Completely Automated System for Evolutionary Design Optimization with Unstructured Computational Fluid Dynamics

We have fulfilled a completely automated system of evolutionary design optimization with unstructured computational fluid dynamics. Until now, we cannot automatize aerodynamic design optimization to deal with geometry with high degrees of freedom. However, we achieved to simply iterate large-scale (it takes huge time to evaluate objective functions) and real-world evolutionary multiobjective optimizations. As a result, we efficiently obtained design knowledge for a next-step problem by applying the system to the design problem of a booster stage for two-stage-to-orbit. Moreover, this study yields the hypothesis regarding the appropriate algorithm of evolutionary computation for not only mathematical benchmark but also large-scale real-world problems.