Research on airfoil design space and the application of niching techniques in airfoil design

Design space of the RAE2822 airfoil is investigated. We find that there exists an optimum region in airfoil design space where considerable number of optimal airfoils locate. In this region, optimal airfoils have similar aerodynamic force coefficients, but distinctive geometries which result in different aerodynamic performances. To find multiple optimal airfoils in the whole design space, we introduce niching techniques into airfoil design and propose a gradient-enhanced niching particle swarm optimization algorithm (GE-NichingPSO). GE-NichingPSO employs a priority-ranked fitness sharing technique to enhance its ability to locate multiple optima. In order to speed up convergence, GE-NichingPSO is incorporated with a gradient-based algorithm. Numerical experiments show that GE-NichingPSO has obvious advantage in finding multiple optima over two famous niching PSOs. The proposed approach is used to solve the drag minimization of the RAE2822 airfoil in transonic viscous flow provided by AIAA aerodynamic design optimization discussion group (ADODG). Ten diverse optimal airfoils are obtained simultaneously, confirming the effectiveness of niching techniques in airfoil design.