Challenges and remedies for accurate performance prediction of a multistage large-clearance axial compressor

This paper addresses the challenges of conventional steady Reynolds-Averaged Navier-Stokes (RANS) simulations in predicting the aerodynamic performance of large-clearance multistage axial compressors and proposes practical remedies to enhance prediction accuracy. The study is based on a 4-stage high-speed Mitsubishi Heavy Industries (MHI) compressor, which represents the middle-rear stages of a one-shaft gas turbine compressor. Conventional RANS simulations predict the performance well for the small-clearance case, but they over-predict the 3D tip blockage and losses in the front stages of the large-clearance case, leading to premature stall onset. To address these shortcomings, an advanced combination of turbulence model corrections is identified, which includes the quadratic constitutive relation (QCR) correction, relaxing the linear constitutive relation assumption of the Boussinesq approximation, and the vortical pressure gradient correction, incorporating the vortex stretching effect due to the pressure gradient in non-equilibrium 3D separated flows. The advanced turbulence model effectively reduces the prediction error of the conventional model by about 76% for total pressure ratio, 50% for isentropic efficiency, and 77% for stall margin. It also demonstrates strong predictive accuracy for overall performance and stage matching for both small and large clearance compressors. The proposed advanced turbulence model could be a promising tool for future small-core engine compressor design cycles.