How to simulate thermal stresses?

[marlonsaveri]

Hi,


When I simulate a simple block, 2D, free, and I apply an uniform thermal condition (cooling in my case), It starts a shrinkage (as we expected) and we can see a displacement in the result.

However, Ansys also shows a stress happening. Ok, One can say there's stress since Sigma = E*Epsilon (Hook's Law), however, since the body is free in all directions, there was no stress. It's just a shrinkage. What can we do to disconsider this stress in the result?

For example:
A material has yield strength = X at 295K and yield strength = Y at 80K (Y>X). So, if I aply a load < X at 295K, this material will be ok.
But, when I simulate the shrinkage and Ansys shows me there's a stress = X-10MPa acting. Therefore, just 10MPa would be enought to damage the body, and it's not true.


Now, the real problem: imagine we have two bodies, of different materials, conected side by side. Since they have different expansion coefficients, now we really expect a stress caused by this difference of shrinkage. I'm interessed in this values, disconsidering that effect Ansys shows about just cool a body.

 

[ChadV]

Is your 2D model using plane stress? If you're using plane strain you will see out-of-plane stress unless you've turned off thermal expansion out-of-plane.

If you are using plane stress, what magnitude of stresses are you seeing? I did a simple plane stress model of a 2D square (100mm x 100mm), structural steel (default properties), cooled to -200C, and the resulting stresses are negligible (order of 10E-6).

 

[marlonsaveri]

Thanks very much, ChadV,

That was the problem. I was using "Plane Strain", than I changed to "Plane Stress" and the problem was solved (and next to the result using a 3D analysis)