Abstract: The advanced high-speed and high-lift Natural-Laminar-Flow (NLF) airfoil has played an important role in improving the aerodynamic performance of the new generation of High-Altitude Long Endurance (HALE) Unmanned Air Vehicles (UAV). However, shock waves and separation bubbles are likely to occur on the surface of this kind of airfoil, which are very sensitive to the aerodynamic characteristics such as fluctuation of Mach number and angle of attack. The aerodynamic shape designed with the traditional method has low robustness, and is thus difficult to be used in engineering practice. Although the Robust Aerodynamic Design Optimization (RADO) method is a very promising solution, it encounters the difficulty of large computational cost. In this paper, we study the key technologies affecting the efficiency of RADO and develop a sparse PC reconstruction algorithm based on the Adaptive Forward-Backward Selection (AFBS) method, greatly improving the efficiency of Uncertainty Quantification (UQ) and RADO. We also develop an efficient RADO method considering multi-parameter uncertainty, which solves the difficulty of the traditional airfoil design method that requirements for high-lift design, NLF design, and supercritical design cannot be met simultaneously. Finally, we successfully apply the proposed methods to design a class of robust high-speed NLF airfoils with significant characteristics. Results demonstrate that compared with the classical Global Hawk UAV airfoil, the airfoils designed with the proposed methods can provide better aerodynamic performance, larger low-drag range and more robust performance, which validate the effectiveness of the proposed RADO method and advantages of the proposed method over the deterministic optimization method.

Key words: high-speed natural-laminar-flow airfoil, natural-laminar-flow design, high-altitude long-endurance unmanned air vehicle, robust design, uncertainty quantification, polynomial chaos expansion