Response Surface Methodologies for Low-Boom Supersonic Aircraft Design Using Equivalent Area Distributions
Trent W. Lukaczyk, Francisco Palacios, Juan Jose Alonso · 12th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference and 14th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference · 2012
The use of response surface models (also known as surrogate models) to describe the performance of aerospace systems within a given design space is by now fairly well established in the MDO community. Despite such widespread use, surrogate-based optimization techniques can incur signicant computational expense when high-dimensional spaces must be modeled. In this paper we describe an adaptive, constrained optimization methodology that is based on (a) Gaussian Process Regression (GPR), (b) the use of gradient information, inexpensively obtained via adjoint methods throughout the design space, (c) an adaptive sampling technique based on the notion of expected improvement, and (d) the treatment of nonlinear constraints (within the adaptive sampling loop) so that the number of unnecessary infeasible samples is signicantly reduced. The methodology is demonstrated using a number of analytic examples as well as with CFD-based low-boom airfoil and aircraft design problems using constraints based on target equivalent area distributions. This work is part of a larger eort (jointly with Lockheed Martin) to develop a multidisciplinary analysis and optimization framework for the high-delity design of lowboom supersonic aircraft within the NASA N+2 Supersonics eort.