Publications

Aerodynamic shape optimization based on computational fluid dynamics has the potential to become more widely used in the industry; however, the optimized shapes are often criticized for not being practical. Techniques seeking more practical results, such as multipoint optimization and geometric constraints, are either ineffective or too time consuming because they require trial and error. We propose a data-driven constraint for the aerodynamic shape optimization of aircraft wings that ensures the overall practicality of the optimum shape, with a focus on achieving a good low-speed performance. The constraint is formulated by extracting the relevant features from an airfoil database via modal analysis, correlation analysis, and Gaussian mixture models. The optimization results demonstrated that this approach addresses the thin leading edge issue that had plagued previous optimization results, and further analysis demonstrated that this data-driven constraint ensures good low-speed off-design performance without sacrificing the transonic on-design performance. The proposed approach can use other airfoil databases and can even be generalized to other shape optimization and engineering design problems.