Stochastic finite element methods for the seismic response of soils

Some of the available stochastic finite element methods are adapted and evaluated for the analyses of response of soils with uncertain properties subjected to earthquake induced random ground motion. In this study, the dynamic response of a soil mass, with finite element discretization, is formulated in the frequency domain. The spectral density function of the response variables are obtained from which the evaluation of the root-mean-squared and the most probable extreme values of the response are made. The material non-linearities are incorporated by using strain compatible moduli and damping of soils using an equivalent linear model for stress-strain behaviour of soils and an iterative solution of the response. The spatial variability of the shear modulus is described through a random field model and the earthquake included motion is treated as a stochastic process. The available formulations of direct Monte-Carlo simulation, first-order perturbation method, a spectral decomposition method with Neumann expansion and a spectral decomposition method with Polynomial Chaos are used to develop stochastic finite element analyses of the seismic response of soils. The numerical results from these approaches are compared with respect to their accuracy and computational efficiency. (C) 1998 John Wiley & Sons, Ltd.