Comparison of Reynolds-Averaged Navier-Stokes Turbulence Models under Condition of Oscillatory Boundary Layer Flow

This study aims to compare Reynolds-averaged Navier-Stokes equations (RANS) turbulence models (k-epsilon and k-omega models) under the condition of oscillatory boundary layer flow at high Reynolds number. For the condition of negligible spanwise flow (y-direction in this study), the U-tube experiment is replicated by a one-dimensional vertical (1DV) numerical model. The most reliable conclusion of this study is that a k-epsilon model and a k-omega model calculate very similar flow condition and turbulent kinetic energy (TKE). This finding is different from previous study. Puleo et al. (2004) also applied a 1DV model incorporated with k-epsilon and k-omega models to U-tube experiment and calculated TKE. They concluded that the result of k-omega model is more accurate than that of k-e model. From Wilcox (1993), it is known that a k-omega model has an advantage over k-e model under the condition of large spanwise pressure gradient. As mentioned, spanwise flow of U-tube experiment is assumed to be negligible. Therefore, it is questionable that the result of k-epsilon model has less accuracy compared to the result of k-omega model. Pope (2000) insisted that a k-omega model is sensitive to calibration of empirical parameters. Based on results of this study and Pope (2000), it is concluded that a k-epsilon model does not a disadvantage compared to k-omega model under the condition of negligible spanwise flow. Therefore, it is also concluded that a k-e model has a capability to calculate turbulent flow and TKE when studying cases such as straight channel flow, estuarine flow and oscillatory flow of which spanwise (or along-shelf) pressure gradient is not significantly large.