A practical approach to combined probabilistic analysis of slope stability and seepage problems

This research is concerned with studying issues and providing tools to improve some of the aspects required to perform consistent and accurate reliability assessments of embankments. Probabilistic slope stability and probabilistic seepage analyses reported in the literature have generally been applied independently to special, simple problems. This study focused on the assessment of how practicing engineers could best use and combine practical, existing, analysis tools in a probabilistic framework that handles complex problems with a simple approach rather than simple problems with a complex approach.; One aspect investigated in this research is the effect of different deterministic and probabilistic models on reliability analysis. The effect of adopting a lognormal distribution for strength parameters was also investigated. The results showed that, for the problems considered, the shape of slip surface and the assumption of lognormally distributed soil parameters have a significant influence on embankment reliability. The effect of both deterministic and probabilistic methods were found to be not significant.; The uncertainty in pore pressure for end-of-construction analysis was studied in terms of the variability in pore pressure coefficient (r{dollar}rmsb{lcub}u{rcub}).{dollar} This parameter was modeled as spatial random variable and as a perfectly correlated random variable with both total and systematic variance. The two cases were shown to provide reasonable reliability results that compare well to the case of considering the variability in an actual r{dollar}rmsb{lcub}u{rcub}{dollar} grid values.; A practical and simple algorithm to locate the critical probabilistic slip surface was developed. This approach is based on systematically examining combinations of soil parameters other than the design (mean) values. The approach was examined and applied to an extensive parametric study on simple problems as well as several case-studies. It proved to provide a practical tool to locate the critical probabilistic surface. The algorithm also has the advantage of being applicable within the context of existing available slope stability computer programs. In many of the problems analyzed, the reliability index associated with the critical probabilistic surface was significantly lower than that associated with the critical deterministic surface.; The uncertainty in pore pressure u due to spatial variability of hydraulic conductivity k was investigated by assuming k as a spatially correlated random variable. The iterative finite-difference relaxation method was applied using spreadsheets both to generate spatially correlated values of k and to solve the steady-state flow equation. The spreadsheets provided simple tools and proved to produce reasonable results with considerably reduced computation effort. The study included the effect of the coefficient of variation of k, the scale of fluctuation of k, and the anisotropy in the correlation structure of k on the variability of u.