Reliability based optimization using approximations with applications to multi-disciplinary system design

Abstract

A methodology for performing Reliability Based Optimization (RBO) using Limit State approximations for multi-disciplinary system design is presented. Two versions of the methodology are developed, one that does not employ decomposition and concurrent optimization techniques used in Multi-Disciplinary Optimization (MDO), and the other that does. The method allows for the use of both deterministic and probabilistic constraints. Finally, the proposed approach is implemented for two test problems, one a simple demonstration problem, and the second a variation of a control augmented structure problem used in a number of MDO studies. The methodology works successfully for the problems considered and is computationally more efficient than a traditional RBO. The MDO version of the RBO is shown to be more efficient than the non-MDO version, for one of the test problems.