Ansys divergence

Hello All,

I have a question regarding Ansys solution divergence in nonlinear static analysis, with geometry and material nonlinearities. In this case the model is shell but that should ba valid for bricks as well.

Usually, a divergence occus when the element distortion is too large, thickness of the shell is reduced or simply equilibrium (default F and M) cannot be satisfied. Stress-strain curve is multilinear with cut out on ultimate, behind which is a perfectly plastic behavior. The element distortion and thinning is well captued as even the software gives the element number to verify that. But in some circumstances the solution diverges without error message. In such case, there usually is a node or series of them with stress and strain value at ultimate, or in some cases the strain gets even farther beyond the ultimate (which is normal as the material is perfectly plastic beyond). But what is the background behind such divergence? And why sometime the divergence occurs right on the ultimate strain while other time the strain is allowed to follow the perfectly plastic cuve. I would expect a plastic hinge to be formed and move along the part while strains would grow indefinitely.Could someone please shed some light on this divergence situation please?

My second question is connected to the previous one. What values are used for the convergence criteria and how is it exactly calculated? Is it a nodal value (average and interpolated) or integration point for each element?

Thank you all.

George