Simulation of a 3D Axisymmetric Hill: Comparison of RANS and Hybrid RANS-LES Models

Abstract

Computational fluid dynamics (CFD) prediction of high Reynolds number flow over a 3D axisymmetric hill presents a unique set of challenges for turbulence models. The flow on the leeward side of the hill is characterized by the presence of complex vortical structures, unsteady wakes, and regions of boundary layer separation. As a result, traditional eddy-viscosity Reynolds-averaged Navier-Stokes (RANS) models have been found to perform poorly. Recent studies have focused on the use of Large Eddy Simulation (LES) and hybrid RANS-LES (HRL) methods to improve accuracy. In this study, the capability of a dynamic hybrid RANS-LES (DHRL) model to resolve the flow over a 3D axisymmetric hill is investigated and compared to numerical results using a traditional RANS model and a conventional hybrid RANS-LES model, and to experimental data. Results show that the RANS model fails to accurately predict the mean flow features in the wake region, which is in agreement with prior studies. The conventional HRL model provides better prediction of the flow characteristics but suffers from grid sensitivity and delayed transition to LES mode. The DHRL method provides the best agreement with experimental data overall and shows least sensitivity to grid resolution. Results also highlight the importance of using a low dissipation flux formulation for flow simulations in which a portion of the turbulence spectrum is resolved, including hybrid RANS-LES.