How do I model thermal contact between two bodies whose temperatures at the interface between them is not known a priori? If I define a heat transfer coefficient as a boundary condition at the interface without specifying the external temperature, a value of 0 is assumed by the software. This results in two disconnected thermal problems.
Apparently, the interface should not be defined as a boundary condition. How then?
I attached a small example of what I tried. An aluminium and a copper block of 2x2x5cm, with common interface and a heat transfer coefficient of 2.5 MW/K/m2. One side is kept a 20C, the other at 30C, but the interface ends up at 0C.
The image of the temperature distribution shows that the temperature at the interface is 0, though it is not specified as such in the SIF file.
Modelling thermal contact
Modelling thermal contact
- Attachments
-
- contact.zip
- All files relating the example.
- (648.54 KiB) Downloaded 466 times
Re: Modelling thermal contact
Hi,
The solver thinks you have a degenerate 3D model with no volume.
For 2D analysis, node points should be placed on the xy-plane.
In your case, you can alternatively change the coordinate map in case.sif as follows:
If you are using ElmerGUI, the map can be found from Model -> Setup... -> Coordinate Mapping
The solver thinks you have a degenerate 3D model with no volume.
For 2D analysis, node points should be placed on the xy-plane.
In your case, you can alternatively change the coordinate map in case.sif as follows:
Code: Select all
Simulation
...
Coordinate Mapping(3) = 2 3 1
...
End
Re: Modelling thermal contact
That is an answer to my other question, but it works. Thanks.
Re: Modelling thermal contact
Hi,
there is an example of such a case in source distribution in folder "fem/tests/HeatGap".
Firstly, you need to define the contact boundary as having double nodes using ElmerGrid
option '-discont id' (for example) , as in
ElmerGrid 2 2 mesh -discont 1
if you have a Elmer mesh files is 'mesh' and your contact boundary id within the mesh
files is 1.
Then, in the .sif file you specify 'Heat Gap = Logical True' in addition to the 'Heat Transfer
Coefficient = value'.
Hope it helps, Juha
there is an example of such a case in source distribution in folder "fem/tests/HeatGap".
Firstly, you need to define the contact boundary as having double nodes using ElmerGrid
option '-discont id' (for example) , as in
ElmerGrid 2 2 mesh -discont 1
if you have a Elmer mesh files is 'mesh' and your contact boundary id within the mesh
files is 1.
Then, in the .sif file you specify 'Heat Gap = Logical True' in addition to the 'Heat Transfer
Coefficient = value'.
Hope it helps, Juha
Re: Modelling thermal contact
Hello,
I think it could be possible as well to model the problem with heat flux BC, because your BC is continuous heat flux that is as well the natural BC so you don't need to prescribe nothing at the boundary, just specify that flux BC is applied.
BR, Martina.
I think it could be possible as well to model the problem with heat flux BC, because your BC is continuous heat flux that is as well the natural BC so you don't need to prescribe nothing at the boundary, just specify that flux BC is applied.
BR, Martina.