Use of a physically based erosion breach model for water and tailings dams within a probabilistic framework

November 19, 2024

Breach analysis requires the selection parameters which describe the geometric dimensions and rate of progression of a breach. Parametric equations derived from historical dam failures are often used to select dam breach parameters. While parametric equations are readily applied, their efficacy is limited by the size and quality of the databases from which they are developed and their ability to allow for explicit consideration of material properties or represent complex embankment geometry. Physically based breach models allow for a more complete assessment of dam material properties, geometry and zoning. Several physically based breach models are available of varying formulation and complexity. The models require a larger amount of input data than parametric equations, and the inputs are subject to uncertainty. This paper presents a method for the estimation of breach parameters using a simplified physically based dam breach model within a probabilistic framework. The impact of the uncertainty associated with input parameters is considered using a Monte Carlo approach where statistical distributions for input parameters are defined and randomly sampled. Two case studies are presented which demonstrate that the approach can be used to replicate observed parameters from a historical failure and be employed in design situations where predictive modelling is required.

Palmen, L. and N. Kent. 2024. “Use of a Physically Based Erosion Breach Model for Water and Tailings Dams Within a Probabilistic Framework,” in Proceedings of the 2024 ANCOLD Conference: Never the Same Dam Twice. Adelaide, SA, Australia, 11-14 November 2024.