Reliability Based Multidisciplinary Design Optimization

Design for engineering system is often confronted with multidisciplinary austerity challenge due to involving multiple disciplines for exploiting complex products, such as aerocraft, automobile, whose system design is complicate and time consuming, high cost. Hence, multidisciplinary design optimization(MDO), aimed to complex engineering system, is emerged as the times require on the basis of design optimization, which can shorten development time, reduce product cost and raise quality. Almost every practical engineering problem contains some random factors which can lead to random fluctuation, such as loading condition, material characteristics, accuracy of simulation models, geometrical feature, manufacturing tolerance and service condition. Under a multidisciplinary design environment, the uncertainties of one discipline may be propagated to another discipline through the linking variables and the final output from the integrated multidisciplinary system has an accumulation of the uncertainties from the individual disciplines. Uncertainty will affect ultimately performance if it is not handled well. Therefore, typical MDO not only searches the best collaboration between system optimization and subsystems optimization, but also considers uncertainty leading to impact for design process and results.Safety factor is use to handle with random factors for traditional design methods so that it leads to excessive design and can't describe the impact of all kinds of random factors as well as actual safety margin. In recent years, uncertainty and randomness can be dealed with by stochastic model and probability analysis with the development of quality engineering. When these techniques are applied in the framework of optimization design, they can search feasible design plan not only in theory, but also in practice.To find the approaches for managing uncertainty, MDO under uncertainty is studied by analyzing the sources, model and analysis methods of uncertainty and reliability based MDO is focused in this paper. iSIGHT is used to perform traditional optimization and reliability based optimization for two examples to analyze the impact of uncertainty. Collaborative reliability analysis method under multidisciplinary structure and a new reliability based MDO approach are proposed aimed to the requirement of efficiency, applicable of reliability based MDO. The former inherits advancements of CO, and has accurate, efficient computation because of employing simulated annealing algorithmic in system level and response surface instead disciplinary optimization. The latter, which is a single loop, can decouple MDO and reliability analysis by using decomposition strategy to avoid computation complexity. Besides, it shifts location of constrains in design space, which can obtain optimal rapidly and increasecomputation efficiency.