Probabilistic treatment of the sliding wedge

Abstract

Rock slope stability problems are often encountered in engineering environments. In nature, most variables governing rock slope stability problems such as the orientation of discontinuities, the strength parameters and the loading conditions are random variables. Therefore the safety factor itself is a statistic. This thesis analyzes the variation of all the parameters that enter into the standard stability analysis through the use of the Monte Carlo simulation technique. To meet this goal, a Windows based program EzSlide written in Visual Basic 4.0 has been developed. EzSlide follows the concepts and procedures of an earlier DOS program, PROSLIDE/GEOSLIDE. Several new routines are added however. Now all the strength parameters and the loading conditions can be modeled using a theoretical probability distribution. Five common probability distributions: the normal, the lognormal, the Weibull, the exponential and the triangular can be chosen by the user to model these variables. Three new routines for the optimization of the slope angle and the slope strike, and the slope height are added to evaluate their influence on the probability of failure. An application of the program using field data from a highway rock cut along TransCanada highway 17A is undertaken. The factors that influence the safety factor and the probability of failure are investigated. The water pressure and strength arameter JRC have the most significant impact on the safety factor and the probability of failure. It is found that the probability of failure changes dramatically with slope strike. The shape of the safety factor distribution is also examined. The probability distribution of the safety factor is found to be multi-modal. Multimodality is caused by the fact that discontinuities are not continuously distributed, but occur in sets and that only certain combinations of the discontinuities form sliding wedges.