Learning Forum

[Simone Del Nero]

Hi everyone,

I'm performing topology optimization on a vessel subjected to 3D loads, so it is not possible to simplify the analysis. In particular I have the cylindrical shape (with a predefined initial thickness) and I'm interested in defining the best material distribution around it with the less material as possible, maximizing the stiffness and keeping the stress below the yield stress. 

I run different topology simulation with different mass constraints and constant stress constraint; I noticed that some combinations of mass and stress (stress is constant) converge, while others have a very slow convergence history and the solutions do not change appreciably between iterations. 

Can anyone explain why or suggest some references?

Thanks a lot for helping me!

[John Doyle]

Failure to converge, or slow convergence, can be caused by many factors. 

Certain combinations of objective and response constraints might be creating a mathematical over constraint (i.e. not all the criteria can be satisfied simultaneously).

Or perhaps it can be solved, but you need more unconstrained DOFs (finer mesh) in the design domain to capture the optimal solution.

I would recommend when your model fails to converge, start by reviewing the solver output for any relevant feedback as to what the code is struggling with.

 

Since you mention that this is a vessel, I assume that one of the criteria is that it needs to envelop a "watertight cavity."

Ansys supports this feature as a manufacturing constraint.  Are you including that?  If not, this might help with convergence and with quality of result.

For more details, please refer to Mechanical User's Guide =>Analysis Type=>Structural Optimization=>Opt Workflow=>Obj & Constraints=>Manufacturing Constraints…

[Simone Del Nero]

Hi John,

thank you for your reply, it was very helpful!