Experiences with Applications of Importance Sampling in Structural Reliability Computations

This paper presents some experiences and observations relating to the use of importance sampling in structural systems reliability computations. These are presented with three examples, each highlighting a certain aspect of such an application. In all the examples, the components are deemed to follow the familiar bi-linear mechanical model and the random variables are independent. A failure path (or sequence) is defined here in terms of the failed components. Importance sampling density is shown to be clearly crucial to the effective use of importance sampling in structural reliability computations. Important modes and hence a subset of the basic random variables, play a dominant role in the selection of the sampling density.