Advanced Analysis of Structural Reliability Using Commercial Fe-codes

Recent trends in structural mechanics applications of Finite Element codes show an increasing demand for efficient stochastic analysis tools. While in principle virtually all stochastic problems can be solved by Monte-Carlo simulation, the computational cost may be prohibitively large. Therefore more effective approximations based on e.g. the Response Surface Method become quite popular. Alternatively, advanced Monte-Carlo methods such as importance sampling or adaptive sampling can be utilized. Depending on the class of problem to be analyzed, these methods may even be more efficient than the Response Surface Method. The paper focuses on two main issues, i.e. • The application of existing code for probabilistic analysis together with existing Finite Element Code for advanced Monte Carlo analysis • Presentation of a local-global approximation strategy for the response surface method. This so-called “weighted radii” approximation appears to provide the flexibility and accuracy needed for the reliability analysis of complex structures. A numerical example from structural analysis under static loading conditions shows the applicability of these concepts. In particular, the stochastics package SLang is utilized together with the FE-code ANSYS in the context of geometrically and materially nonlinear analysis.