FDS Heat Transfer 1D vs HT3D

[oriolc...@gmail.com]

 

Hi!

I am currently working on a model consisting of a metallic tray with some components inside and a fire just underneath. I tried to implement two heat transfer methods available in FDS; the default 1D heat transfer and the Solid Heat Transfer (HT3D) which is in Beta testing on the FDS version I’m using (6.6.0).

The case “METAL_TRAY_FIRE_1D.fds” doesn’t have HT3D implemented and “METAL_TRAY_FIRE.fds” does.

The results I am obtaining from both simulations are different, since the tray is not heating at the same speed and intensity. The 1D case works as expected but in the model with HT3D implemented we can see that the tray takes x4 the time to heat at the same temperature as the 1D case.

The mesh used is divided into 32 MPI_PROCESS. A single mesh is also available and it’s found commented just before the former one.

Can you notice any mistake in the input? How could this difference be explained?

 

Thank you very much!

O.

[Kevin]

It is really difficult to examine a case like this when you have 32 meshes. Can the scenario be simplified for a single mesh, and a very simple heat transfer example?

[oriolc...@gmail.com]

There is a single mesh as well commented in the code. The only thing that has to be done is uncomment it and comment the 32 lines from the original mesh.

[fde]

Have you tried running in FDS 6.7.0?

[TimoK]

Your 1D case has 0.0015 m thick steel (1.5 mm). And your 3D model has 0.02m thick steel (20mm).

Do you think that these should heat up similarly? Make a simple test. Modify your 1D case MATL line for steel, set the thickness to be 0.02m. See, if you get something similar to your 3D case. If so, then your 3D case might be set up correctly, but it is not representing the case that you are trying to model. The 3D model is a model of really thick steel trays (2cm thick).

FDS 6.7.0 user guide:

If two-way coupling with the gas phase is desired, then the
SURF associated with the OBST face should have HT3D=.TRUE. and, generally, no other specified thermal
boundary condition. For example,

&MATL ID='steel', .../
&SURF ID='s1', HT3D=.TRUE./
&OBST XB=..., HT3D=.TRUE., MATL_ID='steel', SURF_ID='s1'/

So, the thickness of your steel is the size of your OBST that is 20mm in the z-direction.

Your MESH definitions are not nice, you have dx=0.1m, dy=0.05m, dz=0.02m so the aspect ratios of your cells is not good:

FDS user guide:

Numerical Instability Errors:

* mesh cells that have an aspect ratio larger than 2 to 1

TimoK

[o...@aquacoustics.biz]

I haven't had time to run your models just yet, but looking at the TRAY_1D model you have specified the INSULATED &SURF parameter for the thermal boundary of CABLE and CABLE_2.  Please refer to the FDS Users guide to understand the consequences of this definition.  I suspect that you want to be defining these as EXPOSED.

t.

[oriolc...@gmail.com]

Dear Timok,

Thank you very much for your answer, I'll modify the mesh aspect ratio and the 3D model steel plates so they have a similar thickness to the 1D case.

[oriolc...@gmail.com]

Hi!

No, I haven't installed the new version yet.

O.

[oriolc...@gmail.com]

Hi!

Thanks I'll take a look at that too!

O.

[o...@aquacoustics.biz]

Okay, I have had a look at your model and what you have rediscovered with your 1D model is a fundamental limitation of 1D conduction implementation in FDS.  Again referring to the Users Manual, Section 8.3.3, there are no gas cells between the cable plastic and the metal tray at the the base and sides of the section.  So the 'wall' (comprising four single cell obstructions in the Z direction, and many many cells in the Y direction) is now more than 1 cell thick.  So the backing of steel tray seen from the fire defaults to an air gap at ambient temperature.   The consequence is that no heat transfer occurs between the steel and the plastic.  

If you run your 1D model at a reduced resolution with a fast ramp to a 1 MW/m2 fire (so we can see stuff happening quickly), put a temperature slice through the X axis, and examine the slice temperature (remember to to turn on solids in the boundary dialogue) you will see the effect clearly.  The back of the steel, where it contacts the plastic wall is stuck at 20°C.  Where the steel backing is exposed to air above the cable, the inner surface of the steel heats up with convective heat transfer to the air and radiative heat transfer to visible obstructions.

Attached is a somewhat simplified version of your 1D model to 'play' with.

Although I haven't looked at your 3D implementation this is probably the reason for the significant difference that you are experiencing between the two models.

t.

  

If you rn   

On Tuesday, 10 July 2018 04:07:27 UTC+12, oriolc...@gmail.com wrote:

[oriolc...@gmail.com]

Thank you for your time. I will take a look and determine what's the best option in my case.

O.