Mesh deformation resulting in intersecting boundaries

[skaal]

Hi all,

I have the following rather weird behavior:
I set up a simulation with a rather thin and soft material. When solving for the deformation, I see the upper wall protruding through the lower one, which seems a bit weird, physically :

protruding_walls.png

(132.55 KiB) Not downloaded yet

I cast this into a very simple and reduced test case, that shows exactly this behavior in the second time step. Solver input and mesh file are attached.

Am I am doing something wrong here? Did I find a bug? Is this a genuine feature?

[Rich_B]

Hello,

Attached is a modified sif file, adding a slowly increasing load over time. This improves the behavior somewhat, up until the solver stops early.

Rich.

[kevinarden]

It is possible in the FEA simulation to shove one side of the plate through it's thickness and out the other side. This is not possible in real word physics. However, you only gave the FEA model linear elastic material properties, and therefore it doesn't know that it should be shearing and failing. You basically defined F=Kx, Hooke's law. Force = stiffness times displacement. If you want any thing else to happen during the simulation than you would have to define many more physical terms in the FEA simulation. It is possible to simulate geometric and material non-linear behavior. However the code solved exactly what you modeled.

The result looks exactly like it should given the boundary conditions, load, and solver used.

[skaal]

Thanks!

It seems that this works quite a bit better when running with a nonlinear / plastic material model.
@kevinarden: your UMAT library works pretty well, here

However, also revisiting the time step helped quite a lot -- the case is significantly more stable with a larger time step, even when using the linear material model as in the provided sif file.