Multi-levels Kriging surrogate model-based robust aerodynamics optimization design method

Abstract

Robust design optimization has a great potential application in many engineering fields. In the conventional robust aerodynamics design optimization method, the main difficulty is expensive computational cost related to a large number of function evaluations for uncertainty quantification (UQ). To alleviate the expensive burden for UQ, two levels Kriging surrogate model was introduced. The first level is for the mean value and the second level is for the variances. Through the second level Kriging surrogate models, the method of Monte Carlo Simulation (MCS), which requires a huge number of function evaluations, can be effectively applied to the analysis of variance. Efficient Global Optimization algorithm (EGO) was employed to achieve the global optimized results. To validate the performance of the design method, both one-dimensional function and two-dimensional function were applied. Finally, robust aerodynamics design optimization was applied for a low-drag airfoil. The results show that the optimal solutions obtained from the uncertainty-based optimization formulation are less sensitive to uncertainties to small manufacturing errors.