P2.27
Assessment of explicit filtering and reconstruction turbulence models for large-eddy simulation over complex terrain
Bowen Zhou, Univ. of California, Berkeley, CA; and F. K. Chow
In this study, large-eddy simulation (LES) is performed over complex terrain for both idealized and real cases. We use an explicit filtering and reconstruction turbulence modeling approach which splits the subfilter-scale (SFS) turbulence stress into resolvable subfilter-scale (RSFS) stresses and subgrid-scale (SGS) stresses. The former are recovered from a velocity reconstruction approach, and the latter are represented by either a TKE-1.5 model or a dynamic eddy-viscosity model. This approach has been proven advantageous in our previous work in terms of turbulence representation and agreement with similarity theory in neutral and stable atmospheric boundary layer flow over flat terrain. It has also been successfully applied in LES over Askervein Hill, Scotland, but with certain limitations. In this work, the limits and applicability of the explicit filtering approach are explored using flow over a 3D bell shaped mountain. This idealized test case has two objectives: to evaluate the performance and computational cost of reconstruction models with conventional eddy viscosity closures, and to assess the grid aspect ratio and hill slope requirements for using the reconstruction model under a terrain following coordinate system. Using the results from the idealized study as a guideline, we perform LES in Owens Valley, located in the lee of (east of) the California Sierra Nevada Range, which was the site of the 2006 Terrain-Induced Rotor Experiment (T-REX). A set of nested grids are used to resolve the region of interest, beginning with 9 km horizontal spacing down to a fine grid with 350 m spacing. At 1 km and coarser grid levels, the TKE-1.5 LES closure is used, as it is more appropriate for such resolutions. The 350m grid resolution run is carried out with explicit filtering and reconstruction models. We will evaluate conclusions from the idealized study using this real test case and validate our simulation results using radiosonde and other data from the field campaign. Results with explicit filtering and reconstruction will be compared to conventional LES closure models.
Poster Session 2, Wednesday Poster Session
Wednesday, 1 September 2010, 10:00 AM-11:30 AM, Alpine Ballroom B
Previous paper Next paper