Influence of the procedure for determining the characteristic values of strength parameters on the probability of failure of unsupported cuts

Estimating the characteristic value of strength parameters for design is an important matter in geotechnical practice. In this study, Latin Hypercube Sampling (LHS) is used to generate statistical meshes with the spatial variability of the friction angle and the true effective cohesion of the soil. For this, typical values of the coefficient of variation and of the horizontal and vertical scales of fluctuation of the friction angle and of the effective cohesion are adopted. Soil characterization is simulated in each of these statistical meshes. The characteristic values of the friction angle and effective cohesion are estimated following two procedures regarding the assumed knowledge of the soil properties. With these characteristic values, unsupported cuts are designed using the Load and Resistance Factor Design (LRFD) method. Slope stability analyses are performed with the software mechpy, using the Random Finite Element Limit Analysis (RFELA) and considering the above-mentioned meshes. Then, the probability of failure of the designed unsupported cuts is estimated using direct reliability analysis. The results show that adopting this procedure using simulation of site characterization has a substantial effect on the probability of failure