Beta Geom / Sphere / Cylinder

[Koranten]

Hi,
I've been working with FDS 5 for quite a long time and I'm finally considering moving to FDS 6 for its new GEOM beta capabilities but after reading the dedicated chapters in the documentation, some things are quite unclear to me.
If I create obstacles with the new sphere model, will they be computed in a 2-way coupling, i.e., will they suck/give energy to the surrounding fluid, or will they only suck energy in 1-way like a sensor device &DEVC would?
To be more clear I'd like to know if I can model a "bed of rocks" with spheres that would store and release mass thermal energy.
If I understand correctly, GEOM beta obstacles cannot use the new 3D conduction capability, so if I model a cylindrical wall, the heat transfer inside will only be 1D, in the normal direction, is this correct?
And for the spheres, I understand they would be 1D in spherical coordinates, so the heat can go inside but not ACROSS the sphere, or am I wrong?
Thanks.

[marcosvanella]

You would be able to do 1D heat transfer using the back of the sphere temperature (across the sphere), but on a model with three dimensions it would not be a spherical coordinate model, it would be 1D using the cross section you have in the surface face. To do 3D heat transfer on these we would need a 3D unstructured solver in a mesh inside the geometry hooked to the gas-phase, something we don't have at the moment.

[marcosvanella]

1. Yes, see the case fds/Verification/Heat_Transfer/back_wall_test_2.fds
2. Yes they will affect/block the air flow.
3. Do you mean a sink term in the 1D heat equation for each face in the sphere?

[Koranten]

I mean like if the spheres could dissipate energy, like there was a Joule effect in them.
I saw it's possible in cable as targets in FDS, but is it also possible in the spheres?

And what about the heat transfer in a GEOM cylinder wall obstacle? 1D in the normal direction?

[marcosvanella]

Yes, the solid 1D heat transfer model is the same in OBSTs and GEOM, you should be able to specify the GEOMETRY, INNER_RADIUS etc. for the 1D heat transfer. This includes the "SPHERICAL", "CYLINDRICAL" models. Note many of these features are hooked up but not exercised regularly. You'll be a very much appreciated beta tester.

[Koranten]

What if I have sphere about the same size as my cells mesh and they are close together? Will the fluid be able to flow between them?
Suppose I have cells mesh of 5 cm and I create spheres of 5 cm not completely packed together but almost. Will the flow go between them? Again, it's to simulate some sort of "bed of rocks", not fully packed, but close.

[marcosvanella]

The resolution is still driven by the cartesian grid. We don't re-mesh at a finer resolution the gas region around the GEOM. Therefore there is no gain from sub cell size features. You would need to use several cells to represent the rocks and also to have resolution of the gas space between them. For that you'd have to do a grid refinement study.
Seems to me that the model you are looking for is a particle model?

[Koranten]

That's what I thought, it makes sense. I'll try the refinement.
What about heat generation ? Not sure you replied on this. Possible to inject W/m3 in the spheres ?
What do you mean by a "particle model"?
Anyway I'll try the spheres and a cylindrical wall when I can.

[marcosvanella]

Yes same way as for !D heat transfer in obstacles you should add heat sources, sinks. See the User Guide sec. 8.3.6.
For solid particles see Sec 15.4 of the User Guide.

[Koranten]

Solid particles are interesting indeed.
They can be static to remain in the same place, correct? They will create like a porous media. Good. The documentation says they absorb but also emit radiation. Good. Is it true for convection too? They absorb energy through convection but can they also release energy through convection ? If yes, I think it's the best way to model of "bed of rocks" indeed. Can they also have heat dissipation? It seems to.

[Koranten]

Particles radiate to the environnement but do they radiate to one another ? I guess not.

[marcosvanella]

Maybe @ericvmueller can clarify this last question.