Mitsubishi Heavy Industries Technical Review
    Vol. 52 No. 1 (2015)   New Products & Technologies
    Technical Papers

    Aerodynamic Design Optimization of Centrifugal Compressor Impeller Based on Genetic Algorithm and Artificial Neural Network

    SEIICHI IBARAKI
    ISAO TOMITA
    KOICHI SUGIMOTO

    The centrifugal compressor in an automotive turbocharger is required to have a high pressure ratio, high efficiency, and in particular, wide operating range. Performance improvement has been carried out utilizing computational fluid dynamics and experiments, and further performance improvement through conventional design methods has become increasingly difficult, requiring a significant amount of time for aerodynamic design. Therefore, we developed an optimized design method with computers using a genetic algorithm and artificial neural network in place of the conventional design method. This method was applied to the design of the centrifugal compressor impeller. The performance results of the two designed impellers attained higher efficiency and a significant extension of operating range, respectively, compared with the baseline impeller.