Improved knowledge-based clustered partitioning approach and its application to slope reliability analysis

Abstract

A knowledge-based clustered partitioning (KCP) approach is improved to determine the reliability index and probability of failure of a rock slope. The Nataf transformation is adopted to transform the correlated non-normal random variables involved in the KCP approach into independent standard normal variables. An improved KCP technique is proposed to search the design point and calculate the reliability index. Two illustrative examples are presented to demonstrate the capability and validity of the proposed approach. The results indicate that the improved KCP-based reliability method can be applied to evaluate the reliability of rock slopes involving multiple correlated non-normal variables accurately and efficiently. Its accuracy is shown to be higher than that of the traditional KCP using the bisection method, and it is much more efficient than Monte Carlo simulation. The improved KCP-based reliability method is especially suitable for dealing with an implicit performance function with a large number of random variables, which is often involved in slope reliability analysis.