Comparison of Robust Optimization Methods Applied to Hypersonic Vehicle Design

Abstract

The use of various robust optimization methods to solve engineering problems with environmental parameter uncertainty was investigated. Comparisons were made between a novel multi-objective-based robust optimization formulation and conventional robust regularization-based and aggregation-based methods. The results, performance, and philosophies of each method are discussed. A hypersonic vehicle design problem was used as a test bed for the methods presented here. A focus was put on studying the effect of uncertain roughness-induced boundary-layer transition locations on vehicle controllability. The robust regularization results show that a flat wedgelike vehicle design is best for a worst-case scenario, whereas a pyramidal-shaped vehicle design minimizes the expected detrimental effects on vehicle controllability. The analysis proved that the multi-objective robust optimization method was able to identify these two types of results within an overall set of results while also capturing tradeoffs within the design space.