Correct Structural Boundary Conditions for Symmetric Thermal Stress Model


I am working on a thermal-stress/expansion analysis of a laser diode array using the following half unit-cell geometry which comes from the red outlined portion of the repeated unit cell structure shown at bottom left. This half-unit cell has symmetry down one plane and a linear periodic condition across the top and bottom surfaces, reflecting the repeating geometric unit for stacked laser diode bars. These boundary conditions are shown in the bottom right sketch of the half unit cell model.

The thermal load comes from the results of a steady state thermal analysis and is not my concern here. My major concern is getting the right boundary constraints in the structural model to allow free thermal expansion in the boundaries shown above with the above symmetries while preventing rigid body motion and rotation, as well as mitigating artificially induced boundary stresses from those constraints. I am familiar with the 3-2-1 approach; however, I do not know how it would apply in the case of a one plane symmetry (U_x = 0, U_y = U_z = free) and linear periodic conditions across two other faces.

I would appreciate some tips/pointers. Thanks!



  • sdeogekasdeogeka Member
    edited April 24


    One possible option is to use Weak Springs to stabilize the simulation and limit rigid body motion (keeping the boundary conditions you have currently unchanged). You can go to Analysis Settings --> Solver Controls --> Weak Springs: On. Although you will still get the warning about insufficient constraints while solving the problem, you shouldn't see very large deformations (that occur when rigid body motion occurs). After getting the Solution, you can use Probe --> Force Reaction on the Weak Springs to ensure that the reaction force on the weak springs is negligible (i.e. the weak springs didn't do a lot of work on the system to alter its response).


    Hope this helps,


  • andeca10andeca10 Member
    edited April 30

     Thanks Sai, this is exactly what I was looking for! The resultant force reactions were on the order of 10^-9 - 10^-10 N, which is significantly less than the forces experienced in the device itself.



Sign In or Register to comment.