Calibrated Rotation-Helicity-Quadratic Constitutive Relation Spalart-Allmaras (R-H-QCR SA) Model for the Prediction of Multi-Stage Compressor Characteristics

Abstract

Compressor performance prediction is still one of the significant interests in the turbomachinery research field. The two critical parameters for compressor design are adiabatic efficiency and stability margin. The Spalart-Allmaras (SA) turbulence model and modified SA models are widely used in that design process. However, the prediction accuracy is not always satisfactory. In most cases, the SA model predicts larger stall mass flow, and the RC-QCR SA model underestimates efficiency. This study proposes a new combination of the modified SA model (R-H-QCR model). R-H-QCR stands for Rotation-Helicity-Quadratic constitutive relation. The model increases or decreases turbulent viscosity based on flow rotation, energy backscatter, and anisotropy of turbulence flow field. The Bayesian inference framework calibrates the model parameters to predict accurately both efficiency and stability in the 3.5 stage compressor. The R-H-QCR, RC-QCR, and default SA models are evaluated in the multi-stage compressor. For the performance prediction, the R-H-QCR model predicts a better stability margin than the SA model and better efficiency than the RC-QCR model. In addition, the spanwise distribution of normalized total pressure is well captured by the R-H-QCR model, indicating that the R-H-QCR model improves flow field prediction.