Numerical assessment of turbulent models at a critical regime on unstructured meshes

Abstract

The present study mainly aims to investigate the performances of different turbulent models for the flow simulation around a circular cylinder at a critical Reynolds number regime (Re = 8.5×10⁵, Tu = 0.7%). A hybrid RANS/LES model (SAS model), a correlation-based transition model ($\gamma - \widetilde{\operatorname{Re} }_{\theta t}$ model), and a fully turbulent RANS model (SST model) were used to simulate various flow features, such as laminar-turbulence transition inside the boundary layer and the unsteady vortex shedding in the wake region, and their feasibilities for the flow simulation at a critical Reynolds number regime were demonstrated. A vertex-centered finite-volume method was used to discretize the incompressible Navier-Stokes equations, and an unstructured mesh technique was used to discretize the computational domain. The inviscid fluxes were evaluated using 2^nd-order Roe’s flux difference splitting, and the viscous fluxes were computed based on central differencing. A dual time-stepping method and the Gauss-Seidel iteration were used for unsteady time integration. The parallelization strategy using METIS and MPI libraries was used to reduce computational costs. The unsteady characteristics and the time-averaged quantities of the flow fields were compared between turbulent models. The numerical results were also compared with experimental results. The turbulent models showed quite different results at the critical regime because of the different abilities of each model to predict various flow features, such as laminar-turbulence transition and unsteady vortex shedding.