Aerodynamic design optimization of a centrifugal compressor impeller based on an artificial neural network and genetic algorithm

Centrifugal compressors are applied to turbochargers, industrial compressors and turbo shaft gas turbine engines because of their high pressure ratio, relatively wide operating range and cost benefits. The internal flow in a centrifugal compressor impeller is however three dimensional and shows very complex flow phenomena, which makes the understanding of the loss generating mechanisms difficult and requires a considerable design effort to reach good performance. Especially turbocharger compressors impose a challenge to the designer when both a very wide operating range and high efficiency are required. The design effort can however be reduced by applying an advanced design optimization system as an alternative to a conventional manual design based on the experience of the designer. In this study a centrifugal compressor impeller for an automotive turbocharger was designed by means of an aerodynamic design optimization system composed of an artificial neural network (ANN) and a genetic algorithm (GA). This resulted in two newly designed centrifugal compressor impellers which were further studied both numerically and experimentally. One has higher efficiency with slightly wider operating range compared to the baseline impeller. The other one has a twice as wide operating range compared to the baseline impeller with a minor decrease in efficiency.