There is a sealing with a rubber core and a polymer shell.
The sealing is positioned into a groove and then is subjected to pressure from a one side.
I've built a 2D model with plain strain elements and set two steps:
Step 1 - Pushing into the groove (everything is okay, no problems).
Step 2 - Apply the pressure and got convergence problems.
I tried
1) Various mesh size, and ALE. Conclusion: Finer the mesh --> earlier getting no convergence
2) NLGEOM is off. Get convergence, but I have doubts regarding results and accuracy.
3) To play with Poisson's ratio. Initially for the rubber core I input nu=0.45 and get no convergence. With nu=0.49 no problems.
4) To use Implicit and even Explicit solvers. Same thing.
5) Also tried full and reduced integration, linear and quadratic plane strain elements.
6) I have kinda ideal plasticity for the polymer shell and remain only elastic properties, but results are the same.
7) Of course I tried to reduce the time step up to 1e-12 and use more iterations for the increments and use solver stabilization.
Do you have any ideas to get convergence with NLGEOM to achieve accurate results? I would be really really appreciated.
Inp. file is attached.
Really, Poisson's ratio for rubber is near to 0.49 (practically uncompressible material). In this case, according to your conclusion 3, convergence is good?
Serge V. Shil’ko, I know about it, but let's it is a some kind of a polymer with such properties.
Yes, if Poisson's ratio for rubber is near to 0.49 there is no problems with convergence, but I guess it is because elements have "no big deformations".
Aleksandr, how much strain is realized under pressure and what model of material do you apply?
Serge
Serge V. Shil’ko I applied a simple elastic model for "rubber" with E=7-9MPa, and linear plastic for shell (plasticity starts at 20 MPa and going up to 20.1 at 300% of plastic deformations, almost pure plasticity).
I got strains in the FEM about 50% (compressive strains).
There are large strains which requires non-linear model (for example, Mooney-Rivlin hyperelastic model) application.
Serge V. Shil’ko , I have tried non-linear models Mooney-Rivlin and Ogden based on my data. It works only if assumed that the material is almost incompressible.
Serge V. Shil’ko I don't think another hyperelastic model could help if input similar material data.
I was able to calculate using explicit analysis with second-order precision, but results are strange. It seems that combination of low elastic stiffness and Poison's ratio