• backward facing step
• CFD
• channel flow
• Direct numerical simulation
• Large eddy simulation

Computational fluid dynamics plays an important role in engineering and nuclear applications. In this work the performance of Large Eddy Simulation (LES) is investigated. Before applying LES methods in engineering applications, it is important to validate the models, therefore in the present study Dynamic Smagorinsky model by Lilly (1992) and standard Smagorinsky model (1963) are employed. Another goal of this study is to determine the mesh requirements for an accurate prediction using LES.
A number of nuclear and engineering applications include turbulent wall-bounded flows with separation and reattachment regions.
As a first case of validation, a fully-developed turbulent channel flow is considered.
Direct numerical simulations (DNS) of channel flows at Re_tau=180 and Re_tau=395 are carried out with different mesh resolutions to validate the accuracy of the CFD package: OpenFOAM. Then, channel flows simulations using LES with the two different sub-grid scale (SGS) models are carried out on various grid sizes and the results are compared against their respective DNS reference results.
As a second validation case, among the flow geometries used for the study of separated flows, the backward-facing step is selected.
The Dynamic SGS model by Lilly (1992) is validated for the flow through the backward-facing step geometry using same mesh resolutions as in the channel flow case.
The results obtained by performing LES for the backward-facing step are compared against experimental data performed by Kasagi et al. (1995) and LES results obtained by Toschi et al. (2006).