Aerodynamic Optimization of the SRV2 Radial Compressor Using an Adjoint-Based Optimization Method

Abstract

Abstract The design of radial compressors is typically a trade-off between peak performance at the design conditions and maintaining a good efficiency over a wide operating range. In the present paper, a gradient-based optimization of the well-known SRV2 radial compressor is presented. The optimization problem is formulated to increase the total-to-total efficiency of the compressor at peak efficiency while maintaining a wide operating range. To achieve an efficient optimization strategy, a gradient-based optimization method is used. Gradients of the RANS solver are obtained through the adjoint approach, leading to a computational cost that is almost independent of the degrees of freedom. A total of 44 design parameters are used to define a rich design space for the radial compressor impeller. A Sequential Quadratic Programming algorithm is used as optimizer and achieves an optimal design in less than 20 major iterations. The optimization process allows to significantly increase the efficiency of the SRV2 compressor at peak efficiency while maintaining the operating range of the baseline. Detailed analysis of the flow field in baseline and optimized compressors reveal the mechanisms which allow to improve the performance of the baseline compressor.