Surrogate-based optimization applied to benchmark aerodynamic design problems

The AIAA aerodynamic design optimization discussion group has defined six benchmark aerodynamic design problems for reseachers to present and compare their optimization methods and results. This work applies a surrogate-based optimization (SBO) method based on kriging model and expected improvement (EI) to three benchmark cases, for demonstrating the SBO-type method’s capability and applicability of expensive aerodynamic shape optimization. These cases include drag minimization of a NACA0012 airfoil and a NACA0012-based rectangular wing in inviscid flow, and a common research model wing in viscous flow. An adaptive design space technique is used to reduce the computational cost of the optimizations. For the first NACA0012 airfoil case, using CST parameterization method, the SBO can get comparable result with gradient-based method. For the rectangular wing case, with the twist distribution parameterized by B-spline, a lift distribution close to elliptical is obtained. For the optimization of CRM wing using FFD parameterization method, the drag coefficient is reduced by 4 counts at present stage and a practical design is obtained. The results indicates that the results of SBO method is comparable to that of other optimization methods, and has been matured to a stage that can be used to efficiently address expensive engineering design problems.