Performance improvement of centrifugal compressors for fuel cell vehicles using the aerodynamic optimization and data mining methods

Abstract

Centrifugal compressors are one of the most important auxiliary components in polymer electrolyte membrane fuel cell vehicles, which tend to operate at a narrow area with low specific speed. Here, the optimal design goals of centrifugal compressors are investigated on the basis of a lumped model for fuel cell systems. A three-dimensional multi-objective and multi-point aerodynamic optimization and data mining method for centrifugal compressors named ODM is presented via integrating a multi-island genetic algorithm, Reynolds-Average Navier-Stokes solver technique and self-organization map based data mining technique. Data mining indicates that compressor geometry would move to a small inlet diameter ratio and a narrow region of the outlet width ratio. Based on the optimization results, a centrifugal compressor for 100 kW fuel cell stack is manufactured. The experimental results show that the improvement of isentropic efficiency near low mass flow has been achieved, which indicates that the proposed ODM is effective in the performance improvement of centrifugal compressors for fuel cell vehicles.