Why does a simulation considering only adhesion converge slower and give weird results, as compared to a simulation considering both adhesion and plasticity?

[sitangshugk95]

I am trying to simulate the evolution in contact area for different displacements for a rough surface for three different scenarios:

• Only elastic deformation

For this, I am using the Hardwall simulation example in Contact Mechanics and getting results that make sense:

• Elastic + adhesion + plastic deformation

For this, I am using the Adhesive plastic cone example in Adhesion and getting results qualitatively that make sense (more contact area than 1):

• Elastic + adhesion only (ignoring plasticity effect)

For this, I use the same code as before, but replace PlasticSmoothContactSystem by SmoothContactSystem. However, the convergence takes significantly longer, and gives results that does not make sense (i.e. contact only along the periphery, very high pressure values as visible in the colorbar autorange):

Questions:

1. Is there a reason why the adhesion+plastic simulation converges faster than the simulation which has ONLY adhesion? It seems non-intuitive since ignoring plasticity might make the system simpler, but I might be wrong.
2. What might be causing this contacts to be formed only on the periphery? The contact regions look very weird.

[sitangshugk95]

@sannant

[sannant]

Difficult to tell.

1. Are you sure the simulation is converged?
2. Are you sure about the paramtrisation ? Try starting with weaker and longer-ranged adhesion. Example of parametrisations that correspond to rather weak adhesion: https://link.springer.com/article/10.1007/s11249-021-01454-6

[sitangshugk95]

Thank you for your inputs @sannant .

1. What is the general procedure for ensuring that the simulation is converged?
   Is it by seeing if the contact areas are remaining same by changing the following:
   (A) increasing the force/penetration tolerance? OR
   (B) increasing the maximum number of iterations? OR
   (C) increasing the number of data points per um of the surface? OR
   (D) any other way I am not aware of?
   (Generally for FEM models I would make the mesh finer but here I am kind of restricted by the AFM scan already taken. Ofcourse I can take a new scan, but I just wanted to get your feedback on the order of preference, based on your experience)

2. When you say parametrization, do you mean the interaction law being used? (Eg; use cubic DMT-like interaction, ias in the Contact Mechanics challenge, instead of the vdW82)?
   (Just wanting to make sure I'm understanding parametrization correctly)

[sitangshugk95]

Thank you for your inputs @sannant .

1. What is the general procedure for ensuring that the simulation is converged?
   Is it by seeing if the contact areas are remaining same by changing the following:
   (A) increasing the force/penetration tolerance? OR
   (B) increasing the maximum number of iterations? OR
   (C) increasing the number of data points per um of the surface? OR
   (D) any other way I am not aware of?
   (Generally for FEM models I would make the mesh finer but here I am kind of restricted by the AFM scan already taken. Ofcourse I can take a new scan, but I just wanted to get your feedback on the order of preference, based on your experience)
2. When you say parametrization, do you mean the interaction law being used? (Eg; use cubic DMT-like interaction, as in the Contact Mechanics challenge, instead of the vdW82)?
   (Just wanting to make sure I'm understanding parametrization correctly)

@sannant do you have any suggestions for these? I am planning to do a benchmark test on the Contact Mechanics challenge - is that potential defined in any of the Interactions in the package?