Contact between a sphere (TET) and a plate (WEDGE)

[RGX124]

Hi,

I am trying to simulate a sphere that is shot on a plate.
The sphere is meshed with TET elements and the plate with WEDGES due to symetry BC.
The contact between the sphere and the plate is a normal hard contact with Default constraint enforcement method and a tangential friction coefficient of 0.3.
When the mesh of the sphere does not correspond to the mesh of the plate (i.e. the first node in contact from the sphere does not enter exactly in contact with a node on the plate), the simulation runs fine but when the first node in contact from the sphere is directly above a node from the plate, the simulation diverges.

Can I solve this numerical problem without changing the position of the sphere or the mesh of the plate?

Any suggestions?

Thanks

Rgx

 

[RGX124]

I found the way to avoid it (at least for the few examples I run).
The solution is to put a surface to surface discretisation instead of a node to surface one.
Surface to surface is also more accurate.

Rgx

 

[gwolf2]

>I am trying to simulate a sphere that is shot on a plate.
The sphere is meshed with TET elements and the plate with WEDGES due to symetry BC.

Your mesh is the worst kind of wrong for this type of analysis. The element types will cause all sorts of solution problems and if you do get it to work the results will be garbage.

Re mesh both components with well formed c3d8r elements and use c3d8i elements on the contact faces. Use contact surfaces again not node contact as you have already discovered. All then will be well. Expect to have to re-mesh very finely around the perimiter of the contact.

If symmetry is present (i.e. a normal impact) then drop down to axisymmetric elements for quicker, more accurate results with a super fine mesh.

Good luck.

Gwolf.

 

[RGX124]

Thanks Gwolf2 for your advice.
I ve noticed that the wedge elements are not good at all for contact problems.
However, the elementary plate that i'd like to mesh is an hexagone (one half or one sixth of it). It is not natural to have C3D8 meshing this geometry. It is like meshing a triangle with C3D8 elements.
Same remark with a sphere.
Can C3D4 elements be used for that purpose?

Thx

Rgx

 

[gwolf2]

>It is not natural to have C3D8 meshing this geometry. It is like meshing a triangle with C3D8 elements.
Same remark with a sphere.

It's perfectly possible to create a nice c3d8 mesh in these volumes using mapped meshing - which is rapidly becoming a lost craft skill. A lot of pre-processors can't really do it though - TET's or nothing...........

It all rather depends on what you are using to make your mesh - if it is ABAQUS CAE then I can't advise.

Don't use TET4 elements for anything - they should be banned by international treaty for crimes against FE. They often appear to work but sometimes give garbage results with no warning.

gwolf