Compressible Large Eddy-Simulations (LES) combining high-order methods with a wall model have been developed in order to compute wall-bounded flows at high Reynolds numbers. The high-order methods consist of low-dissipation and low-dispersion implicit finite-volume schemes. In a first part, the procedure used to apply these schemes in nearwall regions is presented. This procedure is based on a ghost cell reconstruction. Its validity is assessed by performing the LES of a bi-periodic turbulent channel flow at a Mach number of M = 0.2 and a friction Reynolds number of Reτ = 395. In a second part, to consider flows at higher Reynolds numbers, a LES approach using a wall model is proposed. The coupling between the wall model and the high-order schemes is described. The performance of the approach is evaluated by simulating a bi-periodic turbulent channel flow at M = 0.2 and Reτ = 2000, and an isothermal subsonic round jet at M = 0.6 and ReD = 5.7 × 10 5 . The results are in agreement with Direct Numerical Simulation (DNS) data and experimental results. In particular, the turbulent intensities obtained in the logarithmic region of the boundary layers of the channel flow and the jet far-field noise are successfully predicted.

No abstract

No abstract

In this study, a technique of flux reconstruction is proposed to perform aeroacoustic computations using high-order methods on multiblock structured meshes with non-conforming grid interfaces. The use of such grids facilitates zonal mesh refinements, and flows at high Reynolds numbers can thus be simulated at a reasonable cost. The high-order methods consists of low-dissipation and low-dispersion implicit finite-volume numerical schemes. Using a flux reconstruction method, they can be applied on non-conforming grids. In a first part, the method is described. It is based on the application of non-centered spatial schemes and the use of ghost cells. The flow variables in the ghost cells are computed from the flow field in the grid cells using local meshless interpolations with radial basis functions. Then, the performance of the method is evaluated by carrying out two-dimensional simulations of vortex convection and of a mixing layer. The results show that no significant spurious acoustic waves are produced at the grid interfaces. Finally, the flux reconstruction approach is applied to the computation of a three-dimensional jet at a Mach number of 0.6 and a Reynolds number based on the jet diameter of 5.7 × 10 5. In particular, nonconforming grids are used to obtain 384 points in the azimuthal discretization in the jet shear layers, while using less points at the center of the jet. Preliminary results regarding the jet development are shown and compared with experimental data.

Compressible large-eddy simulations combining high-order methods with a wall model have been developed in order to compute wall-bounded flows at high Reynolds numbers. The high-order methods consist of low-dissipation and low-dispersion implicit finite volume schemes for spatial discretization on structured grids. In the first part, the procedure used to apply these schemes in near-wall regions is presented. It is based on a ghost cell reconstruction. Its validity is assessed by performing the large-eddy simulation of a turbulent channel flow at a friction Reynolds number of Re τ 395. In the second part, to consider flows at higher Reynolds numbers, a large-eddy simulation approach using a wall model is proposed. The coupling between the wall model and the high-order schemes is described. The performance of the approach is evaluated by simulating a turbulent channel flow at Re τ 2000 using meshes with grid spacings of Δx 100, 200, and 300 in the streamwise direction and Δ 50, 100, and 150 in the wall-normal and spanwise directions (in wall units). The effects of the choice of the point used for data exchange between the wall model and the large-eddy simulation algorithm, as well as of the positions of the ghost cells used for the coupling, are examined by performing additional computations in which these parameters vary. The results are in agreement with direct numerical simulation data. In particular, the turbulent intensities obtained in the logarithmic region of the boundary layers of the channel flow are successfully predicted.