Reliability analysis of a vertical cut in unsaturated soil using sequential Gaussian simulation

This paper proposed a practical approach to reliability analysis of a vertical cut in unsaturated soil. This approach extends the application of the conditional random field to unsaturated soil. A real case study of a vertical cut was considered, and three boreholes were drilled to investigate the subsurface layers. The Sequential Gaussian Simulation (SGS) was used to generate a conditional random field considering the possible uctuation of soil properties between known data. The undersampled parameters were estimated by the cokriging method, while the kriging method was used to estimate other stochastic parameters. In order to verify the effciency of simulations, it was checked that all data were reproduced at their locations and the input semivariogram model was reproduced within acceptable uctuations. To predict the unsaturated soil behavior, the Soil Water Retention Curve (SWRC) was estimated by the physico-empirical method with the aim of determining suction stress for finite element stability analysis. The vertical cut was analyzed with and without considering suction. It was concluded that considering unsaturated condition shifts the mean of safety factor from unsafe to safe ranges. It was illustrated that the number of known data affected the construction of conditional random elds and led to different probability levels of failure.