idaholab / moose

Multiphysics Object Oriented Simulation Environment
https://www.mooseframework.org
GNU Lesser General Public License v2.1
1.77k stars 1.05k forks source link

Weird contact behavior in Tensor Mechanics (RZ coordinates) #11968

Open GregVernon opened 6 years ago

GregVernon commented 6 years ago

Rationale

Am having weird contact behaviors with a seemingly simple problem in Tensor Mechanics (RZ coordinates). Several of my collaborators at INL have been assisting with this issue and have requested this be posted here.

Description

Primarily, the issues seem to be in computing initial penetration/gap/contact which seems to be resolved by flipping master/slave assignments (which by doing (possibly) adversely affects friction calculations due to where the contact pressure is being computed). This issue manifests itself by contact surfaces "sucking together" over large distances, or not even enforcing/recognizing contact at all. The cases that correctly initiate contact seem to struggle with large sliding displacements, eventually diverging in their solutions and failing.

I've got several cases (several simplification steps) that should help describe / troubleshoot the issues. I will post each in forthcoming comments to this issue. Each will have a Cubit journal file, geometry, and input file, description, pictures, etc.

Impact

In general, robust contact will be necessary for success of Tensor Mechanics in general engineering analysis applications and resolution of this issue (whether user-error or something in the code) should improve external capabilities.

GregVernon commented 6 years ago

Example 1: Flat zig-zag with additional 1/2 symmetry

There are two contact specifications, top_surf and top_surf_flip below images show the behavior with top_surf activated. top_surf_flip has the master/slave assignment flipped. initial_time_markup

This is the eventual state of the simulation, using the same top_surf contact as the above image.
image

In the below image, I've hidden the upper platen to reveal the computed penetration value on the deformable part. Note that the penetration value is negative which suggests that a gap exists between the two contact surfaces, i.e. that there is no contact. image

Interesting behaviors can be seen by moving the upper platen to have a gap or initial overlap prior to the simulation. I recommend testing the following configurations

makeZigZag_flat_half-CubitJour.txt zigzag_flat_half.i.txt