[Tobi]

Hey foam community,

I might ask you an interesting question. I attached a case (nothing special) a room with a window. The window does not play a big role right now. The top wall (ceiling) is heated to 300 K. The bottom is set to 295 K. Hence, I expect a uniform temperature layering from bottom to top and almost no flow. The only interesting point is that the room is a closed volume (no inlets/outlets). The facts are:

• Without turbulence model the case crashes after two/three iterations
• With turbulence model it works (guess due to the artificial diffusion)
  • However, the temperature rises - for any reason > 300 K (around 380 K) - no Idea why that happens.
  • Solution is not physical (at least the temperature does not make sense at all)
  • Schemes are upwind
  • Mesh is pure hex-mesh (no orthogonality)

Even with 10 years FOAM experience, this simple test case makes me getting crazy but I guess I just made some small mistake. An interesting point is, that p and p_rgh is initialized by 1e5 Pa but the results show me 0 Pa. Is it related to the closed volume? Probably it shouldn't. At least the p field should have 1e5 Pa as we use it for calculating the density (even though I use incompressiblePerfectGas )

Case is given for v8.
https://Holzmann-cfd.com/forum/room.tar.gz


Any ideas?

[NiklasW]

Have you tried to use a different equation of state?
I haven't looked deeply into your case, but at first glance I'd say incompressible perfect gas might cause some problems in a closed system, because it calculates density independent from pressure.

[Yann]

Hi Tobias,

I think your 0 Pa pressure is related to the pRefValue in fvSolution:

SIMPLE
{
    nNonOrthogonalCorrectors 0;
    pRefCell        0;
    pRefValue       0;

   [...]
}

Since you have a closed volume there is no pressure value defined on any patch and the solver should look for a pRefCell and pRefValue in fvSolution.

Not sure how it affects the solution though.

Yann

[Tobi]

Hey both.

• the incompressiblePerfectGas was the mistake (my fault) - thanks for the hint and you are totally correct. This will destroy the solution
• the pRef data in the fvSolutions is correct too. However, after changing the EOS, these data are not used anymore.

So ... thank you!

[Tobi]

Okay now we are ready to go to discuss. By using the case provided and changing the EOS to perfectGas, everything works (even though the velocity field looks weirdo - probably based on the steady-State analysis).

However, if we deactivate the turbulence model and calculate in a laminar manner, the temperature goes to 350 K. I have no idea why. That`s actually something I am not sure about.