The main research goal of this thesis is the development of computationally efficient Reliability-Based Multidisciplinary Design Optimization(RBMDO) methodologies under uncertainties. Designs of multidisciplinary system considering uncertainties may give profound potential for improving product quality and for reducing system failures. The design should synthesize multidisciplinary analysis for system integration, probabilistic analysis for considering uncertainty and optimization for performance increase.
To realize the research goal, three separate works are conducted. The first one involves Reliability Analysis of Multidisciplinary systems(RAM) in which a conventional nested architecture is decoupled to make a sequential loop. The key point of this work is that Global Sensitivity Equations(GSE) is synthesized with the Advanced First Order Method(AFORM). Since the AFORM is mainly driven by the sensitivity information, concurrent subsystem analyses are enabled in reliability analysis of multidisciplinary systems. The second one is incorporating a single-level Reliability Based Design Optimization(RBDO) approach with the trust region-SQP technique for a single system. The conventional RBDO has a double-loop structure. Whenever the outer loop searches probabilistic feasibility, the inner loop should perform the reliability analysis. The efficiency of RBDO is improved by decomposing the double-loop structure into a sequential one and incorporating an approximation strategy. However, approximation strategies may give difficulties in convergence by its poor approximation quality. To tackle the problem, the trust region-SQP strategy is proposed to improve the robustness of convergence. The third work is developing the RBMDO strategy. The proposed strategies for RAM and RBDO are synthesized in the RBMDO strategy. In the third work, the RBDO strategy using the trust region-SQP technique is applied to multidisciplinary systems. The Bi-Level Integrated System Syn...