Comparison of RANS Turbulence Models for Prediction of Film Cooling Performance

Abstract

The realizable k-ε, standard k-ω, and RSM turbulence models were used to simulate flat plate film cooling experiments that are commonly described in literature. Adiabatic effectiveness simulations revealed that using the standard k-ω model resulted in the closest agreement with experimentally determined laterally averaged adiabatic effectiveness, but the worst agreement with centerline adiabatic effectiveness. Conversely, the realizable k-ε model agreed worst with experimental laterally averaged adiabatic effectiveness values and best with centerline values. Use of the anisotropic RSM model was not found to predict more realistic coolant spreading than the other models. Simulations to find heat transfer coefficients without film cooling showed good agreement with correlations for all three models, and the closest agreement resulted from using the realizable k-ε model. Heat transfer coefficient augmentation was also examined for two configurations: unit density ratio with and without upstream heating. Laterally averaged heat transfer coefficient augmentation simulations using all three turbulence models agreed well with experiments. However, the spanwise variation in heat transfer coefficient augmentation in all cases was greater than is typically seen experimentally.