Probabilistic assessment of rock slope stability using response surfaces determined from finite element models of geometric realizations

A methodology is developed for probabilistic rock slope stability assessment using numerical modelling that incorporates statistical analysis of the variability of joint set geometric parameters. Each probabilistic input parameter is substituted by its two point estimates. Half-factorial and central composite designs are implemented to obtain a minimum number of representative slope realizations to model. The output from the numerical models is used to construct mathematical prediction models or response surfaces. A response surface can be used to predict the factor of safety of arbitrary realizations without further numerical modelling and can be used to determine the probability of slope failure.