Multifidelity Methods for the Design of a Re-entry Vehicle

The design and optimization of space systems presents many challenges associated with the variety of physical domains involved and their coupling. A practical example is the case of space vehicles designed to re-enter the atmosphere upon completion of their mission. For these systems, aerodynamics and thermodynamics phenomena are strongly coupled and relate to structural dynamics and vibrations, chemical non equilibrium phenomena that characterize the atmosphere, specific re-entry trajectory, and geometrical shape of the body. Blunt bodies are common geometric configurations used in planetary re-entry. These geometries permit to obtain high aerodynamic resistance to decelerate the vehicle from orbital speeds along with contained aerodynamic lift for trajectory control. The large radius-of-curvature allows to reduce the heat flux determined by the high temperature effects behind the shock wave. The design and optimization of these bodies would largely benefit from accurate analyses of the re-entry flow field through high-fidelity representations of the aerodynamic and aerothermodynamic phenomena. However, those high-fidelity representations are usually in the form of computer models for the numerical solutions of Partial Differential Equations (e.g. Navier-Stokes equations, heat equations, etc.) which require significant computational effort and are commonly excluded from preliminary multidisciplinary design and trade-off analysis. This work addresses the integration of high-fidelity computer-based simulations for the multidisciplinary design of space systems conceived for controlled re-entry in the atmosphere. In particular, we will explore and discuss the use of multifidelity methods to obtain efficient aerothermodynamic models of the re-entering systems. Multifidelity approaches allow to accelerate the exploration and evaluation of design alternatives through the use of different representations of a physical system/process, each characterized by a different level of fidelity and associated computational expense. By efficiently combining less-expensive information from low-fidelity models with a principled selection of few expensive simulations, multifidelity methods allow to incorporate high-fidelity costly information for multidisciplinary design analysis and optimization. Modern multifidelity methods leverage active learning schemes to optimize the selection of the sample points while searching for the design optimum. This thesis discusses multifidelity methods and compares different implementations to assist the design of aerospace systems. In particular, active learning frameworks based on multifidelity expected improvement (MFEI) are implemented and assessed for the optimization of different benchmark analytical functions. Additionally, an original MFEI algorithm is proposed and implemented for the specific case of the design optimization of the Orion-like atmospheric re-entry vehicle. The results show that the proposed MFEI algorithm gives better optimization results (lower minimum) than single fidelity active learning based on low-fidelity simulations only. The outcome suggests that the MFEI algorithm effectively enriches the information content with the high-fidelity data. Moreover, the computational cost associated with 100 iterations of our multifidelity active learning strategy is sensitively lower than the computational burden of 6 iterations of a single fidelity framework invoking the high-fidelity model.