Reliability Analysis of Strain-Softening Slopes Using the First Order Reliability Method (FORM)

摘要

The paper pertains to the slope reliability analysis under a probabilistic framework in strain-softening cohesive soils, using the first order reliability method (FORM). The performance function is based on the Bishop simplified method modified to take strain-softening into account in terms of the residual factor R(f) over a potential slip surface, estimated based on a progressive failure model proposed in the literature. The reliability analysis is performed on the surface of minimum factor of safety determined by using the sequential quadratic programming (SQP). The random shear strength parameters are assumed to follow normal distribution while the residual factor has been considered both as a deterministic parameter and a beta-distributed random variable. The results obtained for an illustrative example shows substantial reduction (21%) in the value of reliability index when R(f) is considered as a random variable with an assumed COV of 0.3. Results of FORM-based sensitivity analyses also reveal that R(f) has the most dominating influence on reliability and thus justifies its inclusion as one of the random variables. A parametric study, varying the assumed correlation coefficient between the random shear strength parameters from 0 to 1, shows that there is a maximum reduction of 16% in reliability index.