Data-Driven Aerodynamic Shape Design with Distributionally Robust Optimization Approaches
Please login to view abstract download link
Optimization under uncertainties remains one of the major challenges for aerodynamic design optimization. Uncertainty has the potential to render an optimal shape design worthless, even if obtained using sophisticated numerical approaches, as their conclusions are not realized in practice due to inevitable variations in problem data. On the other hand, the increasing available data allows an unprecedented insight into the uncertainties. For example, it is nowadays not difficult to acquire data about various key flight conditions (Mach number, altitude, etc.), but the incorporation of these data in design optimization is challenging. In this work, we formulate and solve data-driven aerodynamic shape design problems with distributionally robust optimization (DRO) approaches \cite{delage}. Building on the findings of the work \cite{gotoh}, we study the connections between a class of DRO and the Taguchi method in the context of robust design optimization. Our preliminary computational experiments on aerodynamic shape optimization in transonic turbulent flow show promising design results.
