Aerodynamic Design of a Flying Wing Using Response Surface Methodology

Abstract

The design of a subsonic e ying wing with maximized lift is considered. A novel method of surface generation, known as the partial differential equation (PDE) method, is used to parameterize the e ying wing. Because this method is able to parameterize complex geometries in terms of a small number of shape parameters, the computational costs that are normally associated with optimal aerodynamic design are dramatically reduced. The lift data, which are estimated using a low-order potential e ow panel method, are subject to numerical noise and yield, therefore, a design space that contains many spurious noise-induced local maxima. Standard methods of local optimization are severely hampered by this noise and may converge prematurely in nonoptimal plateau regions where the variation of the lift is small relative to that of the noise. To combat this inefe ciency, techniques from response surface methodology (RSM) are used to construct smooth analytic approximations of the noisy lift data, which can be optimized successfully. This combination of the PDE method and RSM results in a design approach that is both efe cient and robust. Three e ying-wing design problems areinvestigated, and the results are presented.