Convergence Problems with Non-Linear Static

[equitus.engineering]

I am running a Non-linear analysis on the attached assembly. The analysis converges when I run it as a linear static analysis. But when I change to non-linear, the solution starts to diverge. It has a steel frame and a plastic assembly with separation contact between them. The steel is fully constrained in space and a 50N force is applied to the plastic on the edge as seen in the model.

Given the dimensions of the components, I've midsurfaced them.

The model is attached.

Any assistance anyone can provide in helping this model solve will be highly appreciated.

Regards,

Raam

[sohan.kumar.thiruvakkadu · Accepted solution]

Hi @equitus.engineering 

I did do a quite few changes to the model. Taken a quarter symmetry to make it more stable and changes in the plastic part to have Offset Idealization just to simplify the model to have a node to node connection. I had made few splits on the surfaces just to differentiate the node assignments w.r.t fixed/symm constraints/Contacts/Load, It may be also solving without any of these splits. Have attached the model. The model converged with the contact as "Sliding/no separation"

But still could not get a converged result with the contact set as "Separation". Suspecting that is the loading realistic? It seems the load may be too high causing a 'pull thru' event or catastrophic failure.

[sohan.kumar.thiruvakkadu]

Hi @equitus.engineering 

I wanted to follow-up with the latest update using contact as separation. I did attach the model, please go through the surface contact, Nonlinear setup and its advance settings. Looks like the load is causing a 'pull thru' event and could not converge after 50% of the load being applied. If the displacement is know, It is better to apply the load as an enforce motion.

-------------------------------------------------------------------------------------------------

If my post answers your question, please click the "Accept as Solution" button. This helps everyone find answers more quickly!

[equitus.engineering]

Hi Sohan,

I eventually got it to work. Thanks though.

Regards,

Raam