Slope reliability analysis using length-based representative slip surfaces

Abstract

This paper develops an empirical approach to slope reliability considering 2D spatial variability of soil properties with length-based representative slip surfaces. The correlation coefficient between factors of safety for two slip surfaces is analogous to that between two slip surfaces which is quantified based on the lengths within different elements of the slip surface. A large number of potential slip surfaces are categorized into separate slip surface groups within each of which the correlation coefficient between any two slip surfaces is higher than the threshold value. A slip surface with the minimum factor of safety is considered to be the representative slip surface for each of slip surface groups. The factor of safety is minimized among all the representative slip surfaces, and the minimum one is regarded as the output for each set of random samples generated by Monte Carlo simulation (MCS). The methodology is validated through comparison with other methods. It is found that slope failure probability, p f, increases as the scale of fluctuation, λ y (in vertical direction), and λ x (in horizontal direction) increase. The effect of non-circular failure mechanism was studied at different λ x and λ y values. Significant larger failure probabilities are provided with non-circular failure mechanism than with circular failure mechanism. For 2D spatial variability case, two orders of magnitude larger failure probability is noticed.