Best way to support a shell

I am trying to simulate a 4 point shell. Each support is rounded off in order to distribute the forces and avoid stress concentrations.

The first approach was to support the shell support ends with hinged supports for each vertex at z=0. However there are still stress concentrations and due to the missing co-linearity of the points, a kind of fixed support is created. This (probably) results in tension unnecessary forces in the upper shell.

To avoid tension forces, I also tried to enlarge the support, however this now created an even more fixed support, which is not what I want to achieve.

The goal is to introduce no bending moments at the supports, howver I havent figured out a way to achieve this yet.
An idea would be to use a unidirectional support, which only support the shell in the z+ direction, but not in z-.

Any ideas how this can be solved?

Thanks a lot,

Hello @Rudi,
try to place very stiff but weightless dummy beams at mesh edges of the shell boundary. Check different stiffness values. If chosen too large they may lead to a badly conditioned stiffness matrix and inaccurate results.
Then add a few beam elements to tie these dummy boundary beams to one point per edge and support this point only.
Rotate the supports in such a way that the supported directions are exactly in-plane.
– Clemens

Hey Clemens,

thank you for your answer! I did not quite understand your suggestion.

You mean at the shell boundary - of the support edges right? (skteched in red)

So the shell would effectively be held by 4 points in total, right?

in which plane? Could you maybe draw a quick sketch of what you mean?

This is what I could understand so far from your description:

Also I dont qute understand how this approach could support the shell in one direction, but not in the other.

Thanks a lot!


Maybe a workaround using a bunch of beams with a spring cross-section applied to them as vertical supports could work?

Hello @Rudi,

yes, I meant the shell edges at the supports.

yes, each corner should be supported in one point only.

In order to avoid bending moments in the shell, the supports must not introduce shear forces. Therefore support reactions need to be in the plane of the shell only, to avoid transverse forces.

In Karamba3D 2.2.0 springs tie together the DOFs of two nodes without kinematic relations. This means that if one node rotates, the other does so as well but there is no corresponding translation imposed. So in your situation it is better to use dummy beams.

– Clemens