ABSTRACT
Probabilistic analyses are often employed in slope stability analyses to account for the uncertainty associated with local variations in subsurface conditions. Practically, the consideration of spatially and stochastically varying soil properties is commonly achieved in combination with limit equilibrium methods to calculate the factor of safety. However, such analyses require advanced searching methods to locate critical slip surfaces with the minimum factor of safety. Surface altering optimization (SAO) is a novel searching method which minimizes the factor of safety by altering the geometry of a given slip surface using spline curves in 2D. By altering the points on the spline curves, the slip surface can be fine-tuned to trace the critical failure path in a slope. It is a local search algorithm that when combined with a coarser global search method, forms a powerful hybrid optimization technique to identify the critical slip surface. In this paper, a case study of the James Bay Dyke is presented whereby a probabilistic analysis is conducted on a slope with spatially varying shear strength. It was found that the use of random limit equilibrium method (RLEM) and SAO combined with a global search, provides accurate results while curtailing computational effort. It was also found that longer correlation lengths (coarser sampling resolution) can result in large regions of reduced material strength, leading to multiple failure modes and shorter slip surfaces. The results, which highlight multiple failure modes, were verified using the shear strength reduction (SSR) method.
