RAS Turbulence Model for Natural/Forced Convection
 

Hello.

I've been using k-omega SST model for water pumps (pimpleFoam) and I get very good results compared to actual, real-life measurements even on far-from-perfect meshes.

However, when I use this model on (compressible) chtMultiRegionFoam for simulation of passive heat sinks, I get temperatures that are much too high (again compared to real-life measurements). Besides that, the whole case is very mesh-sensitive (smaller cells > higher temperatures). I have thoroughly checked my mesh, boundary conditions and other settings a zillion times over.

Does anyone have any experience with natural/forced convection and can suggest which model to use? It is a case with external flow, compressible, low velocities (Re ~ 8000). Currently I'm running an RNGkEpsilon case that gives results closer to expected but I would like to know more than just trying this and that.

Any help would be greatly appreciated.
Thanks!

Hi Foamers,

I am roughly looking at a same conditioned wall function/turbulant model, to simulate conjugate heat transfer mechanism with forced convection.

What would be the suitable wall function and the turbulent model to go with, Re is around 4000-5000.

Thank you.
Dasith

For natural convection and especially for the low Reynolds numbers you want to simulate, a transition model is required, like kOmegaSSTLM:

https://turbmodels.larc.nasa.gov/lan...nter_4eqn.html
https://www.openfoam.com/documentati...megaSSTLM.html

Also, it is pretty much mandatory to have a wall-resolved mesh, that is y+ < 1.
Wall functions are standard, I suppose nutUSpaldingWallFunction for nut, kqRWallFunction for k and omegaWallFunction for omega.

Both those are the reason for my failed attempts, to answer my own question from the last century...

Thank you for the links and description provided. I will give it a go