[a_habib]

Hello foamers
I'm trying to simulate a non-isothermal flow in a room.
Using buoyantBoussinesqSimpleFoam, I created three meshes and used kEpsilon, kOmegaSST, RSM on these meshes, but I've got confusing results.

- Using the coarse mesh, the simulation overestimated the temperature profiles compared to an experimental results recorded in several locoation in the domain, even after reducing the relaxationFactors.

- Using the medium mesh, the results had a good agreement with the experimental results .

- BUT when I used the fine mesh, the simulation underestimated the temperature profiles even after reducing the residualsControls.

The average relative errors using kEpsilon are (24.7% for coarse mesh, 6.34% for medium mesh and 18.8% for the fine mesh)

I 've got the same behavior using the three turbulence models mentioned above, you can see an example in the attachment picture

I've tried to find an explanation to this results, but I came up empty handed, so if any one can help I'll appreciate that very much.

Regads.

[piu58]

As you can see in your simulations the results come closer to a final result when you refine the mesh. You may try an even finer mesh and you will see that it represents more or less the solution of the now finest mesh.

You have to accept that this is the computational result. I checked bBPF against a theoretical result so I believe in this program if temperature is the only source of flow. (I have problems with additional external flow, but that is another story).

I think you have to look for the deviations elsewhere. I would suspect first the material constants.