Applying interference fit contact

[AminKB]

Hi,
I'm trying to model a cuboid sandwiched between two less stiff blocks, as shown in the cross-sectional view below.

This model is intended for some iterative analysis where the parts are expected to have an arbitrary level of initial overclosure. I applied a surface-to-surface, hard contact with a friction coefficient of 0.5. To resolve the overclosure automatically in a robust way, I applied an interference fit condition as well.
The model has two steps. In both steps the bottom cuboid is under encastre. a zero displacement boundary condition is applied to the top block in the initial step, which is propagated to the second step (the first analysis step). Also, in the second step, a uniformly distributed pressure is applied on top of the top block to simulate pressure on the cuboid.
The analysis would take a long time without converging and would abort in the second step (the first analysis step).
The results show stresses and strains, as expected from an interference fit, but the overclosure is not actually resolved. This is also true for all the other trials I did, with multiple meshes and boundary and loading conditions.
Any advice on how to get this analysis to converge and successfully resolve overclosure using interference fit?

Thanks

 

[FEA way]

Try without friction and with different overclosure levels as well as with other settings of contact and interference fit options.

Can you share the file ?

 

[NRP99]

Check whether node to surface formulation helps in solving the interference fit. Alternatively you can use general contact algorithm which is more robust in terms of convergence. Also use frictionless contact in step 1 (interference fit) and then introduce actual friction in step-2 (Dummy step with no load applied) and then now actual assembly is done, use the loads to either load or relax the interference.

 

[AminKB]

Thank you for the responses.
Replacing the BC on the upper block with a 1N uniformly distributed pressure solved this issue and the model would converge quite quickly.
I started having issues again when I tried tilting the cuboid in the middle a little, so the overclosure is not uniform. It seemed then that the friction had a huge effect. The model was failing to converge. I then reduced the friction coefficient from 0.5 to 0.1. The model is now working, but it's extremely slow, it takes about 24 hours to resolve the overclosure. So, my next step is to try Step-1 without any friction, as suggested.
I'll let you know how it goes!

 

[NRP99]

Instead of pressure loading, you can try displacement loading for initial contact step-1 and dummy step-2 and then apply your load in next step-3. Displacement loading helps in contact convergence/initial contact setup efficiently than force loading. Also, to reduce the time you can add as many dummy steps as you need. This ensures the analysis is converging with stable equilibrium in each step before moving actual loading step.