Linear solution keeps failing to converge

[flyforever85]

I have to parallel plates with a simple honeycomb in between (basically a sandwich structure). The bottom plat is fixed, the top one moves of a small amount in one in-plane direction. I need to see what happens to the honeycomb in shear buckling. The solution fails to converge. I reduced the time increment to 1e-07 and still I received a message it's not small enough. Shall I just keep decreasing it or is there another strategy I'm not aware of?

 

[FEA way]

Further decreasing initial time increment won’t help in this case. The error that you get is just a generoc message indicating nonconvergence. Here’s what you can try:
- check all warning messages
- make sure that all interactions (contact, tie constraints and so on) work properly and there are no rigid body motions/unconnected regions
- turn NLGEOM on if it’s off now
- use automatic stabilization in step settings
- refine the mesh and make sure that its quality is good
- consider using explicit solver

To give more detailed advice I would have to see the model. So, if you can, please attach a picture of it or, even better, share its files.

 

[flyforever85]

Here the input file. when the increment is 1e-05, it aborts after 5 increment. The lower I go with the increment the more iterations I get but still not enough for convergence. I thought about moving to explicit and I think I'll go there now. Thanks!

https://files.engineering.com/getfi...8f8a-18a36ae53600&file=0-Honeycomb-stress.inp

 

[FEA way]

To solve this in Abaqus/Standard you should include imperfections (imported from eigenvalue buckling analysis). And it might be easier to run the postbuckling study using Riks method.

 

[flyforever85]

Thank you for suggesting the Riks method, I was not aware of it and I will definitely look into it.

Regarding the *imperfection command: as far as I understood, I impose imperfections to my model that follow the path of buckling analysis. Is there a rule for the magnitude of the imperfections?

THank you

 

[FEA way]

When defining impoerfections you have to specify scale factor for each mode. In case of shells it is advised to use a value order of magnitude lower than the shell thickness for the first mode and decrease the factor by half for each of the subsequent modes.