180 likes | 406 Views
Subgrid models in LES of drainage flows. Craig Smith Fernando Porte-Agel WIRE, EPFL, Switzerland craig.smith@epfl.ch. Craig Smith WIRE, EPFL. Why study drainage flows?. Source: Whiteman 2008. z jet = 5-10m – requires very fine resolution, or highly anisotropic grids. Craig Smith
E N D
Subgrid models in LES of drainage flows Craig Smith Fernando Porte-Agel WIRE, EPFL, Switzerland craig.smith@epfl.ch Craig Smith WIRE, EPFL
Why study drainage flows? Source: Whiteman 2008 zjet = 5-10m – requires very fine resolution, or highly anisotropic grids Craig Smith WIRE, EPFL
Science questions • Predictability of drainage flows • Subgrid model performance • Monin Obukov similarity (MOS) Rattlesnake Ridge – SE WA, USA Horan and Dorst, 1983 Craig Smith WIRE, EPFL
Observations Rattlesnake Ridge – SE WA, USA Horan and Dorst, 1983 Dx = 7m, Dy = 10m Craig Smith WIRE, EPFL
Numerical setup B C A B A 500x20x400 900x30x620 Dx = 20m, Dz = 2.5-3.8m Dx = 13.3m, Dz = 1.6-2.4m WIRE LES code – spectral difference in horizontal, center difference in vertical, 2nd order Adams Bashforth in time, z0 = 0.1 m, qsfc = -15-40 Wm-2 (not at peak) Subgrid models – Smagorinsky (wall corrected), Dynamic, Lagrangian scale dependent dynamic Craig Smith WIRE, EPFL
SGS models Craig Smith WIRE, EPFL
SGS models resolved scales unresolved scales log(Ek) Dissipation subrange Inertial subrange log(k) kc =p/D Oregon State University College of Oceanic and Atmospheric Sciences • Smagorinsky (wall corrected) – model coefficients are constant (isotropic homogenous TKE) • Dynamic – uses a 2nd test filter to dynamically optimize model coefficients • Scale dependent dynamic – allows model coefficients to vary with filter size
Results Scale dependent Lagrangian dynamic after 45 minutes of integration Velocity Potential temperature Craig Smith WIRE, EPFL
Results Scale dependent Lagrangian dynamic after 45 minutes of integration U q TKE Craig Smith WIRE, EPFL
Results Rattlesnake ridge observations comparison Craig Smith WIRE, EPFL
U q TKE Smagorinsky ->low near surface temperature -> 10-20% more cold air flux Craig Smith WIRE, EPFL
U q TKE Smagorinsky ->low near surface temperature -> 10-20% more cold air flux Craig Smith WIRE, EPFL
Results Smagorinksy Dynamic Scale dependent dynamic Assumption of constant model coefficients in Smagorinsky Craig Smith WIRE, EPFL
Results Dynamic Scale dependent dynamic Assumption of scale invariance in dynamic model Craig Smith WIRE, EPFL
In context Observations suggest zj = 5-10m Author location D bc sgs Burkholder et al 2010 - very small periodic many Axelsen 2009 glacial 2.6x2.6x0.4 periodic ??? Skyllingstad 2003 SE WA 0.75x0.75x0.75 slope FSF Smith and Skyllingstad 2005 VTMX 3x3x3 slope FSF 100x100x5 valley 1.5 TKE (ARPS) Chow et al 2006 MAP 150x150x20 valley 1.5 TKE and DRM (ARPS) Catalano and Cenedese 2010 - 50x50x2 valley WRF in LES mode Zhong and Whiteman 2008 VTMX 250x250x2.1 valley 2.5 TKE (RAMS) Craig Smith WIRE, EPFL
Future work • Numerical setup • Grid aspect ratio • Spanwise heterogeneity Smagorinsky model for katabatic winds – constant Prsgs Craig Smith WIRE, EPFL
Subgrid models in LES of drainage flows Craig Smith Fernando Porte-Agel WIRE, EPFL, Switzerland craig.smith@epfl.ch Craig Smith WIRE, EPFL
Results Smagorinsky – original vs wall corrected Craig Smith WIRE, EPFL