Probabilistic Evaluation of Local Overstress on Slope Stability Problems

A methodology for the probabilistic evaluation of slope stability, capable of disclosing local reliability levels for a certain failure mode of an excavated slope, is proposed in the present work. Probabilistic slope evaluation is conducted upon a deterministically defined slip surface, exposed by the strength reduction method. Computations are performed by combining the probabilistic point estimate method with finite elements. Soil shear strength parameters, φ′ and c′, are treated as random variables of assumed constant variation, following normal distributions. The disclosure of remarkable variation of local slope reliability levels along its critical slip surface, highlights the potential for occurrence of local overstress, in cases where global slope stability is secured. High local probabilities of failure in the overstressed areas of the slope are manifested, despite its low global slope probability of failure. The noteworthy influence of cross-correlation between shear strength parameters, as well as the impact of their uncertainty level on the local probability of failure is discussed. The updated local and global reliability levels, during progressive failure propagation, are also obtained by the model.