Heat transfer 3D with thin surface

[cunco...@gmail.com]

Hi everyone!

I want to simulate the heat distribution inside an lockers.

The heat source is some devices in charge in a box. The lockers's wall is very thin (about 1 to 2 mm). 

I want to use HT3D to simulate the heat transfer from the heat source through the wall (the device is placed on the horizontal wall of the box) to the nearby boxes. But the wall's thickness is to small for the meshing. 

Is there any way to use HT3D for a thin material  layer or to simulated the conduction from a volume with HT3D to a thin layer with HT1D?

Thanks!

[dr_jfloyd]

HT3D only operates at the resolution of the FDS grid. Given the scenario you have described and the wall thickness, the 1D model should be pretty good in terms of accuracy. 

[cunco...@gmail.com]

Hi dr_jfloyd,

Thanks for your answer. My problem is like below.

I'm wondering about the interface between the heat source and the wall. I want to compare my simulation to an experiment.

In the test, we have some thermocouple to  measure the surface temperature of the walls under and beside (non exposed sides) the heat source. 

Thanks!

[dr_jfloyd]

One option is to not use HT3D. The larger the source size is compared to the wall thickness, the less the error is in just doing a 1D solution.   

One option is to define the OBST for the walls to be one cell thick but reduce the density and increase the conductivity. Then you can use HT3D. For example if your wall is 1 mm thick but your grid is 1 cm, then decreasing the density and increasing the conductivity by a factor of 10 would result in the same heat transfer.

[cunco...@gmail.com]

Hi dr_floyd,

I'm running a case test. At least the simulation are running with FDS 6.7  but i'm not sure about my setup. 

My lockers is placed inside a closed room and it has 3 boxes. I have three hot objects inside the medium box. The upper  box and the lower box are blank. All boxed are closed.

So i set four pressures zones : one for each box and one for the room with some leakage surfaces (from boxes to toom and from room to the outer environment)

 I'm testing the HT3D. The lockers's wall are made by steel and the wall's thickness is abour 0.5 mm. in my test case the mesh size is 5 cm, so 100 times larger.

I do as your advice : i decrease the steel's density 100 times and decrease  the heat conductivity 100 times  like below :

&MATL ID='STEEL_2',

      FYI='Drysdale, Intro to Fire Dynamics - ATF NIST Multi-Floor Validation',

      SPECIFIC_HEAT=0.46,

      CONDUCTIVITY=4580.0,

      DENSITY=78.5,

      EMISSIVITY=0.95/

&MATL ID='STEEL',

      FYI='Drysdale, Intro to Fire Dynamics - ATF NIST Multi-Floor Validation',

      SPECIFIC_HEAT=0.46,

      CONDUCTIVITY=45.8,

      DENSITY=7850.0,

      EMISSIVITY=0.95/

I use 'STEEL_2'  for the lockers 's wall. I define the heat source like in a normal FDS simulation :

By example :

&SURF ID='HOT',

      RGB=102,0,0,

      TMP_FRONT=200.0/

&OBST ID='Obstruction', XB=1.05,1.2,0.85,0.95,0.8,0.85, SURF_ID='HOT'/ 

the hot object lays over a horizontal wall :

&SURF ID='S1',

      HT3D=.TRUE./

&OBST ID='Obstruction', XB=0.8,1.2,0.8,1.15,0.75,0.8, RGB=51,0,255, SURF_ID='S1', HT3D=.TRUE., MATL_ID='STEEL_2'/ 

I wonder how FDS will treat the heat transfer through the common surface between the wall and the object  from (like from 1D to 3D): the surface limite by XB=0.8,1.2 ; YB =0.8,1.15, and ZB = 0.8. Does the surface temperature of this region on the horizontal steel plate  is sets to the temperature of the hot object.

********************************

I have another cas test with 1D heat conduction. I have however the same situation :

&SURF ID='STEEL',

      COLOR='GRAY 20',

      MATL_ID(1,1)='STEEL',

      MATL_MASS_FRACTION(1,1)=1.0,

      THICKNESS(1)=5.0E-4/

&OBST ID='Obstruction', XB=1.05,1.2,0.85,0.95,0.8,0.85, SURF_ID='HOT'/  

&OBST ID='Obstruction', XB=0.8,1.2,0.8,1.15,0.8,0.8, RGB=51,0,255, SURF_ID='STEEL'/ 

As i can remember with 1D heat conduction, we can not have two surfaces at the same location ;so the surface which is defined latter overwrites the other. So if i define the steell plate after the hot source there will be no heat conduction from source to the plate below?

 

************************************

And how HD3T calcul the heat transfer coefficient for the convection (I guest that nothing changes : like  in 11.2.2 of user guide : from natural and forced convection ).     

I relly hope to have your advice for my case setup. I will upload it latter 

Thanks for your time!

[cunco...@gmail.com]

This i my FDS HT3d case 

[dr_jfloyd]

I've just realized I didn't think my advice all the way through. Adjusting the density and conductivity fixes heat transfer through the wall, but will break the along wall heat transfer which I should have realized right away. You are stuck with 1D unless you use a small enough mesh. To do this correctly with HT3D you would need to be able to scale each dimension of the OBST independently for HT3D, which isn't something that can be done right now. 

With 1D you cannot stack OBST one on top of another and have heat transfer between them. One thought is you could define a hollow box with thin surfaces that have a low specific heat, a high conductivity, and a large heat transfer coefficient on the inside face. Then put a large particle in each grid cell in the box with the temperature you want. You probably want to define a pressure zone in the box with some leakage to avoid pressure heating. Something like below where B would be the box, P is particles in the box, and W is the locker wall. If all the interior faces of the box including the locker wall have high heat transfer coefficients (you'll need to define the OBST with multiple SURF so from the inside out the wall is high HTC then normal HTC and from the outside in the wall is normal HTC then high HTC. This might approximate having a constant temperature surface at the particle temperature. 

B B B B B

B P P P B

W W W W W