Nonlinear shell models?

Hi all, is there any reason from a software standpoint/bug that a nonlinear model (plastic properties ) shouldn’t run with shells using tied connections (no rigid body, no contact etc). I solves linearly without issues (400 seconds run time)

thanks

Luke

Are there any errors or does it just abort with no message ? It should work but there are some bugs related to shell elements in CalculiX. Can you share the file (it might be just a minimum model exhibiting this issue) or at least some screenshots ?

It just isn’t converging but that could be how I have defined the plastic properties but I wanted to check there wasn’t an underlying problem before digging into that. The residual force keeps going up with each cutback which isn’t promising.

It is usually user error :slight_smile:

Luke

Yes, it can be a problem with plasticity. But also make sure that parts are properly connected (tie constraints are working as expected) by running a frequency analysis.

thanks for the help, no issues with running the frequency analysis at least.

The old school workstation is keeping my house warm, hopefully it will converge at some point. (it is cold and rainy in southern France at the moment)

The current version of PrePoMax lacks the ability to search for nodes by number right? 1.4 shows the option but it doesn’t appear to do anything.

Try the updated v1.4.0, and it should work.

Hi Matej
I think this was the updated version (unless it is new in the last few days).

I am getting some partial results and it appears that some of the tied connections that work in static and the frequency check are not connected when I run non-linear. Any suggestions of how to proceed with this. I have 180 tied connections but I can’t see any misalignments etc at the problematic spots.

thanks

Luke

Yeah, it was updated a few days ago. Download it again.

Why do you think so ? Check the warning messages generated by the solver and corresponding node sets (missed nodes should be included there). Also, make sure that adjustment is enabled and increase the tolerance.

I will try it again and see if it provides any interesting error messages. I have been running without adjustment but will try it with. I did do one test with adjustment but I think I changed the solver at the same time so I don’t know what caused it to fail straight off

The reason I think there are issues with the ties is that in the partial results I am seeing large displacements in random locations which given the tiny load applied make no sense physically. The confusing thing is that when when I run static or frequency steps these parts do not go flying off into space. In past experience (abaqus) this would mean that if it worked in static aside from contact errors it should run in non-linear as well.

Trying to do real analysis with a less familiar program is instructive if rather aggravating at times :slight_smile:

thanks

Luke

Maybe try a Spooles or Pardiso solver. Sometimes the default Pastix solver gives random resut.

Are those locations anywhere near tie constraints ? Maybe try running the analysis with merged parts if possible in your case. Also, try using tied contact instead of tie constraints.

CalculiX doesn’t offer as many diagnostics related to tie constraints as Abaqus but at least there are the .nam files showing missed nodes.