How to reduce runtime for an HT3D + gas-phase cavity model with thin corrugated steel geometry?

[JetLieju]

Hello,

I am trying to model heat transfer through a composite wall assembly in FDS. The construction consists of:
a steel front plate exposed to an ISO 834 fire curve;
an air cavity where heat transfer should occur through radiation and convection;
a corrugated steel sheet / profiled steel element with small air voids inside the profile;
two insulation board layers on the cold side.

The main purpose of the model is to study the temperature development and heat transfer through this construction.

1. the fire-exposed steel plate heats up;
2. heat is transferred through the air cavity by convection and radiation;
3. the corrugated steel profile heats up;
4. heat is transferred through the contact/interface with the insulation boards;
5. temperatures are measured at the cold side and inside the board layers.

Because the steel profile has 1 mm thick parts, I currently use a 1 mm mesh. The model is simplified in the y-direction, but because of cell aspect ratio issues I am not sure how much I can simplify it without making the gas-phase calculation unreliable.

My main questions are:

1. How can I make this model faster?
   I considered using multiple meshes: a fine 1 mm mesh around the corrugated steel profile and coarser 10 mm meshes in the larger air cavity, the front plate and the insulation region. However, if I keep a 1:1:1 cell ratio, the finer meshes would need more cells in the y-direction, which can result in more total cells than using one uniform 1 mm mesh. Is there a better modelling strategy for this type of problem?

2. Is my modelling approach physically reasonable?
   I am trying to combine gas-phase heat transfer in the cavity with HT3D conduction in the steel and insulation. The cavity should remain gas-filled, including the small voids inside the corrugated profile. The solid parts should use HT3D so that heat can conduct through the steel profile and into the insulation boards. Are there important limitations or mistakes in this approach?

3. Could this cavity setup lead to pressure-zone problems after running for some time?
   The air cavity and the small voids inside the corrugated steel profile are mostly enclosed by solid obstructions and side blockers. I want the cavity gas to participate in convection and radiation, but I am not sure whether the current setup provides a physically/numerically acceptable pressure path. Could this lead to pressure-zone issues, pressure build-up, or numerical instability after the model has run for a while? If so, what is the recommended way to vent or define the cavity?

4. How should I handle the thin blocking objects at the sides?
   I added side blockers to prevent gas bypass around the cavity. Their only purpose is to close the side paths, so that the gas phase participates inside the cavity and profile voids, but does not flow around the wall assembly and heat the cold/back side from outside. So now its not fully correct yet but I had problems earlier with putting obst on the boundary.
   What is the recommended way in FDS to define such side blockers without adding unnecessary HT3D solids or affecting the thermal response?

5. How can I verify device placement in Smokeview?
   I could no t see my devices in Smokeview as the green spots like I saw before, so it is difficult to check whether the wall-temperature and inside-wall-temperature devices are placed on the correct surfaces. I need temperatures on the exposed parts of the corrugated steel profile, on the recessed parts, in the gas cavity, and at/inside the insulation board layers. What is the recommended way to verify these device locations before running a long simulation?

The reason I want to resolve this now is that I still need to add temperature-dependent specific heat and thermal conductivity for several materials. Therefore, runtime and model correctness are the most important issues. I want to avoid running the model for hours and only later finding out that the geometry, HT3D setup, or device placement is not correct.

Any advice on simplifying this model while keeping the relevant heat-transfer mechanisms would be very helpful.

17june.txt

[Er9y714]

Item 5: you can begin with simple case in which you can verify your inputs and outputs.
Do you see the devices in the output files (_devc.csv and CHID.out file device summary)?

[JetLieju]

Thanks for your reply, I already made models with all the individual specifications and that worked but when i started combining Ht3d with gas in a cavity the problems started arising. also a smaller case worked with those two so I think I build it up a bit already? At least I knnow under the previous circumstances what happend with outputs
when I changed inputs.

[drjfloyd]

Is HT3D really needed here? You are only showing a couple of corrugations but for any real application you would have a much larger area of the assembly. With a more or less uniform temperature bottom plate, then central region of that assembly might not be seeing large gradients in heat flux over its surface. If that is the case, then 1D may suffice provided you model a large enough portion of the assembly.

[mcgratta]

I ran the case called 17june and the HT3D is not adding significantly to the CPU time. One thing to consider is that your time step is bound by the Von Neumann (VN) condition, not the Courant (CFL) condition. That is, diffusion is dominating the transport, and you have 1 mm cells.