[macRC]

Hello everybody,

Hope you all are fine.


Usually I do thermal simulations for electronics cooling (CHT simulations). This means, we can simulate the overall thermal behavior of the electronics. For that, we have a quite efficient and proper workflow.



An important detail is often the thermal resistance between two parts. E.g. how is the PCB (Printed Circuit Board) mounted on a heatsink. Due to the roughness of the surfaces and the not perfect planarity, there is always a small gap between two parts, usually in the range of 30 µm air.
With the assumption of a homogenous thermal resistance of 30 µm, we can achieve a very good agreement between simulation and measurements.



The next target now is to model and simulate the 30 µm air gap more properly, as we know, the gap is not constant but can vary in a certain range.



For doing so, I think a common approach is a one way coupled thermal stress simulation:
- - Performing a thermal simulation as usual in order the get the thermal behavior of the parts
- - Map the result and perform a structural simulation considering the different coefficients of thermal expansion (CTE) and observe the deformation of the parts


Now, the gap is not constantly 30 µm any more, but can vary in a range of let’s say 0 – 60 µm due to the deformation (it´s now a function of the place).
Since the gap can have a big impact on the thermal results, I would like to perform another thermal simulation, with the new “air gap function” and not with the homogeneous 30 µm. Thus I would expect slightly different thermal results. This new thermal result will end in a different deformation and air gap function respectively. If this works, I could do a slope thermal, stress, thermal, stress and so on until there is convergence.


Unfortunately the implementation of all that is not so easy. I did some investigations with different approaches, but the success is negligible so far


Do you have some ideas or hints of performing such "2 way coupled thermal stress simulations"?
I use Star-CCM+ and so far there are only two mechanical interfaces – bonded and Small Sliding Firctionless.

Attached you can find a sketch of the domain.
This example shows a simplified LED module mounted on a heatsink.
Thank you very much in advance.
Kind regards
macRC