Reliability analysis of foundation settlement by stochastic response surface and random finite-element method

This paper presents a reliability-based analysis of strip-footing settlement by stochastic finite-element method (sfem). the stochastic response surface method (srsm) and random finite-element method (rfem) are used as two formulations of sfem. the elastic properties of soil are considered as spatial random variables and modeled as cross correlated log-normal random fields. random field discretization is done by karhunen-loeve (k-l) expansion. two programs were coded by matlab so as to take full advantages of its matrix operations, and in an illustrative example, it was shown that the results of srsm are close to rfem; however, the consumed time in rfem is at most 50 times longer than that in srsm. using the faster method, srsm, it is concluded that considering the spatial variability of soil parameters in stochastic analysis is necessary.