Kriging Methodology for Surrogate-Based Airfoil Shape Optimization

Abstract

In this paper, a computational fluid dynamics optimization strategy-based surrogate model, which is used for the predictions of vertical aerodynamic force (Cl), is proposed. An airfoil shape optimization problem has been formulated, and the particle swarm optimization algorithm technique is used to solve the problem along with the inclusion of constructed surrogate model. This is used to replace the actual CFD algorithms. Ordinary kriging and Hammersley sequence sampling approach are applied to construct and enhance the effectiveness of the constructed surrogate model, respectively. The geometry of the airfoils (NACA 2411 and 0012) is described with the help of parametric section method, and the flow around the airfoils is solved by using the Panel method. The constructed surrogate model is more accurate and limits their estimation error within the order of 10−2–10−6. The computational time required to solve the formulated ASO problem, which will take 630 s while using a normal optimization scheme in a computer system which has 2GB of DDR3 RAM, 2.0GHz of processor speed, and 1MB of L2 cache memory, is drastically reduced to 56.7 s while using the surrogate-based optimization, and also, the optimized airfoils have improved Cl when compared to the original airfoils. The optimized airfoils were validated using experimental data.