While the thought of a tailings dam failure is frightening for all stakeholders, even the safest engineered dam might fail. Dam break analyses (DBA) play a critical role in planning for these unlikely and undesirable emergencies. Given the uncertainties associated to these types of studies, mainly when it comes to tailings storage facilities (TSFs), any efforts to understand their implications tend to be widely supported in the mining industry.
Among the different sources of uncertainties, release volume plays a key role on the variability of consequences during a potential failure of a TSF. Instead of using deterministic methods, which typically estimate one single release volume for a DBA, SRK is developing a method that assesses the variability of the input parameters for release volume. Based on the cone-depression method, a generic failure surface has been defined, whose parameters are not treated as fixed values but rather as what they truly are: random variables.
These variables have been assigned probabilistic distributions based on the analysis of records of past failures. A Monte Carlo simulation process is then carried out, with the primary objective of obtaining the cumulative frequency distribution of release volume. As a result, the range of plausible values has an exceedance probability associated to it under the assumptions of the proposed method.
By picking up release volumes with different exceedance probabilities, different flood simulations can be modelled, thus providing the opportunity to plot incremental inundation maps. These allow analysts to assess the incremental downstream consequences instead of relying on one single scenario as in the traditional deterministic approach. Among other benefits of this method, emergency and preparedness plans can be better refined, and the TSF design can be enhanced.
The method that has been developed can be modified following the same logic, with the ultimate objective of defining a cumulative frequency distribution of the credible release volumes. The study of the uncertainty of other parameters involved in a DBA continues to be a subject of concern for SRK. While it is impossible to predict the specifics of an unfortunate event like a dam failure, estimating the range of credible consequences is wiser than just trusting one deterministic scenario.
