Black-box optimization methods for hypersonic flow problems

This review examines the application of black-box optimization methods to hypersonic flow problems, with a particular emphasis on scenarios where computational fluid dynamics (CFD) simulations function as opaque, nontransparent models. The paper begins by outlining the unique challenges associated with hypersonic flows and the core principles of black-box optimization. It then explores how genetic algorithms effectively address these complex optimization tasks. Additionally, the review discusses the use of machine learning-based surrogate models to optimize critical hypersonic system components—such as thermal protection systems and film cooling—where gradient information is unavailable and simulations are computationally expensive. The study highlights recent advances in integrating CFD solvers with optimization frameworks and presents successful cases where black-box strategies have significantly reduced computational costs while enhancing design robustness. Overall, this work offers a focused perspective on how black-box approaches can enable efficient and accurate optimization in the highly complex and computationally demanding domain of hypersonic flows.