Modeling of flows closer to Kolmogorov Scale
 

I am trying to simulate a leakage in a tank through an orifice in CCM+, the diameter of the orifice is 1um. I realize that the scale is smaller or almost closer to kolmogorov scale, as TKE would cease to exist, I am modeling using Spallart Allmaras Model which solves for viscosity alone. It would be quite helpful if someone has done similar work in the past. The literature on leakage rates in quite less.

If there is no TKE then that implies that there is no turbulence and no Kolmogorov scales present. Which means there's also no turbulent viscosity.

That type of problem is really a nano / micro channel flow problem. Navier-Stokes is not very appropriate description of those types of flows (and forget turbulence). Although small, the Kolmogorov scales are still much larger than the mean free path of the mean molecular motion. When your channels are 1 micron in diameter, you are restricting the microscopic and macroscopic motion to much smaller dimensions. Even the continuum assumption breaks down.

Bottom line is, I wouldn't use CFD for those types of problems.

My mistake, i shouldn't have used the term "doesn't exist". TKE would dissipate as heat and the remaining term would be viscosity. I am not sure if DNS would solve the problem. When i run using spallart allmaras model, I can see a huge rise in temperature within the orifice.

Don't get me wrong, I agree with your earlier statements. I would probably say that TKE does not exist. But given that we know TKE is not an appropriate for these flows, then why use turbulence models? Turbulence is already non-existent. DNS is probably inappropriate. DNS is just accurate solution to the Navier-Stokes equations which are already not appropriate.

You really need a nano-microscale fluid flow model to solve these problems, not Navier-Stokes.

That's just my take, maybe someone else has tried it and it works.

I was also coming down on that conclusion, but there is a bit of uncertainty as the little literature i've found mentions about choked flow in the orifice. DNS and turbulence models (rans, k-omega..etc) are two different things, for that reason i was thinking about DNS. The results however are still subject to experiment or previous work done by someone.