Probabilistic analysis of a tailing dam considering spatial variability of soil properties

Non-circular Random Limit Equilibrium Method (RLEM) uses a combination of non-circular limit equilibrium method, 2D random field theory, and Monte Carlo simulation. Non-circular RLEM has the ability to seek out the weakest failure path in a slope stability analysis using optimization techniques such as Surface Altering Optimization, while considering spatial variability of soil properties. This study uses data available from the Mount Polley Tailings Dam breach in order to 1) compute soil property correlation lengths, 2) calculate deterministic factor of safety (FS), 3) calculate probability of failure (PF) using a Single Random Variable (SRV) probabilistic analysis, and 4) calculate PF using the RLEM, using a search of potential composite circular and non-circular failure geometries. The results from both methods are compared including the influence of the choice of coefficient of variation (COV) of soil input parameters. It is shown that non-circular analyses result in higher PF values when compared to composite circular search methods. Also, considering spatial variability of soil properties in the vertical direction reduces the calculated probability of failure from composite circular or non-circular RLEM analyses when compared to the SRV approach

Keywords:

probabilistic analysis , dam , spatial variability , c-ϕ soil , non-circular RLEM