On predicting the turbulence-induced secondary flows using nonlinear k-ε models

Low turbulent Reynolds number direct simulation data are used to calculate the invariants of the Reynolds stress and the turbulent dissipation rate in a square duct. The results show that, depending on the region where the analysis is carried out, the turbulent flow field comes close to one-, two-, and three-component states. Modeling such flows—even at higher Reynolds numbers—will require models that can approach all three states. A number of related nonlinear k-ε models are tested a priori using the direct simulation data. The numerical simulation using Reynolds averaged Navier–Stokes equations with these models was performed. Their ability to predict the secondary flows, with a low-Reynolds k-ε model, cannot be gauged from realizability.