contact coefficient for thermal analysis

[simpledude]

I have a thermal study (SW Sim/Cosmos) in which to begin with I have assumed bonded contact faces for the transfer of heat, essentially 100% transfer (as I understand it). I am now trying to get a more accurate study with inputing a closer to real life heat transfer between faces. Does anyone have a good reference for contact face coefficients? I am using carbon steel, stainless steel, ceramics and possibly teflon and nitrile (seals, not as important).

havin fun, thanks.

 

[salmon2]

First of all, even with bonded contact, you are not getting 100% transfer as far as I know. Because of the mesh, you will have some void. With finer mesh, it will approach to 100%.

But actually this is exactly you will face in real life because surface are never perfect flat or round, etc. But you don't have a control over it, it will depends on the automatic mesher of SW. I am not sure about your problem scope, you just want to have a reasonable heat transfer coefficient to set or this contact with heat transfer is the problem you want to tackle? I have some experience on this topic before, pretty narrowed but deep experience.

 

[simpledude]

Salmon2,
Thanks for your question to clarify what I am trying to accomplish. I would like a reasonable heat transfer coefficient.

Thanks

 

[GRF]

You should get a value that is pretty close to what you will see in the real word without using a coefficient between the two surfaces. That is assuming the contact area is relatively small, 2-3 inches in diameter, the materials are going to be made pretty flat, machined vs stamped or cast.

We usually use a thermal interface material between two parts to take out any surface issues. It those cases I calculated the contact resistance as 1/2 the max flatness differences between the two parts using the thermal values for the thermal interface material.

My results are usually off by less than 5% or for my stuff 2-3 deg C which could be just as much from the measurements as it is from the simulations. I am usually dealing with a 10-20 watt systems and natural convection.

If you are using Cosmos for this test you will have to run the simulation about 4-5 times calculating a new boundary temp after each simulation until it stops changing. I would recommend trying to upgrade to Flow Simulation as it handles the multiple passes for you and will calculate the convection coefficient for you as well. The meshing is also much easier with Flow as it is a finite volume simulator vs finite element like cosmos. It cannot handle any of the stress functions that Cosmos can though.