I got my simulation of a mattress and an indentation into it to work (thanks also to JimB and MarkLamping). The model uses a hyperleastic material model (especially the Foam model) and PLSOLIDs. It works well and allows very large deformations (compressions) as long as the mattress is a whole or the cutouts are beneath the surface. The force-displacement-curves are very realistic and practically not sensitive to mesh quality or element size:
But when the cutout is located at the surface, the deformations in particular spot are so high, that they distort some or even just one element too much. It does not help to use a finer mesh as there are then still elements that totally distort. In my experience it is even more likely that the coarser mesh allows larger displacements. There are sometimes also some problems with contact elements but they can be solved. My Simulation stops because of too distorted elements.
What could be done to avoid this problem? Any suggestions?
Things I thought of so far:
I heard of it very often but do not know if NX can do that at all and especially on an deformed model an map all the strains on the new mesh.
I have some experience with Element Death/Birth but I used it only time depended. It would probably help to turn off the element but this would have to be dependend on the distortion and not on time.
PS: As the .GIF format is the only one that NX allows to export (as long as I know) it would be clever if the forum would allow this format!
As I can't see the videos I add some pictures:
My suggestion is to check the stability criterium of the foam material. The foam material tensor can loose its positive definite behavior under loading - often resulting in distored meshes shortly before this happens. Are you sure that the foam material parameters are tested in a strain range that matches with your simualtion?
you are right, the possible deformation depends a lot the material model. When I use some curve fits from the Polymer Foams Handbook I get large deformations. When I use only one pair of ALPHA and MU I get larger deformations, probably because the model valid for a larger range.
Anyhow, I set up a model for the simple compression test and the test in the mattress. Achieving large deformations in the simple test doesn't necessarily mean to reach large deformation in the mattress test. And it can also be the other way around! At the moment I cannot find any logic to that.
Right now I am looking for a to fit the test data, but I will post this in a new thread!