Comparison of Computational Velocity and Heat Transfer Predictions to Experimental Measurements in a Rotating Cooling Passage With Smooth Walls

Abstract

Numerical predictions of the turbulent velocity field and wall heat transfer for a simulated turbine blade cooling passage are presented. The square cross-sectioned, smooth-walled passage is identical to one for which velocity and heat transfer data are available for comparison. Reynolds number (8000), rotation number (0.2), and buoyancy numbers (0 and 0.49) are typical of gas turbine applications. Predictions are presented for three turbulence models: standard k-ε, Renormalization Group k-ε, and Reynolds Stress. In addition, two wall treatments are evaluated: wall functions and a two-layer zonal model. Results from the three models are comparable, however the two-layer zonal wall treatment provides the best match to both the experimental flowfield data and the Nusselt distribution. Wall functions are shown to be unsuitable for this flowfield. General flow features in the passage are adequately captured by the zonal model including the Coriolis-induced double vortex and the distorted streamwise velocity profile due to the buoyancy effect. Agreement between the calculated and measured streamwise velocity profile (from leading to trailing wall) is particularly remarkable and contributes to an impressive leading and trailing Nu match with the data. This agreement suggests that the model adequately accounts for the buoyancy effect on the bulk flow without any buoyancy terms in the k or ε conservation equations. The model is less effective, however, at capturing the specific vortex position and strength. Specifically, the model vortex has only half the measured vortex maximum velocity and is located forward or aft of the passage centerline (depending on the density ratio of the flow).