Deterministic and probabilistic based analyses of reinforced soil slopes

Methods for deterministic analysis of reinforced slopes and current probabilistic analysis methods that have been applied to slopes with no reinforcement have been reviewed. On the basis of this review and additional studies, a new methodology for the probabilistic analysis of reinforced slopes has been developed. Using this new analytical method, both probabilistic and deterministic analyses of two reinforced soil slopes were performed and the results compared. The slopes chosen for analysis were designed using current design guidelines. Therefore, conclusions drawn from these analyses should be applicable to reinforced slopes currently being designed and built.;The analytical approach consisted of first analyzing each slope deterministically using limit equilibrium procedures adapted from analytical procedures for unreinforced slopes. For the deterministic analysis, the factor of safety was defined with respect to shear strength. Searches were conducted to find the shear surfaces with minimum factors of safety. Numerous local minima were identified. Next, each of the local minima with its corresponding shear surface was analyzed probabilistically, using the mean-value first-order reliability method (MFORM). From the probabilistic analysis, two critical failure modes with the lowest reliabilities were identified. Each of these two modes was then analyzed with the first-order reliability method (FORM). For the final portion of the dissertation, deterministic analyses for an instrumented, full scale embankment were performed. The results of these analyses demonstrated the limitations of using deterministic analyses to assess the stability of reinforced slopes and point to the need to adopt a probabilistic approach.;It was concluded that probabilistic methods have the potential to be better measures of the stability of slopes than current deterministic methods. Practical implementation of probabilistic methods in design will require more information on the variability of design parameters than is currently available. A robust approach for probabilistic analyses of reinforced slopes has been developed. The approach can be modified and improved as more data become available.