A numerical optimization chain combining computational fluid dynamics and surrogate analysis for the aerodynamic design of airfoils

Abstract

The aim of this study is to present a fully automated computational fluid dynamics-based optimization chain, implementing a radial basis function meta-model combined with an improved Latin hypercube design of experiments strategy. The objective function (aerodynamic performance) is evaluated through computational fluid dynamics calculations by using the commercial code ANSYS-CFX. The optimization strategy is hybridization between a stochastic bi-objective non-dominated sorting genetic algorithm and a gradient-based method known as modified method of feasible direction to get benefit from their combined capabilities. The testing of this optimization chain consisted in finding the optimal operating conditions of an airfoil NACA0012. This methodology may help to a great extent in the better exploration of the design space and to guide numerical and experimental studies to the potentially optimal design parameters.