Validation of a low-dissipation Finite-Volume solver for implicit Large-Eddy Simulation

In the field of turbulent flow modelling, implicit Large-Eddy Simulation (iLES) is appealing for its low cost and ease of implementation. Such advantages rely on the absence of a sub-grid scale model, since the dissipation of the numerical scheme is assumed to match the behaviour of unresolved turbulence. The implementation of an iLES model in traditional Unsteady-RANS codes for Computational Fluid Dynamics is not a straightforward exercise, as most of the classical schemes used for the discretisation of the Navier–Stokes equations prove too dissipative. This work presents a low-dissipation fix for the traditional Flux-Difference Splitting scheme of Roe in the context of Finite-Volume discretisations. The fix consists in selectively scaling the eigenvalues of the Roe matrix to lower the numerical dissipation as needed, by means of a scalar parameter. The low-dissipation version of the Roe scheme is implemented in an existing Finite-Volume compressible wall-resolved URANS code, to obtain an iLES model. The solver is first verified on a fundamental test case, i.e. vortex transport in uniform flow. The scalar parameter is then properly calibrated on the decay of Homogeneous Isotropic Turbulence, to ensure physical meaningfulness. A robust validation of the iLES model is finally presented on realistic turbulent flows. Results show that a relatively simple fix can achieve excellent agreement with the benchmark DNS data on a flat-wall channel flow and a bumped-wall channel flow.