Importance of Uncertainty Representation and Design Exploration Method for Robust Aerodynamic Optimization

Abstract This study examines the impact of uncertainty propagation and design exploration techniques on robust airfoil optimization processes. The RAE2822 airfoil geometry is selected as the case study and parameterized using the class/shape transformation (CST) method. The functional relationship between design inputs and outputs is modeled using a computational fluid dynamics (CFD)-driven surrogate model based on Gaussian process (GP) regression. Uncertainty calculations are performed using Monte Carlo sampling (MCS) and polynomial chaos expansion (PCE). In addition, the effects of different design exploration methods are analyzed using deductive design exploration (DDEM) and inductive design exploration (IDEM). The results indicated that the DDEM method is sensitive to the choice of uncertainty propagation technique, leading to variations in the optimization results. In contrast, solutions obtained using the IDEM method are found to be independent of the uncertainty propagation approach and more robust than those from the DDEM method. Furthermore, the robustness metric provided by IDEM makes it a practical and efficient method for robust airfoil optimization processes.