A hybrid technique for the efficient reliability computation of large structures

Abstract

In the case of large structures, one obstacle encountered when the computation of reliability is attempted by most methods, is the presence of large numbers of random variables. Another issue is that computing the reliability of large structures typically requires conducting their structural analysis, which can be computationally time consuming, a large number of times. These concerns may render the reliability analysis computationally expensive or even unachievable. In this paper, a hybrid technique for computing the reliability of large structures is presented. In this technique, the most probable point of failure (MPP) is determined first using modified concepts of the Weighted Average Simulation Method (WASM). The WASM concepts are modified to handle the problem of large random variables present in large structures and also in order to find the MPP in a computationally more efficient manner. Once the MPP is determined, it is transferred into the standard normal space. Hence, the reliability index is calculated in closed-form in the standard normal space. The approach is tested on a truss bridge example.