| description abstract | Reynolds-averaged Navier–Stokes (RANS) simulation is a routinely used tool for turbomachinery research and development, but it often over-predicts the pressure ratio of centrifugal compressors especially in the high impeller tip Mach number regime. In this work, the effects of a series of geometrical and numerical uncertainties and errors on the aerodynamic performance of a centrifugal compressor are investigated systematically. The investigated compressor is the NASA CC3 centrifugal compressor under different impeller tip Mach number conditions. The investigated geometrical/numerical factors include the impeller blade fillet, the impeller hub cavity, the impeller running tip clearance, the averaging method in the post-process, the turbulence model, the inlet duct hub rotation, and the inlet turbulence boundary condition. Results show that the uncertainty of the predicted total pressure ratio generally increases with the impeller tip Mach number. Among the investigated factors, the averaging method, the impeller blade fillet, the turbulence model, and the impeller hub cavity have the most pronounced effects in determining the compressor total pressure ratio. By adopting realistic geometric features, advanced turbulence modeling treatments, and the same averaging method as the experiment, the over-prediction in the total pressure ratio can be alleviated. Detailed flow mechanism analysis with respect to the impeller hub cavity and the turbulence model has been performed. These findings provide valuable guidance for future RANS simulations of centrifugal compressors. | |
