Abaqus contact/constraint problems

[Samy76]

Hi,

I have a curved deformable plate that I wish to flatten using 2x flat plates (1 at the top and 1 at the bottom). My question is, do I choose those 2x flat plates as a discrete or analytical rigid model? In terms of the contact property, I imagine it will be a surface-to-surface contact with penalty and hand contact method? Also, I am slightly lost with the constraints - do you tie top plate to the curved plate and again the curved plate to the bottom plate? Of course in both cases, the 2x flats will be masters and curved will be the slave respectively. Please let me know if more info is needed. Cheers

Example model:

 

[FEA way]

Analytical rigid will be faster. You can leave the default contact settings for now. Instead of using tie constraints, it's better to model contact with friction between the deformable and rigid parts.

 

[Samy76]

So I have tried using the general contact and also surface-to-surface contact with friction, but both of them aren't working. I am getting negative few eigenvalues for some reason. Any advice please?

 

[FEA way]

The model is underconstrained, you have to eliminate the initial rigid body motions of the curved plate. E.g. by utilizing symmetry.

 

[Samy76]

I have applied constraint to the curved plate and the simulation works. However, I'm getting multiple warnings: 1) "Displacement increment for contact is too big" - 2) "There is zero FORCE everywhere in the model based on the default criterion". When running animation, the top plate forces the curved plate downwards (which is what I want), however as soon as the curved plate touches the bottom plate, the simulation stops. What am I doing wrong here? :<

 

[FEA way]

Use general contact and check the force-displacement plot to see what happens before the analysis fails. You might be entering an unstable regime difficult for the NR method to converge. In such cases, it's sometimes best to use explicit dynamics but implicit dynamics (quasi-static application type) may also help thanks to some numerical damping.