Hybrid RANS-LES Modeling Using a Low-Reynolds-Number k-ω Based Model

Abstract

Hybrid RANS-LES modeling is proposed using a Low-Reynold-Number (LRN) k − ω model. The model is demonstrated in a zonal RANS-LES approach and in an embedded LES approach. The model is calibrated and evaluated using Decaying Homogeneous Isotropic Turbulence (DHIT), turbulent channel flow and turbulent flow over a hump. The effect of different LES length scales on log-layer mismatch and turbulence resolving capability is demonstrated using the proposed model. Interface conditions are proposed in the embedded LES approach in order to reduce the grey area zone in the LES domain downstream of the RANS region. To further improve the development of turbulence resolving flow in the LES region downstream of the interface, anisotropic turbulent velocity fluctuations from synthetic turbulence are added. The hybrid RANS-LES modeling approaches that are presented, using the LRN k − ω based model, show that predictions of turbulence resolving flows are in reasonable agreement with experimental data and DNS data. Moreover, the choice of the LES length scale using the proposed model is shown to be of great importance in reducing the log-layer mismatch.