Effects of random field modeling methods on slope stability

Soil properties exhibit an inherent spatial variability, its parameter changes from point to point and can not be neglected in probabilistic slope stability analysis. The commonly used methods to model soil random field in probabilistic slope stability analyses are variance reduction method, midpoint method and local average subdivision method. However, the effects of different random field modeling methods on slope stability has not been studied yet. In the paper, for a simple slope the three random field modeling methods together with first order reliability method are employed to conduct probabilistic slope stability analysis, and the results are compared with each other. The reliability analysis results from different random modeling methods are comparable with each other from practical point of view. The reliability results also show that the stability of slope is influenced by the spatial variability of soil properties, and the influence depends on the relative dimension of the scale of fluctuation and the length of critical slip surface.