Probabilistic analysis of consolidation problems using subset simulation

Abstract

In geotechnical structures, the failure probability (Pf) is usually calculated using the Monte Carlo simulation. This method is very time-consuming, particularly when dealing with small failure probabilities. To overcome this problem and as an alternative to MCS, the Subset simulation approach was applied to study the coupled two-dimensional consolidation. This method aims at performing a probabilistic analysis of the consolidation coupled with a heterogeneous soil with spatially varying Young's modulus (E). The probabilistic numerical results have shown that the probability of exceeding an admissible vertical displacement calculated by Subset simulation is very close to that calculated by MCS, but with a very substantial reduction in the number of simulations. In this study, the random field has been discretized into a finite number of random variables using the Karhunen-Loeve expansion. A parametric study to investigate the effect of the soil variability on Pf was presented and discussed. The effect of autocorrelation horizontal and vertical distances Lx and Ly of E on Pf has shown that increasing Lx and Ly increases Pf for both isotropic and anisotropic soils. In addition, this study has shown that Pf is more sensitive to Ly than Lx. Finally, the increase in COV(E) increases Pf.