Hi,
I recently started to use Abaqus in order to simulate the spherical indentation of a porous layer.
Here is a quick description of my model (see the link below to download related files).
2D axisymmetric model with 2 parts:
- the indenter: discrete rigid body, real shape of the indenter defined by a spline curve (so it's not exactly an arc of circle)
- the sample with 2 sections:
=> the porous coating: elastic material with Drucker Prager Plasticity
=> the bulk: elastic material
Elements: CAX8 and CAX8-R.
Interaction: hard contact, no friction
Step 1: displacement control (loading and unloading)
I have the following convergence issues:
- If the coating is considered as elastic (no plasticity), I manage to get convergence. However I noticed 2 strange things that I don't understand:
=> If I reduce the increment size from 0.1 to 0.01 I lose the convergence beyond a certain displacement
=> If I use a finer mesh under the contact, I lose the convergence beyond a certain displacement
- When I try to use the Drucker Prager plasticity, I cannot get convergence beyond a certain displacement
I’m a bit lost… I don’t know where the problem(s) come from. Is it contact related ? mesh related ? geometry related ?
I would be very grateful if you could suggest a methodology to solve this.
To finish one basic general question: how to know what caused the divergence in a simulation ? Sometimes I don’t have any warning before it stops and it is not clear either in the Message file.
Thank you very much for your help
https://www.dropbox.com/s/496bdfb3472b5lu/Spherical%20indentation.zip?dl=0
I downloaded your inp-file. In Abaqus 6.13, it runs to the end (simulation time 2), loading and unloading your part.
In general, finding where a problem comes from is not easy. Look at the job diagnostics/msg-file to see what warnings are produced in the final solution increments.
In your example, trouble may be due to
- using 2nd-order elements in a contact problem (may be fine because your indenter is very smooth)
- too large compressive stresses under the indenter (use hybrid elements perhaps because large plastic deformation is incompressible)
- mesh distortions becoming too large
and probably a bunch of others...
BTW, your mesh probably could be simplifed a lot if you use a mapped mesh only in the indented part and coarsen the mesh in the elastic lower part.
Thank you very much for your answer Martin. That indeed helped a lot to know that the simulation was working for you. I found that the convergence problem was in fact related to memory allocation (I'm running Abaqus 6.11 on a remote server). That explains why decreasing the increment size and refining the mesh was causing divergence and why I was not finding anything in the jobs diagnostics/message file. Regarding the mesh thanks for your advice !
ABAQUS mailing list
https://groups.yahoo.com/group/ABAQUS
http://imechanica.org/
- Badges
- Science topic