Knowledge Discovery on Cavity-Based Scramjet Combustor Design via Stochastic-Surrogate-Assisted Multi-Objective Optimization

Designing scramjet combustors is one of the most crucial yet challenging tasks in developing a scramjet engine, primarily due to its complex aerothermal and chemical phenomena as well as high computational cost that is inherently incurred in the design process. The cost issue has prohibited the global design exploration of scramjet combustors while further improvement of scramjet performance necessitates such design techniques including multi-objective and global design optimization. The present study enables multi-objective design optimization (MDO) of scramjet combustors by developing a cost-efficient and robust MDO framework. This framework incorporates evolutionary algorithms, Bayesian optimization, and stochastic surrogate modeling. It warrants significant reduction of computational costs by reducing the number of solution evaluations while ensuring the robustness of the solution search against the inaccuracy of surrogate modeling. In the present study, a cavity-based two-dimensional scramjet combustor is optimized for hydrogen combustion with respect to thrust and combustion efficiency. A detailed post-analysis of the optimization results provides valuable physical insights into scramjet combustor design, and the findings are summarized as the optimal design strategies for cavity-based scramjet combustors.