior is central to risk assessment andemergency planning.
The Tailings Model in FLOW-3D HYDRO simulates flows with a non-Newtonian rheology. It treats tailings as a two-species system of fine and coarse particles suspended in water, each with its own concentration, density, settling behavior, and contribution to the mixture viscosity. The shear-thinning behavior of a tailings slurry is captured using a bimodal viscosity model based on the Herschel-Bulkley framework, allowing accurate prediction of flow runout distance, inundation area, and arrival times at downstream locations.
Tailings dam breach simulated using a hybrid 2D/3D mesh in FLOW-3D HYDRO. Fine tailings concentration is shown as the flood spreads across the downstream terrain. The 3D mesh near the breach captures layered tailings with varied concentration and the ensuing vertical mixing; the 2D shallow water mesh covers the larger runout area efficiently.
How the Model Works
Tailings are modeled as a mixture of two particle species: fine particles and coarse particles, separated by a grain size threshold. Each species is transported independently within the flow, governed by advection, diffusion, and settling driven deposition. The mixture density and viscosity are computed in every cell from the local concentrations of each species, so the flow resistance and runout behavior respond dynamically as the mixture evolves.
Bimodal Viscosity Model
The viscosity of a tailings mixture depends on both the fine and coarse particle fractions and their interaction. Fine particles form a colloidal suspension whose rheology is governed by interparticle forces: at low concentrations the suspension behaves nearly as a Newtonian fluid, but as concentration increases, particle-particle interactions dominate and the mixture becomes strongly non-Newtonian, with a measurable yield stress and shear-thinning behavior. The fluid’s rheology is modeled using the Herschel-Bulkley model, where the Herchel-Bulkley parameters, yield stress, consistency, and flow index, all varying with local concentration.
Coarse particles do not directly contribute to the yield stress of the fluid, but at high coarse fractions they significantly increase the overall mixture viscosity. The bimodal viscosity model accounts for this by computing the fluid rheology from the fine particle concentration, then applying a coarse-fraction correction to the Herschel-Bulkley’s yield stress and consistency parameters.
Three empirical correlation approaches are available for defining the effect of coarse particles on viscosity: the distance ratio model, the Thomas correlation, and the Ovarlez model. Alternatively, the rheological parameters can be specified directly from laboratory measurements using tabular data, giving engineers full flexibility to match the model to site-specific tailings behavior.
Particle Species
Fine and coarse particle species can be activated independently or together:
- Fine particles form a colloidal suspension with the fluid. Their rheological parameters can be defined by tabulated data or empirical correlations fitted to laboratory measurements of the specific tailings material.
- Coarse particles are transported by the flow and settle under gravity. Their settling velocity can be defined by tabulated data or computed from standard correlations.
- When both species are active, the fine particle suspension acts as the carrier fluid for the coarse particles, and the mixture viscosity reflects the combined contribution of both fractions.
- Simulations can include Newtonian fluid behavior for areas with low particle concentrations, such as a supernatant water layer above the tailings layers or a river that the tailings breach flows into.
Meshing Approaches
Tailings dam failures and other non-Newtonain flows often span hundreds to kilometers, but the details of the breach and early flow are localized and three-dimensional. FLOW-3D HYDRO‘s hybrid meshing framework allows engineers to match the mesh type to the flow’s modeling requirements in each part of the domain.
- 3D meshing captures the three-dimensional flow structures within the dam and in the near-field immediately downstream of the breach, including mixing, settling, and vertical concentration gradients within the tailings flow.
- Shallow water meshing provides fast, efficient solutions over the large downstream runout area where the flow is fully mixed and the depth-averaged approximation is valid.
- Hybrid 3D/shallow water meshing couples both approaches in a single simulation, allowing 3D resolution of the dam breach and early flow with shallow water efficiency for the far-field runout.
Validation: Brumadinho, Brazil (2019)
On January 25th, 2019, the Corrego do Feijao tailings dam at Brumadinho, Minas Gerais, Brazil, failed catastrophically. The failure resulted in significant loss of life and is one of the most extensively documented tailings dam failures in recent history. The construction materials, dam geometry, and the times at which the tailings flow reached key downstream locations are all well documented, making it an ideal validation case.
The FLOW-3D HYDRO simulation used the documented dam geometry and tailings properties. The simulated runout compares closely with observed inundation extents and arrival times at three key downstream locations: the site canteen, the railway bridge, and the Paraopeba River.
| Location | Observed time of arrival | Simulated time of arrival |
|---|---|---|
| Canteen | 2 minutes | 2 minutes 10 seconds |
| Railway bridge | 10 to 12 minutes | 9 minutes 30 seconds |
| Paraopeba River | 1 hr 30 min to 2 hr 10 min | 2 hrs 18 minutes |
[1] Lumbroso, D., Davison, M., Body, R., and Petkovšek, G.: Modelling the Brumadinho tailings dam failure, the subsequent loss of life and how it could have been reduced, Nat. Hazards Earth Syst. Sci., 21, 21–37, 2021. https://doi.org/10.5194/nhess-21-21-2021
Applications
The Tailings Model is used in FLOW-3D HYDRO for:
- Tailings flow runout assessment for dam failure inundation risk analysis and emergency action planning
- Design and evaluation of breach scenarios for different dam construction methods and tailings compositions
- Comparative analysis of fine-dominated and coarse-dominated tailings behavior under failure conditions
- Supporting regulatory submissions and independent tailings dam safety reviews
- Sensitivity analysis of runout to tailings rheology, supernatant water depth, and breach geometry
Tailings runout sensitivity study in FLOW-3D HYDRO, comparing simulated flow depth and self-channelization in a numerical flume against physical flume experiments. Sensitivity studies like this provide a guide to model uncertainty and help quantify the effect of rheological parameter variation on runout predictions.