A low dissipation numerical scheme for Implicit Large Eddy Simulation

•A numerical method is constructed for Implicit Large Eddy Simulation.•It combines the MDCD reconstruction and the SLAU flux in a Finite Volume framework.•The method has high resolution and good all-speed performance.•The method’s numerical dissipation is validated to be more physically consistent.•Results convincingly show its capability of predicting massively separated flows. The design of numerical discretization with physical consistency remains a challenge in the Implicit Large Eddy Simulation (ILES). Both the reconstruction and the Riemann solver may introduce excessive numerical dissipation in turbulence simulation with Finite Volume Method. In this paper, a numerical method is proposed for ILES by combining two low-dissipation schemes: the Minimum Dispersion and Controllable Dissipation (MDCD) scheme for reconstruction, and the Simple Low-dissipation AUSM (SLAU) scheme for the Riemann solver. The numerical features, including the high resolution (despite second-order accuracy) and the good all-speed performance, are verified by several numerical experiments. Quantified by the Taylor–Green vortex problem, the method’s numerical dissipation and its relation with the relative resolvable scale are discussed. The dissipation’s behavior is found to be more consistent with that in the physical flow, and the robustness is satisfactory. Then the MDCD/SLAU scheme is applied to ILES of the separated flow over periodic hills. The flow statistics agree well with the experimental data. These simulations show that the MDCD/SLAU scheme is a promising candidate for ILES of engineering flow problems.