Response surface methods for opitmization of a subsonic nacelle

Crawford Fredric Smith, Purdue University

Abstract

The design of low drag aircraft components is critical to achieving high levels of performance. Nacelles are significant contributors to the overall drag of an airframe. The research presented centers on an innovative approach for the design of low drag nacelles. Response Surface Methodology (RSM) coupled with Design of Experiments (DOE) is investigated for designing low drag subsonic nacelles using a calculus-based numerical optimization algorithm. The basic idea of RSM is to generate a small number of trial drag calculations at design points determined using DOE and to fit these results to a response surface using a quadratic polynomial. This response surface is used with an optimization code to determine the minimum drag design point. These methods are applied using a semi-empirical analysis code and a Computational Fluid Dynamics (CFD) code to predict the drag. Several design of experiments are studied in terms of their effects on the accuracy of the response surface fit. Also, the use of stepwise regression is examined to determine if the surface fits can be improved, over those obtained using full least squares regression. The use of trust regions is studied to assess the effectiveness of this approach to obtain minimum drag nacelle designs using fewer trial runs than DOE require. The results indicate that no particular DOE provided a significant advantage and are problem dependent. The use of stepwise regression did not improve the response surface fits. The use of trust regions showed a significant reduction in the number of trials needed as the number of design variables increased, although these results are also problem dependent.

Degree

Ph.D.

Advisors

Crossley, Purdue University.

Subject Area

Aerospace materials

COinS