Hi all, In a mesh highly anisotropic and unstructured, what is the best way to calculate delta to be used in LES sgs models (say, Smagorinsky and oneEquation models, for instance)? Obviously, in this case the classic volume cubic root would not work, and the other options I know need a cartesian mesh...

Thanks in advance, sampaio

If your grid is anisotropic and unstructured, you are dealing with a filter that varies in space.

In any calculations, simply make \Delta a function of space..

Is it really only that? Cause my delta is already a function of x,y,z (I am using cubic root of cv)

I wonder what happens if deltaZ is not in inertial range any more... This will certainly happens, if you deal with separated flows (wake behind a cylinder, for instance) and cannot afford the burden of building different dz in the spanwise direction acording to your physics (gradative braking down of vortex, evolving towards turbulence further downstream...)

Thanks for your comments.