Managing and mitigating the risks associated with mine waste facilities are important to mining operators and stakeholders. A tailings flow runout assessment is used to understand the risk of potential failure to downstream infrastructure and environment. Tailings are a common byproduct of mining operations. To retrieve the minerals or metal contained in mined ore, the ore is crushed so that subsequent processes can efficiently extract the needed material. After extraction, the residual material, called tailings, are then deposited in tailings ponds or used to make the pond containment structures called tailings dams. Tailings ponds contain a mixture of tailings particles and water, called slurry. If a tailings dam fails, it can flow with great force and cause significant damage to anything in its path. In addition to the potential cost in lives and infrastructure, the environmental damage can be severe.
FLOW-3D HYDRO‘s tailings model allows users to model the flow of hyper-concentrated mixtures of water and tailings. It includes the ability to model fine and coarse particles separately or together and with multiple initial layers of varying tailings concentrations. A supernatant water layer can also be included above the layers of non-Newtonian tailings material. Tailings flows exhibit highly non-Newtonian behavior with the viscosity depending on shear rate, tailings concentration and particle size. To model tailings flows accurately and efficiently, several advanced developments were required.
Tailings dams consist of tailings materials with a variety of grain sizes, depending on the mining process and tailings dam construction approach. Numerical models must also account for the varying layers of coarse and fine materials that make up the construction of a tailings dam.
Bimodal viscosity model
To model the flow of such structures when they fail, a bimodal viscosity model was developed. The features of this model include:
- Coarse (>40 micron) tailings model
- Methods include distance ratio, Thomas and Ovarlez correlations
- Grain size dependent
- Fine tailings model (<40 micron)
- Tabulated functions
- Concentration-dependent flow index
- Concentration-dependent yield stress
- Empirical correlations
- Tabulated functions
- Two-layer Herschel-Bulkley model in shallow water
- Viscous model
- Turbulent bed shear stress models (Strickler, Manning, more)
In meshing, only one vertical cell is used to model the topography and the flow. Since the tailings material is highly viscous, the viscous boundary layer is difficult to resolve without special treatment. A two-layer Herschel-Bulkley model in shallow water meshes was developed to resolve this issue.
Tailings dams are not only complex in their composition but are typically quite large, often measuring hundreds or even thousands of meters across. To accurately model over such large areas, both 3D and shallow water meshing is available. 3D models allow the flow of water and coarse and fine tailings to be modeled, including mixing and settling. Once the tailings have mixed sufficiently, the shallow water meshes can be coupled to the 3D meshes so that tailings runout flow over large areas can be modeled efficiently.
- Provides accurate modeling of variable composition tailings dams
- Multi-block, conforming meshing approaches available to capture detail in the areas of interest
Shallow Water Meshing
- Fast solutions over large domains
- Linked mesh blocks for greater meshing efficiency
Hybrid 3D/Shallow Water Meshing
- Coupled 3D and shallow water meshes in the same model
- Provides 3D resolution of variable tailings compositions with shallow water meshing in fully-mixed tailings regions
Application of the Tailings Model
The ultimate test of any modeling approach is the application of the model to experimental results or a real-world event. One of the most significant tailings dam failures in recent years is the Brumadinho failure which occurred in Minas Gerais, Brazil on January 25th, 2019. The significance of this failure lies in the number of lives lost and the amount of damage caused. The event provides a valuable case study since the construction materials were well documented. The times at which the tailings flow reached various locations along the valley are also well documented, which allows the simulated runout to be compared with observations. Below is a comparison of the simulated inundation compared with the observed inundation.
The timings for runout compare well with observations as shown below:
- Runout observation times (as reported in Lumbroso, et al. 20211)
- Canteen ~ 2 minutes
- Railway bridge ~ 10 to 12 minutes
- Paraopeba River ~ 1 hr 30 min to 2 hr 10 min
- Simulation results
- Canteen ~ 2 minutes 10 seconds
- Railway bridge ~ 9 minutes 30 seconds
- Paraopeba River ~ 2 hrs 18 minutes
 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