Convergence issue in simulation with buckling and contact in Abaqus

[otaviobarodrigues]

I am attempting to simulate the buckling problem of two concentric tubes in Abaqus, with the larger diameter tube being rigid.

To simulate contact, I have used the following conditions:

- general contact, all with self, type edge to edge, radial direction

The post-buckling analysis is conducted with automatic stabilization using default values, and friction has been neglected.

I conducted a mesh refinement test on an example considering a mesh of PIPE31 elements with lengths of 2m, 1m, 0.5m, 0.25m, and 0.125m. I am considering the first buckling mode and a 20% imperfection.

My expectation was that the problem would converge with mesh refinement, or that the time values would remain consistently constant; however, this did not occur.

Element length - time
2m - 0.056
1m - 0.139
0.5m - 0.054
0.25m - 0.053
0.125m - 0.048

Is this type of behavior common in contact problems with geometric nonlinearity in Abaqus? If not, what suggestions could you provide to solve the problem?
I am attaching some simulation files.

 

[FEA way]

Did you try with regular beam elements too (as advised in your previous thread) ? They are fully supported and commonly used with general contact in recent versions of Abaqus.