Multi-Objective Optimization of Aerodynamic Design of Space Reentry Vehicles Using a Deep Learning Surrogate Model

November 27, 2025

This thesis presents an integrated workflow for preliminary aerodynamic design and multi-objective optimization of non-winged reentry vehicles, namely bicones, cones, and capsules. The focus of the work is on the hypersonic regime, but the purely supersonic regime is also included. Geometries were systematically parameterized — including with Class-Shape Transformation (CST) method — to define a design space that is both tractable and capable of representing relevant features. Deep neural networks were trained as surrogate models to deliver fast predictions of aerodynamic coefficients, heat flux, and volumetric metrics for sampled configurations. Training databases were generated: low-fidelity data was obtained using the Modified Newtonian local inclination method and high-fidelity data was obtained from Computational Fluid Dynamics (CFD) simulations. For bicones, a multi-fidelity strategy was employed, combining scarce and expensive high-fidelity data with abundant low-fidelity data to preserve accuracy while covering a wide region of the design space at low cost. Once validated, the surrogate models were coupled with a multi-objective genetic algorithm, Non-dominated Sorting Genetic Algorithm II (NSGA-II), to perform Aerodynamic Shape Optimization (ASO) under different flight conditions. The objectives were to maximize lift-to-drag ratio and vehicle volume while minimizing a heat flux parameter. The resulting Pareto fronts were of high quality, showed clear and physically interpretable trade-offs, and produced vehicle geometries consistent with hypersonic aerodynamic theory. Furthermore, the low prediction errors relative to both low- and high-fidelity references confirm the robustness of the surrogate models.

Authors in the community:

• 

  L

  Leonor Couto Azevedo

  ist199516

Supervisors of this institution:

• 

  F

  Frederico Afonso

  ist156309

mechanical-engineering - Mechanical engineering

eng - English

Instituto Superior Técnico