Leveraging Multi-Fidelity Aerodynamic Databasing to Efficiently Represent a Hypersonic Design Space
Kevin Quinlan, Jagadeesh Movva, Elizabeth V. Stein, Ana Kupresanin · ASCEND 2021 · 2021
View Video Presentation: https://doi.org/10.2514/6.2021-4245.vid Predicting flight-performance of a hypersonic flight vehicle requires characterization of the complex aerothermodynamic phenomena present across the flight envelope. A stable and accurate trajectory simulation requires an aerodynamic database that covers a wide variety of flight conditions (aka using statistical terminology, a large design space). When preparing such an aerodynamic database, computational cost becomes critical to conserve limited computational resources while ensuring the quantities of interest are well resolved. In this study, a computationally efficient way to generate an aerodynamic coefficient database leveraging multiple fidelity levels will be demonstrated on a sample conic geometry at hypersonic flight conditions. The results show that this improves model fit over a high-fidelity model alone while saving significant amounts of computational time, and the active learning strategies further reduce errors over conventional sampling plans.