Computational modeling of tailings dam failures

Over 250 people died as a result of catastrophic damage following a tailing dam break in Brazil in 2019. The incident happened after a subsequent incident of a tailing dam break in Brazil three years ago. 

As per a report [1] by Reuters, more than 1700 dams hold mine waste, and around 690 are classified as high risk in terms of the probability of dam breach and consequent damage to life and property. 

Mine waste usually called tailings; consists of several components including fines, coarse particles, minerals, metals, chemicals, and water. Due to the presence of various hazardous contaminants in them, these are stored at isolated locations where the risk of large-scale damage is less, yet due to various reasons dams crumble and cause tremendous damage to infrastructure, the environment, and human life.

Dewatering the tailings streams to produce a homogenous non segregating thickened form is a solution proposed by researchers to reduce the damage. This may change their rheological properties and flow behavior, thus reducing the destructive nature of these flows.

Computational Modeling using OpenFOAM is an efficient way of reducing tailings dam disasters and consequent damage 

As much as there may be theoretical solutions proposed for minimizing tailings dam failures and consequent damages, trust can be gained only through on-site experimentation - something that is not possible for real-case catastrophic events. The only solution then left is full-scale computational modeling of such probable events!

OpenFOAM for modeling tailings dam failures

OpenFOAM is the computational software for modeling tailings slurry flows and predicting impacted area, runout length, front velocity, and the energy contained in the flow. OpenFOAM stands for open-source field operation and manipulation. The software is, of course, open-source; fully 3-dimensional; truly multiphase i.e not limited to two phases; can model complex physics including multiple fluids and particles; supports unstructured meshing for accurately capturing the complex terrains; scalable on the cloud; and, comes with an absolute state-of-art data visualizer.

Red Chris mine: A case study

At Paanduv Applications Private Limited, we conduct case studies involving real tailings sites. In this case study, we picked the Red Chris mine because of its beautiful description along with the stunning site photography provided by the Narwhal [2].

The Red Chris mine is majorly owned by Newcrest Mining Limited. The site is situated 1,700 km from Vancouver and primarily copper and gold mining happens here. The tailings dam wall is a whooping 105 m high.

We extracted the real site terrain from Google Maps and got started with the computational modeling of the Red Chris mine tailings dam failure. The goal was to predict the effects of the partial tailings dam wall breach on the downstream areas. The simulation was run for a 5-minute real-time scenario post-breach.

We used the Herschel-Bulkley model, one of the many non-Newtonian models available in OpenFOAM. Based on the data reported by Pirouz et al [3] involving typically thickened gold tailings slurry rheological experiments, we were able to find a representative set of Herschel-Bulkley parameters for gold tailings. The parameters change significantly with the value of solids percentage in the slurry. We used the parameters corresponding to a solids percentage of 55.2%. 

Although there are different aspects of relevance for tailings dam design and construction, we specially investigated the change in the total kinetic energy of the gold tailings flow downstream followed by a partial but sudden dam wall breach. The total kinetic energy is one single parameter that gives much information about the havoc such an event may cause to the downstream life and property.

We also compared this event with another event with exactly the same conditions except that water was filled in the tailings pond in place of gold slurry.