The Impact of Turbulence Model and Numerical Scheme on the Performance of a Linear Compressor Cascade

In the use of RANS models, it is well known that the selection of the turbulence model and the numerical scheme may have a critical impact not only in terms of convergence, but also on the reliability to simulate separated or secondary flows in general. The aim of the investigation, performed using the commercial software FINE/Turbo, is the understanding and the quantification of the effects of these two numerical parameters on the performance and the stability of a state-of-the-art controlled diffusion airfoil compressor cascade. A mesh sensitivity analysis has been carried out at both design and off-design conditions. The behaviour of the main flow parameters have been investigated over the whole incidence working range, considering a variation of the inlet Mach number between 0.35 and 0.65. Five different turbulence models have been tested: Baldwin-Lomax, Spalart-Allmaras, k–ε Yang-Shih, k–ε Launder-Sharma and k–ω SST. In a specific combination of incidences and Mach numbers, the impact of turbulence model settings has been assessed imposing boundary conditions according to different criteria. Two different numerical schemes have been tested: a Jameson central scheme and a second order upwind scheme. The results between the different simulations are discussed in terms of loss coefficient distribution and incidence range; considering the turbulence model comparison, the differences are significant in the whole incidence range, specially approaching the stall limit. Baldwin-Lomax and Spalart-Allmaras simulations present the same value of last stable incidence, while Yang-Shih and SST are characterized by a reduced stall margin. In many operating conditions, simulations computed with centered scheme present negative losses in a wide area of the outlet sections. This problem is reduced if an upwind scheme is used, but causes a substantial reduction of the incidence range.