Hello everybody,

I have a question regarding LES simulations of laminar or near-laminar flows (i.e. low Reynolds number flows).

From what I know, if I do a LES of a flow, as my grid spacing gets smaller, smaller and smaller scales are filtered out, until at some point the simulations actually become DNS (all scales resolved). Unless all scales are resolved, they are modeled by a subgrid model. In the field I'm in, the dynamic Smagorinsky-Lilly model is the subgrid model of choice.

However, the dynamic subgrid model is based on the assumption that there exists an intertial subrange in the turbulent energy spectrum. For the LES to be well resolved and the dynamic model to be accurate, the smallest scales that are filtered out should be of (very) high wavenumber, and the test-filter should be put somewhere relatively close in the inertial subrange.

Now, if I do a LES of a laminar or near-laminar flow, and use a very fine mesh, the influence of the subgrid modeling will be very small (due to the low ratio of the subgrid turbulent viscosity to the molecular viscosity). My question then becomes: If the grid is coarser, so that the subgrid turbulent viscosity becomes influental (but of course still not dominating), what is the physical implication of the output from the dynamic Smagorinsky-Lilly model in this situation? Is it still a useable model, or has the absence of an inertial subrange in the low Reynolds number flow rendered it dubious? How would the results be affected by any inconsistency in the theoretical ground for the dynamic model in this situation?

Many thanks for any input on this!

/Henrik

What I understand from my limited experience is that as long as we are resolving some inertial scales we should be ok. Do you know how can we figure out if our grid is fine enough to resolve inertial scales?

Its confusing why are using LES for laminar flow?

Regards