Many large-scale engineering products and systems are composed of several smaller subsystems or disciplinary units. Multidisciplinary design optimization is a modeling and analysis framework that approaches the design process at the subsystem and component level and evaluates discrepancies or conflicts among the subsystems and components to provide for aggregation into a feasible system design. Significant sources of uncertainty are encountered during design, and the need to incorporate uncertainty and understand its impact on the design process is well documented; however, only limited studies have been conducted to characterize and propagate uncertainty within multidisciplinary design optimization frameworks. This research studies uncertainty within the context of multidisciplinary design optimization. Using a bi-level approach, uncertainty is incorporated in design variables, the operating environment, and customer requirements and propagated throughout the system during design optimization. Using a testbed of mechanical design problems, results will first be validated against traditional optimization approaches. Three design applications serve as test cases for verifying and validating the proposed multidisciplinary design optimization framework: combustion chamber of an internal combustion engine, racecar, and autonomous underwater vehicle. Compared to traditional optimization formulations, the proposed method successfully identifies designs that are robust to the observed uncertainty. |