Decomposition Techniques and Effective Algorithms in Reliability-Based Optimization

The common problem of an extensive number of limit state function calculations in the various formulations and applications of reliability-based optimization is treated. It is suggested to use a formulation based on decomposition techniques so the nested two-level optimization problem can be solved at a single level instead. The suggested formulations are evaluated from the 1. and 2. order optimality conditions. It is concluded that the one level formulation cannot be used without modifications in general but it makes a sound basis for a presentation of two practical solution approaches which reduces the number of limit state function calculations considerably. The ideas are illustrated in two examples.