No abstract

Probabilistic analyses of slopes using Random Limit Equilibrium Method (RLEM) have been extensively reported in literature. However, in these types of analyses, the generated random fields are based on assumed values of horizontal and vertical correlation lengths. In practice, horizontal and vertical correlation lengths can be measured using CPT data and the data can be used to condition the generated random fields. Conditioning random fields reduces the level of uncertainty in the analysis and helps the simulations to render more reasonable results. In this study, the stability analysis of a simple slope is used to investigate the influence of conditional and unconditional random fields. To generate spatially variable fields, first, some artificial borehole data are employed to correlate the spatially variable friction angle field. Then, considering some typical values for the variability of the cohesion random field and the possible cross-correlation between the two fields, a couple of scenarios are defined to synthesize the spatially variable realizations of the cohesion field. Then, the results of cross-correlated conditioned and unconditioned random fields are compared. The results show that conditioning random field and considering the cross-correlation between soil input parameters significantly reduce the probability of slope failure.