Multi-Objective Combinatorial Optimization Design Method for the Compressor Splitter

Considering the present backward situation of the compressor splitter designing level, a multi-objective Combinatorial Optimization Design method is put forward for the splitter design consisting of three core components: the CST parameterized method, the Design of Experiment and the ASA optimization algorithm. In the whole optimization design process, the CST parameterized method is developed for the complex geometry modeling and geometric samples generating of splitter. The Design of Experiment is taken used to qualitatively analyze the multiple design variables and adjust their number and scopes. The ASA algorithm takes charge of the global optimization of splitter geometry samples, and selects the best geometry conform to the design target. All the relevant optimization components and modules are integrated by the Isight software and be interacted and automatically called by the edited scripts. The effectiveness of the proposed design method is verified through a practical splitter example. Results show that: 1) the CST parameterized method of fourth order or above is of great accurately to fit the splitter geometry, in order to generate variable geometric samples. 2) The Design of Experiment is able to identify the different influence of multiple design variables on the design objectives and lead to the optimization process more targeted. 3) The performance of optimization result is improved significantly, in which the core duct behaves more sensitive to splitter geometry changing and the low-static pressure and high-radial velocity areas reduced to half of the initial design.