Risk-based approach for minimising tailings dam failures
Stability of tailings dams and tailings run-out
The total number of tailings accidents has reduced from over 50/decade in the 1960s-1980s to less than 30/decade in the 2000s to 2010s. However, the rate of significant tailings dam failures has not been reduced. The significant tailings dam failures the mining industry experienced in the recent history include: Merriespruit, South Africa, 1994; Los Frails, Spain,1998; Kolontár, Hungry, 2010; Mount Polley, Canada, 2014; and Samarco, Brazil, 2015. The conventional Factor of Safety approach itself may not be enough to further reduce the rate of significant tailings failures due to its limitations such as uncertainty and indirect link to consequence categories. For example, the same factor of safety could have a vast range of annual probability of failure for different tailings dams, and insufficient consideration of failure consequence.
This paper presents a risk based approach for design with the objective being to minimise tailings dam failures, especially those significant tailings failures with high consequence and low probability. This approach includes risk based design and risk management using bowtie tools. In the risk based design, quantitative analysis will be used to estimate the annual probability of failure and the tolerable failure criteria is linked to the consequence category. Bowtie analyses will be used to understand hazards and risks and identify and select independent controls that are effective in preventing and mitigating catastrophic failures. In this paper, examples of risk based design and bowtie analyses are presented and discussed. The examples include risk based designs for different failure models such as piping failure, overtopping and liquefaction et al.