Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Heat Transfer with fins / contact resistance

Please login with a confirmed email address before reporting spam

If anyone can help here, I need to construct a model consisting of a 100W (constant) transistor and an aluminum heat sink directly on top of it with four cylindrical fins. I was able to construct the geometry and got the Temperature distribution within the transistor to plot correctly.

The boundary condition between the transistor and the aluminum block is a contact resistance of 10^-4 C/W.

I can't find how to add this contact resistance. My aluminum block just shows a constant temperature distribution throughout and doesn't appear to be connected to the transistor properly. I made a union between the two parts but it still doesn't work properly.

Any help would be great on how to add a contact resistance between the two parts.

Thank you!

Andrew E

7 Replies Last Post Nov 20, 2009, 3:03 a.m. EST
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 18, 2009, 4:16 p.m. EST
Hi

I believe that you can do this correctly if you use the [htgh] physics (application mode), you need to activate the interiour boundaries, and select the Highly conductive layer tab, and apply your contact boundary conditions there.

Try it out on a simple model that you can check analytically, and why not post a reply back so we all can gain from your experience.

Good luck
Ivar
Hi I believe that you can do this correctly if you use the [htgh] physics (application mode), you need to activate the interiour boundaries, and select the Highly conductive layer tab, and apply your contact boundary conditions there. Try it out on a simple model that you can check analytically, and why not post a reply back so we all can gain from your experience. Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 18, 2009, 5:29 p.m. EST
Thank you Ivar. I believe I have it running correctly now.

The issue was that under Physics>Boundary Settings, under the Boundaries tab, the surface I wanted was grayed out and I couldn't specify any properties.

But, under the Groups tab, I could select the layer as a highly conductive layer and give it a thermal conductivity of 10000 corresponding to a contact resistance of 10^-4.

It appears to be working correctly and there is definitely heat transfer going between the transistor and heat sink base now.


Andrew
Thank you Ivar. I believe I have it running correctly now. The issue was that under Physics>Boundary Settings, under the Boundaries tab, the surface I wanted was grayed out and I couldn't specify any properties. But, under the Groups tab, I could select the layer as a highly conductive layer and give it a thermal conductivity of 10000 corresponding to a contact resistance of 10^-4. It appears to be working correctly and there is definitely heat transfer going between the transistor and heat sink base now. Andrew

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 18, 2009, 5:36 p.m. EST
Hi

if the boundary is "greyed" it's because it's an interiour boundary and by default it is set to "continuity".
To get access to the interiour boundaries you need to select the tick on the bottom of the boundary selection tab (not the groups), it's side by side with he "select by group tick"

Good luck
Ivar
Hi if the boundary is "greyed" it's because it's an interiour boundary and by default it is set to "continuity". To get access to the interiour boundaries you need to select the tick on the bottom of the boundary selection tab (not the groups), it's side by side with he "select by group tick" Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 18, 2009, 6:16 p.m. EST
I tried changing the interior boundary properties, and it lets me. But, then the solver no longer works and it gives me this error:

Error: 6170
Failed to evaluate variable.
- Variable: contact_cp1
- Geometry: 1
- Boundary: 6
Failed to evaluate operator Jacobian
- Operator: if
- Geometry: 1
- Boundary: 6


I did group the two parts using a union pair. Could it be working correctly when I set the conditions under the group tab? It appears as if it does and I don't get the error.

Thanks,
Andrew
I tried changing the interior boundary properties, and it lets me. But, then the solver no longer works and it gives me this error: Error: 6170 Failed to evaluate variable. - Variable: contact_cp1 - Geometry: 1 - Boundary: 6 Failed to evaluate operator Jacobian - Operator: if - Geometry: 1 - Boundary: 6 I did group the two parts using a union pair. Could it be working correctly when I set the conditions under the group tab? It appears as if it does and I don't get the error. Thanks, Andrew

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 19, 2009, 3:24 p.m. EST
Hi Andrew

I would use standard mode (not assembly mode, nor identity pairs), except if you are doing contact physics, or some nice multibody simultions.

If you draw two rectangles adjacent to each other or even overlapping partly, the common border will be defined as an interiour boundary with the "geomanalyse" command (in the GUI it's when you go from geometry editing mode to sub-domain, boundary, edge or point mode, in V4 beta its when you hit the "Finish" flag at the end of the geometry section). You might notice that the first switch of mode requires some time as the programme analyses the geometry, groups items and generates internal pairs, not external ones as you get in "assembly mode".

Then to work on this internal boundaries, you must activate them specifically, otherwise they are greyed out and by default set to "continuity" that is identical values for "up" and "down" side of the boundary.

Reread the chapter in the user guide about the geoemtries, and try a couple of examples, this helps a lot to understand some of the numerous subtilities of COMSOL. In fact its not really subtilities, it's just applying the correct physics to your model, but it is not always very obvious to link a GUI selection to the underlaying physics.

Good luck
Ivar
Hi Andrew I would use standard mode (not assembly mode, nor identity pairs), except if you are doing contact physics, or some nice multibody simultions. If you draw two rectangles adjacent to each other or even overlapping partly, the common border will be defined as an interiour boundary with the "geomanalyse" command (in the GUI it's when you go from geometry editing mode to sub-domain, boundary, edge or point mode, in V4 beta its when you hit the "Finish" flag at the end of the geometry section). You might notice that the first switch of mode requires some time as the programme analyses the geometry, groups items and generates internal pairs, not external ones as you get in "assembly mode". Then to work on this internal boundaries, you must activate them specifically, otherwise they are greyed out and by default set to "continuity" that is identical values for "up" and "down" side of the boundary. Reread the chapter in the user guide about the geoemtries, and try a couple of examples, this helps a lot to understand some of the numerous subtilities of COMSOL. In fact its not really subtilities, it's just applying the correct physics to your model, but it is not always very obvious to link a GUI selection to the underlaying physics. Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 19, 2009, 8:05 p.m. EST
Thank you, that helped a lot. The boundary should be set up properly now, but my temperatures are really high. I have the chip modeled as a silicon chip with the heat source set to 100W/(0.016*0.016*0.002)m^3.

But my temperatures range from like 3500K to 3580K. This seems way too high. I am trying to figure out what to put in the Production/Absorption coefficient box. It is currently set to 0. It is a uniform 100W transistor, so maybe there is an easier way to just specify 100W uniform heat generation?

Perhaps it would better be modeled with just a specified temperature instead of as a heat source?

Andrew
Thank you, that helped a lot. The boundary should be set up properly now, but my temperatures are really high. I have the chip modeled as a silicon chip with the heat source set to 100W/(0.016*0.016*0.002)m^3. But my temperatures range from like 3500K to 3580K. This seems way too high. I am trying to figure out what to put in the Production/Absorption coefficient box. It is currently set to 0. It is a uniform 100W transistor, so maybe there is an easier way to just specify 100W uniform heat generation? Perhaps it would better be modeled with just a specified temperature instead of as a heat source? Andrew

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 20, 2009, 3:03 a.m. EST
Hi

you shouldbe able to do a simple hand calculations to give the order of magnitudes, usually use MaplSim and Maple for that, these are good support tools for FEm model verifications.

100W in such a small volume is a lot, contact condusiton is essential but 3500K is really "lighting up" was it a light bulb you wanted to simulate ;)
more serious, if you ahve more than 300°C diffeence you start to get radiative exchanges if some importance too, and as tat follow a DT^4 law its very non linear, but also often tricky to calculate as you need view factos for the surface exchange, so try to avoid taht, in the beginning, Radiation can anly give you some extra heat dissipation, so youshould consider ignoring radiataion as a worst case

Good luck
Ivar
Hi you shouldbe able to do a simple hand calculations to give the order of magnitudes, usually use MaplSim and Maple for that, these are good support tools for FEm model verifications. 100W in such a small volume is a lot, contact condusiton is essential but 3500K is really "lighting up" was it a light bulb you wanted to simulate ;) more serious, if you ahve more than 300°C diffeence you start to get radiative exchanges if some importance too, and as tat follow a DT^4 law its very non linear, but also often tricky to calculate as you need view factos for the surface exchange, so try to avoid taht, in the beginning, Radiation can anly give you some extra heat dissipation, so youshould consider ignoring radiataion as a worst case Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.