Large Eddy Simulation of Wall-bounded Turbulent Flows at High Reynolds Numbers

Abstract

In the simulation of turbulent flows, resolving flow motions near a solid surface requires a high resolution that is computationally expensive. The present research investigates reducing the computational cost of simulating wall-bounded flows through a technique, called wall-modeling, that introduces the effects of the near-wall flow dynamics as a wall shear stress to the outer layer. Turbulent wall bounded flows were studied using large eddy simulation at moderate to high Reynolds numbers to evaluate the performance of the wall-modeling.

The results of wall-modeled turbulent channel flow at Re = 2000 were in good agreement with the experimental data. However, a log-layer mismatch was observed in the mean velocity profile below the matching point due to the inconsistency between the local grid resolution and that required by the subgrid scale model. Moving the matching point further from the wall mitigated the mismatch. The effects of time averaging and temporal filtering schemes on the performance of the wall model were also investigated. It was found that smaller time periods for time averaging result in a wall model that is more responsive to the flow structures in the outer layer. The results indicated that the temporal filtering scheme is strongly dependent on the location of the matching point.

Next, the wall-modeling was implemented in the simulation of a turbulent boundary layer. Inflow generation methods were reviewed, and a recycling rescaling method was employed to generate realistic turbulence at the inlet boundary. Zero pressure gradient turbulent boundary layers over a wide range of Reynolds numbers up to Re = 25 523 were studied in terms of the mean velocity profile, Reynolds stress, and skin-friction coefficient. It was found that a wall-modeled turbulent boundary layer can be resolved using a much lower grid resolution in the wall layer.

Finally, the wall stress model was implemented to introduce the effects of wall roughness into the wall-modeling via the eddy viscosity. The proposed wall model was examined for transitionally and fully rough channel flows and successful results were achieved. For high-Reynolds number wall-bounded flows, wall-modeling can effectively couple a large eddy simulation to the wall via the wall shear stress without the need to fully resolve the inner region.