A localized probabilistic approach for slope stability analysis

A Localized Probabilistic Approach for Slope Stability Analysis, LPSLOPE, is introduced. In LPSLOPE, the localized probabilistic analysis of a slope was solved by combining the local joint probability model of the in-situ strength parameters and the deterministic slope model. The measure of safety is now expressed in terms of the “Probability of Failure” at a certain confidence level.; Input to LPSLOPE consists of a localized probabilistic model of material strength parameters and a critical boundary line. The in-situ strength parameters are given in terms of cohesion (c) and internal friction angle (&phis;), therefore these strength variables are randomized in the bivariate joint probability model. The geostatistical conditional simulation, called Sequential Gaussian Cosimulation (SGCOSIM), was introduced to achieve the localized probabilistic characterizations of the in-situ strength variables. The bivariate normal distribution was selected to model the joint variation of c and &phis;.; The three dimensional slope stability analysis method of columns, called Hovland's method, was adopted as a deterministic slope model for this analysis. The back calculation of the Hovland's slope model made on the failure surface generates a series of the necessary strength values (pairs of c and &phis;). This is the strength required in order to maintain the design safety criteria within a slope.; Outputs from LPSLOPE are the local probabilities of failure (the probabilities of failure of columns) and the global probability of failure (the probability of failure of a slope) at a desirable confidence level. These are the conditional probabilities of failure. They are conditioned to the available information, the state of knowledge of a slope and the design parameters.; LPSLOPE was applied to the slope at an open pit coal mine in Lampang, Thailand. The objectives were to illustrate the capability of the model in handling a variety of problems, and to examine the sensitivity of the local and global probabilities of failure to the changes in the design parameters. The results have shown that LPSLOPE is a powerful method for drawing a detailed and realistic picture of slope stability conditions. The spatial variability of the strength variables, which is considered to exist in the slope area, played an important role in slope stability, and should be included in the analysis model. Moreover, the sensitivity analyses allow the adjustment of the design parameters corresponding to the risk acceptable in the final slope design. However, the most important conclusion that can be drawn from this research is that LPSLOPE can be applied at the stage when only sparse sample data is available within a slope. (Abstract shortened by UMI.)...