Optimization and Knowledge Discovery to the Design Problem of Fuel Injector in the Supersonic Combustor

A Multi-Objective Genetic Algorithm (MOGA) is applied in search of the optimal fuel injector shape in the supersonic combustor. The optimal shape is investigated in terms of rapid mixing of fuel jet and supersonic airstream. Two goals, those are to maximize the jet core height and to minimize the total pressure loss, are adopted for the optimization. The numbers of maximum injector holes are restricted to 1, 2 and 3 during the evolutive process. Pareto fronts under the 2 and 3 maximum holes restriction evolve successfully. Self-Organizing Map (SOM) is applied, in order to realize the role of optimal solutions in actual design, to analysis of structure on design space by means of design database obtained in evolutive process of optimization. The design database is mapped onto the two-dimensional SOM, where design space is successfully visualized. The visualization that clarifies relativity of performance parameters acquires valuable design information.