Investigation of high-fidelity turbulence modelling for corner-flow separation using delayed detached-eddy simulation

Corner separation, a common flow separation phenomenon in aviation compressors, significantly compromises aerodynamic performance. Conventional Reynolds-averaged Navier-Stokes (RANS) turbulence models, such as the Spalart–Allmaras (SA) and Menter’s shear stress transport (SST) models, systematically overpredict the extent of corner separation. This inaccuracy stems primarily from their failure to capture turbulence non-equilibrium in the separated flow region. To provide reliable benchmark data for model improvement, an improved delayed detached-eddy simulation (DDES) approach is employed in this study to construct a high-fidelity, scale-resolving turbulence database for a highly loaded aero-engine compressor cascade. This database is then used to evaluate the capability of helicity-corrected turbulence models in predicting both the turbulence non-equilibrium and the corner separation extent—the latter characterized by the location of peak turbulent kinetic energy production near the corner. Validation with two practical engineering test cases confirms that the helicity-corrected models, calibrated and justified against the DDES results, substantially enhance the accuracy of Computational Fluid Dynamics (CFD) predictions for corner-separated flows, demonstrating strong potential for engineering applications.