Hi there
Trying to model a metal-coated polymer cylinder in axial compression using axisymmetric modelling in Abaqus, while assuming no adhesion between the metal coating and polymer core.
I understand that by assuming no adhesion, the contact modelling must be used in Abaqus. As a very simple approximation I am just using the "general contact" method with simple contact properties (hard normal contact, no friction tangential contact). I am using Abaqus/Standard (not Explicit)
I am setting up an axisymmetric model and introducing the load via a displacement BC while keeping the other side fixed (no axial or radial movement on fixed side).
As expected, there is buckling occurring on the metal coating (see image), as the contact modelling inherently introduces an imperfection into the system. This is causing the Abaqus loading step (set up as "Static, General") to not converge with the prescribed amount of displacement as specified by the boundary conditions.
I read online that the Riks method is meant to handle situations like this, especially where contact is considered. However, this only gets me a bit further, and still does not completely converge.
I believe the next step would be to try and resume or redo the analysis using Abaqus/Explicit, but I am not sure if this will work with contact modelling in an axisymmetric simulation. I am about to try and refine the mesh to see if it will allow me to converge...
Any thoughts?
So it turns out that the Riks method won't work for my case:
"For postbuckling problems involving loss of contact, the Riks method will usually not work;"
(source: http://abaqusdoc.ucalgary.ca/books/usb/default.htm?startat=pt03ch06s02at03.html)
Anyone have an idea if there is a "standard" of stabilizing the problem using dashpots or springs? Or any literature that exists for a problem like this?
[p.s. guess I'm talking to myself for now :D]
http://abaqusdoc.ucalgary.ca/books/usb/default.htm?startat=pt03ch06s02at03.html
Hello Bharat,
Prior to using Riks Method, did try you using static stabilization with the *static step?
Give that a try and see how it goes. The static stabilization will atleast allow the contact to get separated and prevent the job from diverging.
Resorting to a dynamic approach would be the last resort.
Hello Rajas,
Thanks for your response!
Wanted to say that my intention is to obtain the buckling load from these simulations, so while I want buckling to take place, I also want the simulation to converge so that I can get the peak of the force-displacement plot...
As you suggested, I tried using the static stabilization. Did this by changing "automatic stabilization" to "specify damping factor" and entering a damping factor of 0.002 (10 times the default value provided).
The simulation (just static, not Riks) still did not converge but I am going to try and increase the damping factor gradually to see if that helps.
I'll see if there is a difference in using Riks with a damping factor.
I realize that a last resort will have to be a dynamic approach and I'm trying to stay away from that. At this point I believe that this is do-able in the Abaqus/Standard and as you are probably thinking we should exhaust all the avenues there first.
The newest version of Abaqus I have access to Abaqus 2016 (version 6.16 I believe), whereas everything I have been doing so far has been on Abaqus 6.12. According to the documentation on Dassault's website, contact modelling was improved for that version so I'm going to see if that makes a difference too...
- Badges
- Science topic