Sheet metal meshing

[fridtjofZM5ME]

What we have to choose from in the mesh department are the tet-elements that you aren't suppose to use for thin sheet metal, and the quads and triangle shell elements. That is all good and well. Until you have contacts on both sides of the sheet metal parts that is. For all the penetration offsets and max activation distances in the world, everything inevetabely tends to go south with this concept in the context of an assembly with almost all sheet metal parts.

So, is there some clever way of getting to use shell elements on sheet metal parts that still allows you to define contacts relating to both sides of the elements? The shell element method creates a 2d-mesh and adds a uniform thickness to it as far as I understand, the only thing missing is getting to use the oposite side of the mesh for anything.

I'm currently attempting a workaround where I select both sides of the sheet metal part (not the edges) and define a shell thickness equal to half the sheet thickness placed with an offset of -0.25*sheet thickness. I generates twice as many elements as would be neccesary if I could have used both sides of the one sided elements. I suppose you could regard it as shell elements behaving like solid elements within the Nastran context in away...

[John_Holtz]

Hi @fridtjofZM5ME 

The answer may depend on the analysis type, the contact type (bonded, offset bonded, etc.), how many layers need to be in the contact, and the gap between them.

Figure 6 in the following article shows one way to define contact when you have 3 shells in a sandwich. See Understanding maximum activation distance and contact type in a Simulation. If you can provide a "typical model" or a sample test model, that would help. There may be some Parameters that can be changed to allow a special case like yours to work.

P.S. From a programming perspective, contact tries to avoid creating contact when parts are penetrating. So the question is how much penetration/interference is real and how much is existing because the mesh approximates the surface. I am hopeful that there are parameters that will tell the solver to allow the penetration AND not try to eliminate the penetration at time 0.

John