Surrogate Approaches for Aerodynamic Section Performance Modeling

Abstract

The use of surrogate models (response surface models, curve fits) of various types (radial basis functions, Gaussian process models, neural networks, support vector machines, etc.) is now an accepted way for speeding up design search and optimization in many fields of engineering that require the use of expensive computer simulations, including problems with multiple goals and multiple domains. Surrogates are also widely used in dealing with uncertainty quantification of expensive black-box codes where there are strict limits on the number of function evaluations that can be afforded in estimating the statistical properties of derived performance quantities. This paper compares and contrasts a wide range of surrogate types on an aerodynamic section data set that allows for design variation, manufacturing uncertainty, and damage in service, all solved with a high-quality industrial-strength Reynolds-averaged Navier–Stokes solver. This paper examines speed of training and model quality for different sizes of problem up to one where there are 26 input variables and nearly half a million CFD results in the available data set.