Heat conduction with GEOM

[ESett5555]

Hi everyone,

I have made a few tests regarding heat conduction in GEOM obstacles, and I want to share with you what i have found while doing these tests. All of them are made with the help of BlenderFDS.

Each test looks the same:

• I have a 4 x 4 x 2.5 meter mesh, with a burner (HRRPUA=1000) on a box at z=1.01 m and a rectangular steel 4 x 4 m roof above (centered at z=1.8m), made using GEOM.
• My cells are cubic with a 10 cm side.
• The SURF line on the roof is always specified with BACKING='EXPOSED'.
• All exterior walls (except the ground) are set to be adiabatic, with a &SURF ID='ADIABATIC', ADIABATIC=T/ and a &VENT line.
• To make the roof heat faster, the MATL is set with very unrealistic values: &MATL ID='STEEL', DENSITY=7900, SPECIFIC_HEAT=0.005, CONDUCTIVITY=15000.0 / (I know that this is probably not the best way to do it).

The only things I change between simulations are :

• The "real" thickness of the roof
• The parameter THICKNESS on the SURF line corresponding to the roof which I always set to be equal to the "real" thickness

The goal of these tests was to see when would the top side of the roof (and the fluid above it) would heat.

I found out that the top side of the roof only heats if the "real" thickness (and the parameter THICKNESS) is lower than sqrt(3) x 10 centimeters (approx. 17.32 cm). Does someone knows why this happens?

I have also tried to reduce the size of the gas phrase grid cells. If I set it to be 5 cm, the top of the roof only heats if the thickness is lower than sqrt(3) x 5 centimeters (approx. 8.66 cm), so it seems to me that it is linked with the size of the gas phrase grid cell. Would it be something with FDS setting BACKING='VOID' like it is explained in paragraph 8.3.3 of the User Guide?

If someone knows how it works I would be interested to understand how this works, and i would be happy to share others similar tests that I have been doing on this matter (which do not deserve a special discussion).

Thank you and have a great day.
Elias

One of the .fds I used :
GEOM_173milieu_conduction_mur.txt

Edit : I have also tried with a tilted roof and the limit of sqrt(3) x cell size do not change.

[marcosvanella]

Hi @ESett5555,

Would it be something with FDS setting BACKING='VOID' like it is explained in paragraph 8.3.3 of the User Guide?

Correct, in this case the diagonal cell size is used. If the distance is larger than the cell diagonal size then BACKING is set to 'VOID'.

[ESett5555]

Hi @marcosvanella ,

Thank you for your answer !

Then how is the solid distance calculated?

Let's take the example of a GEOM 2D triangle (in black), which is divided into faces separated by vertices (in red). How is the solid distance corresponding to face F calculated? Is the distance the green line? the yellow line? the orange line? or any other?

[marcosvanella]

Let me look at this, we should be able to make the BC VOID when the distance between surface and back surface is sqrt(3)*DX larger than the THICKNESS provided in SURF, instead of the hard threshold sqrt(3)*DX. That should give more flexibility.

[ESett5555]

It would be great, and it would match the way it works with OBST. Thank you !

[marcosvanella]

@ESett5555 do you know how to compile the code? If so do a repo update and recompile. Your case should pick up exposed back walls now. Also, you should be able to run with multiple meshes, picking up back wall bcs across meshes. Let us know if you see any issues.

[ESett5555]

Unfortunately, my workstation where FDS is installed is not connected to internet so I can't compile the code.

I tried to install the latest nightly build on [https://github.com/firemodels/test_bundles/releases/tag/FDS_TEST], but it is version 6.11.0 related to commit 280fc88 (if I understand it correctly), so the code modifications you made are not available.

I will try to see if i can compile the code on another workstation and wait to download a release when it is available on my primary workstation.

Thank you and have a great day!

[marcosvanella]

It will be the normal going back (green). A ray is cast in the normal direction to the face you are looking a back face for, into the geometry triangulation.