Good morning everyone.
I am struggling with this rather simple model that represent the insertion of a steel washer into an aluminum recess.
The model is axisymmetric and plasticity is considered in the case of washer material.
I tried to apply the displacement to the washer (as can be understood from the attached model) or to the aluminum seat, to see if any improvement in convergence can be achieved.
I tried every combination of master/slave configurations and different meshes densities I could think about: nonetheless, the model does not converge.
I even tried to create small fillets in correspondence of sharp edges to allow a smoother contact, but nothing changed (I followed some tips & tricks from the snap-fit tutorial: PrePoMax (CalculiX FEA) - Tutorial 43 - Snap-fit .
It is possible to see that there’s still little compenetration between master and slave surface:
The image shows also the best result I got so far: the washer (blue) is supposed to clip into the seat (red) as a snap-fit.
The model I attached is the last version of many different tests, can you give a look and give some advice? Thank you very much!
Did you try different surface interaction properties as advised in the snap-fit tutorial you’ve mentioned ? That (along with the mesh setup) was crucial to achieve convergence in the case discussed there.
yes, starting from the standard “hard” configuration I tried with different K values. Being the less rigid material aluminum I considered 5-10-50x Young modulus in defining K parameter, but nothing changed.
Sorry, my bad: in a last attempt I tried to solve it as a simple 2D problem trying to figure out if axisymmetric condition was to blame for the non-convergence. I linked the file with wrong elements type.
I’m not sure I got the geometry right (or the scale) but I took the 2d model meshed it with quads (eventually using C3D8R elements) and rotated into a quarter solid model. I used mecway with a bit of playing around with time steps and 10000 GPa/m contact stiffness
I agree, I think it was in meters when I exported the CAD to do some cleaning/slicing.
I would go the same route as @fatmac , and mesh it with quads. I would start with surface-to-surface contact, and adjust the contact parameters. I try a direct time stepping procedure for this type of problem.
That’s a very good job! The scale is mm, and in fact the final model would be made with solid elements, since the geometry isn’t truly axial symmetric: I simplified things a bit for debugging purposes and moved to 2D axial simmetry to have quick results, since the 3D model I made gave me similar problems.
