Slope Reliability and Response Surface Method

Abstract

A method to analyze reliability of soil slopes using the response surface method is described. The soil slope is modelled and analyzed by a finite element code as in prevalent deterministic studies. The simulation which is usually expensive, is repeated a limited number of times to give point estimates of the response corresponding to uncertainties in the model parameters. A graduating function is then fit to these point estimates so that the response given by the finite element code can be reasonably approximated by the graduating function within the region of interest. The approximating function, called the response surface, is used to replace the code in subsequent repetitive computations required in a statistical reliability analysis. The procedure is applied to a sample problem in slope stability involving uncertain soil properties. It is shown that the slope stability statistics from the response surface is within 1–9% of the statistics based on a direct Monte Carlo simulation using the finite element code. The merits of such an approach to reliability analysis are examined.