Issues with contact in ABAQUS

[D_Bioeng]

Hi!

I am trying to model a simple cylinder between two NON-parallel plates. The cylinder is subjected to both compression and lateral bending (or other pure moments). The surfaces of the cylinder and the plates are already in contact (the cylinder does not present parallel end-surfaces) and I impose a simple surface-to-surface "Hard" contact with friction. The problem is really simple, but it failed immediately. It is not even able to perform a single increment. The geometry should not present any initial compenetration between the surfaces in contact. I tried to modify some parameters of the contact formulation (passing to the node-to-surface or General contact) but nothing change. I don't know what to do.

Can someone help me?

Thank you very much,
Davide

 

[FEA way]

Apparently, there are initial rigid body motions. What are the boundary conditions ? Can you share a screenshot showing them ?

 

[D_Bioeng]

I applied an encastre at the inferior plate and a moment/force at the superior one. Based on the fact that perfect match is present between the surfaces of the plate and the cylinder, I did not apply any constraint on the cylinder itself. I attach the picture of the simple model with its boundary conditions.

 

[FEA way]

Then there are initial rigid body motions, at least before contact is established. It would be best to add some BCs holding the remaining parts to allow contact to close. You could use contact stabilization too but it may not be sufficient here. Also, displacement control (prescribed rotation in this case) is always better in terms of convergence than force control (torque here).

 

[D_Bioeng]

What type of boundary conditions would you apply to the remaining parts of the model? And what part are you meaning? the cylinder?

 

[FEA way]

You have to eliminate all rigid body motions until contact is established so all parts should be fixed completely except for DOFs used to move them towards each other. That would be the first step and then there would be a second step with actual load and selected DOFs released (only those that have to be released to allow for the desired operation of the mechanism). You could also try changing the assembly so that there are small initial interferences between the parts (to be resolved by strain-free adjustment at the beginning of the analysis) but it may not be enough. Then there are other, more complex tricks but try those (and stabilization) first.

 

[D_Bioeng]

But they are already in contact and it is allowed separation. Is your suggestion still applied?

 

[FEA way]

Even if the geometry is touching, it doesn't mean that contact is established in a simulation (especially when the geometries are curved). You may have to help it. Contact stabilization often solves the issue but sometimes more tedious workarounds are necessary.