Divergence in symmetrical thermal model in Workbench

Hello everybody, 

I hope I am right here and you can help me with my problem  

I guess, the basic problem is rather hard to describe, but I'll try it anyways:
For a project I want to simulate the thermal behavior of a heat sink. The model consists of a chip, that emits a constant heat flux, a heat sink made from copper and the surrounding air. Since the chip and the the heat sink are both symmetrical, I wanted to simulate only one quarter of the problem. 

So what I did was, that I modeled one quarter of the chip and one quarter of the air. Afterwards I applied the loading conditions, which is the constant heat flux. To make things easier, I set the conductivity at the contact area to a constant value (to simulate an easy convection) and the temperature at the outer shell of my air as a constant value. 

Afterwards I simulated the whole thing for the element sizes 15, 20 and 25 mm for the heat sink and 120 mm for the air. For the contact size, I always chose the size of the elements of the heat sink. 

Now comes the part that I don't understand: For some reason, the temperature diverges, when I refine the elements, and I really can't think of a reason why. Is there something wrong with my mesh, or the mesh that is created by Workbench? If I simulate half the chip, I get converging temperatures.

I attached a screenshot of the heat sink for a better understanding

I hope you can give me a tip, where my problem might be. Thanks in advance! 


  • peteroznewmanpeteroznewman Member
    edited November 2018

    Hello StayFun,

    I'm inserting the image you attached above since ANSYS Employees on this site are not allowed to open attachments.
    Please insert images inline in future posts.

    A simpler model would be a mechanical Steady-State Thermal analysis. That way, you don't need to mesh the air, but you do have to provide the convective film coefficient.  If you don't have any idea of a good value for the convective film coefficient, then a model of the air and the heat sink can calculate that value, which is maybe why you have the model described above.

    You might get a better mesh if you switch to Multizone Method on the solid bodies.  The mesh in the above image looks too coarse, espcially for meshing the air, which size does that image represent?

    You don't show the air mesh, but it would need an inflation layer. Did you include an inflation layer in the air mesh?


Sign In or Register to comment.