TIE CONNECTIONS AND SHELL SURFACES

FatmacFatmac
For some time now I have avoid using shell surfaces in the majority of my work as I have had trouble making them work in non-linear analysis for various reasons. Just recently I have had the need to use shell surfaces again and have come across a few issues -many I have had before. Perhaps some one may be able to help me with these?. First I have had success using *Tie connections between shells/shells and shell/solids but I get the red waring in MW26 suggesting that only solids can be used? (see attached pic).
Is there any need to be concerned about that waring? Secondly I frequently can not run non- linear analysis (even if I relax the convergence tolerances). Is there a work around or is that some thing I'm doing wrong

Comments

  • VictorVictor
    edited September 25
    You can usually use the elastic option instead of *TIE. It works on shells.

    *TIE kind of works sometimes but it's unreliable and I'm not sure what all the failure modes are.

    Nonlinear analysis not converging is usually because of unconstrained rigid body but there are many reasons, buckling can also be one, or very large deformation, or contact too hard. You can run it as linear to make sure it's properly constrained, and if that doesn't help, delete big parts of the model, especially contacts, to isolate what's causing the problem. You can also add fixed supports to many parts to ensure no rigid body motion, then remove them until it fails.
  • FatmacFatmac
    Thank you for ur input. I not had too may issues with *TIE but was put off by the red warning -I'll continue ignoring it. My battle with NL and shell elements has be long and I generally avoid them where possible! Where I have had problems I have not had any issues with solving with the linear solver. Some times the NL ones work with very low loads then don't converge with slightly higher loads so perhaps there is some thing going on with early buckling instability with shells??. Most of the time I'm confident its not an unconstrained body as I frequently have a solid set up in the same analysis running in parallel with the same constraints and normal it runs with the linear solver. I wasted lots of time not realising that RNE(2) connections don't always work with shells (even though I had noted the warning MW gave me!) - they normally work in the static linear but not the NL. I have found some work arounds for the node (nonshell!) surface connections (some with your kind help!) but still frustrated I can't always get the NL models working with shells -particularly when i want to do a no line post bucking analysis. I'll keep battling with the shells when i have to and otherwise avoid them like the plague!
  • disladisla
    Would you share one of those models that are giving you issues in NLG?. Light version if possible.
  • VictorVictor
    I see these error messages start to look useless when they say something's not allowed but it still works. Maybe it should explain the reasons too. Nonetheless, I still wouldn't use *TIE on shells because you never know if it's quietly over-stiffening them or something. Elastic contact is more robust on shells because it doesn't use MPCs like *TIE which can interfere with the MPCs of the shells' knots.

    Is it practical to extrude the shells to solids? The value of shells is that they're easier to mesh, not more accurate or faster to solve. CCX does the extruding internally but with a lot of ways to fails as you're seeing.
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In Register