Approximations to non‐normal random variables in structural reliability analysis

Modern reliability analyses of large structural system designs require substantial computing times, and need to be repeated frequently in the process of design optimization. In the important case when underlying random variables are not normally distributed, the widely used approximation scheme by Rackwitz and Fiessler further increases the computational requirement by a substantial factor. This paper proposes a simple alternative approximation scheme which requires significantly less computation than the Rackwitz‐Fiessler algorithm, at an acceptable loss of accuracy.