No success in the validation case of DIN EN 1993-1-2 with FDS

[GebChr]

Hello everyone,
I would still consider myself a newbie to FDS and I am trying to improve and test my skills for an upcoming job.
I came across simple validation examples in the DIN EN 1991-1-2 National Annex (Germany). The examples are intended to validate calculation programs for use in fire protection concepts in Germany.
The task in that example is: Heat transfer in multiple layers (steel hollow cross-section with filling)

I assumed FDS would be able to find the required solutions, but I'm far from it. Attached is the task (in excel and pdf) and my code. I'm using Pyrosim as a platform, so I hope the code can be used without problems.
I am wondering whats the problem and would love to find a solution. Or get the Info that FDS is not suitable for that kind of taskswhich would be a surprise for me.

After 30 minutes, the temperature should be between 337.1 and 343.9°C in the center of the filling. However, I'm only getting a temperature of approximately 231°C, which is significantly different. Does anyone have any new approaches for me or find any errors?

I tried (in desperation) already to change between LES and DNS;
changed my mesh size;
Changed Wall Increment and radiation increment and also changed the TIME_STEP_FACTOR.
When i tried to test with Fourier Number Constraint on Misc line i got an error, so couldnt try that one.
Changed between a multilayer obstruction and seperate obstructions.
A change of the absorption coefficient didnt change to the better.

As an additional question. As soon as i changed my mesh size to < 0,05 i also got an error for my solid temperatur profil (ERROR(429): PROF 1 requires repositioning) and could not find any valid position. Any Idea aswell how to solve that?

Exampel Case 3_FDS Code.txt
Exampel Case 3_task.xlsx
Example case3_task.pdf

Thanks a lot to anyone taking some time :)

[drjfloyd]

You are not replicating the test problem. The test problem is idealized with the gas temperature adacent to the surface is always 1000 C. In your inputs, the moment the simulation starts you will convect heat from the gas to the object, and the gas temperature adjacent to the object is going to drop. This will result in a flow downward along the sides of the object. This will bring back warmer gas from the surroundings but you are not going to have a 1000 C temperature at the block. In the image below plut a slice file of temperature at Y=0.1 m. You can see at 10 s the temperature adjacent to the block has dropped ~20 %.

In addition to this the FDS heat conduction you are using is one dimensional. However, this problem is a 2D problem where the center of the insulation is getting heat transfer from all sides of the block. You are not modelling the additional dimension and over the long term will have reduced heating as a result.

[GebChr]

Thank you for your detailed answer! I really appreciate it.

As an additional question before I put more work into it. Can this test case be recreated with FDS? As 3D heat conduction is implied in the beta phase.

The thing about heat conduction annoys me, because I've already read about it and could have figured it out. Damn it :D I wouldn't have thought of the first point. Saw a temp drop at my thermocouple but didnt figured out it was a problem.

[mcgratta]

ht3d_filled_column.txt
If you are just interested in performing the heat transfer through a cross section of the column, this may get you started. Use the latest version of FDS.

[GebChr]

I'm currently mainly looking for problems with predetermined results that fit my upcoming work. Then I try to solve the problem to determine where my shortcomings lie. This will hopefully help me better understand FDS.

The ht3d_filled_column file will help me better understand my mistakes and perhaps find the adjustments. So, thanks a lot!
But to progress, I need problems and models to recreate myself.

Ps.:
It worked great with FDS, but when imported into Pyrosim, I get different results, which I now need to investigate.

[drjfloyd]

Pyrosim runs the version of fds that it installs not the version you have installed.

…

On Fri, May 2, 2025, 06:52 GebChr ***@***.***> wrote: I'm currently mainly looking for problems with predetermined results that fit my upcoming work. Then I try to solve the problem to determine where my shortcomings lie. This will hopefully help me better understand FDS. The ht3d_filled_column file will help me better understand my mistakes and perhaps find the adjustments. So, thanks a lot! But to progress, I need problems and models to recreate myself. Ps.: It worked great with FDS, but when imported into Pyrosim, I get different results, which I now need to investigate. — Reply to this email directly, view it on GitHub <#14593 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADBUZ4I5ITTRXHWHNTOM7R324NFARAVCNFSM6AAAAAB4H5XOAGVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMBRGI4TIOA> . You are receiving this because you commented.Message ID: ***@***.***>

[GebChr]

Holy Moly, thanks a lot,
I knew but never imagined that between the Versions (minor release) is such a huge difference between the 2 FDS Versions.
(Pyrosim version: FDS Version 6.9.1 and later in Comparison to FDS 6.10.1).
Just installed FDS Version 6.9.1 and get same results as Pyrosim (which than makes totaly sense)).

FDS 6.9.1 - temp (Boundaries wall temp) around 520 °C at end of simulation (36000 sec);
FDS 6.10.1 - temp (Boundaries wall temp) around 520°C at 15300sec and at around 940 °C at end of simulation (36000 sec).
To be honest I find the results of FDS 6.9.1 strange.

Lesson learned, thanks a lot!!!

[drjfloyd]

HT3D is a beta feature still under development.

[mcgratta]

We are making changes to HT3D often. Most of these changes have to do with functionality; that is, the input parameters and their interpretation. In your case, we have not tried to do 2-D heat transfer where the top and bottom of the obstruction are essentially mirror boundaries. In my example, I explicitly added an adiabatic boundary to mimic this effect. This could be useful in comparing with 2-D analytical or benchmark cases.

[GebChr]

Thanks a lot to both off you.

[mcgratta]

If you succeed in setting up a case that models this standard exercise, could you contribute the input file so that we can use it as a verification case.

[GebChr]

If I am successful, I would be very happy to contribute.

[mcgratta]

This Discussion thread inspired me to take a look at some simple verification cases for a 2-D or 3-D heat conduction solver. I remember that about 10 years ago I co-authored an appendix to an SFPE Standards document. This appendix contains 16 heat transfer problems with which one can verify the accuracy of a multi-dimensional solid phase heat conduction solver. The cases and some explanatory notes are found here in the file named SFPE_Verification_Cases. Some of these cases originated in the German standard, and some were developed by Ulf Wickstrom in developing TASEF. I have begun to add some cases to the FDS Verification Guide. The input files for Cases 6, 7, 8, 9, 13, and 16 can be found here. The cases are documented in the FDS Verification Guide, which should be available the next time the nightly build updates.