Coupling suspension in multiobjective multidisciplinary design optimization

The design of systems is increasing in complexity as a result of advances in the tools available for the designer. The design process is usually broken down into smaller, more manageable subsystems. The specialists from different disciplines work on the subsystem objective. The outputs of different subsystems are interlinked with each discipline requiring information from other subsystems for its evaluation. In practice, design problems have more than one objective, which might be conflicting in nature. Rather than having one solution, the output of such design problems is a set of solutions. The designer has to select a solution depending on his preference. Problems in practice have a large number of couplings between the subsystems, which makes design process computationally expensive. The designers are always working on deadlines and are looking to get to solutions as fast as possible. Coupling suspension is a method where some couplings are removed with little effect on the solution accuracy. Previous work has proved the benefits of coupling suspension in case of single objective multidisciplinary design optimization (MDO) problems. No work has been done to investigate and understand the impact that coupling suspension has on multiobjective MDO problems. Previous researchers have used genetic algorithm to solve coupling suspension problem. In this work branch and bound method is used to give consistent results. In this thesis, a modified MDF method is developed and tested to implement coupling suspension analysis in multiobjective MDO problems. The method is compared to Hybrid MDF/IDF method to compare its trade-off in terms of accuracy and efficiency. The method provides time savings in low volatile systems and at lower allowable errors with high volatile systems.