Form finding on a closed surface

Hi everyone,

I’m currently doing some work on the “form-finding” field. Following the work that I did in this past post (Varying the material Properties (young's modulus)). Now I’m trying to do something similar but with a closed surface, like a cylinder. Basically, I applied a “mesh load const” in order to simulate the pressure of the concrete into the textile formwork and the pinned supports are in both the top and bottom parts of the cylinder (see picture below).


Looking at the results, in some parts of the geometry, it’s possible to see some lines/points getting out abnormally (see picture below). If I increase the mesh refinement those abnormalities increase quite much (see the second picture below).


So, as I said my goal here is to simulate the pressure of the concrete into the textile formwork. Is there a way that I could correct those abnormalities? I’m open to suggestions. Thanks!

Besides that, I have two more questions:

  1. Is there a better way to simulate a pressure load onto a shell than the “mesh load const”?
  2. I notice that when I use the “mesh to shell” element, it seems to divide the mesh into triangles and not squares. Is there a way that a can force the mesh elements to be squared? Usually, they have the advantage of having those singularities (pointy encounters).

Here is the grasshopper file:
Form Finding 2 - (37.1 KB)

Thanks again in advance,
Gustavo Silva

Hi @gustacsilva,
the strange behavior comes from the fact that for your structure the stiffness-matrix is ill-conditioned: The size of the structure is 200m and the shell has a thickness of 0.25cm only. This effects a huge stiffness difference between the in-plane and bending stiffness of the elements.
A solution could be to use membrane elements (which are new in Karamba3d 2.0.0):
Form Finding 2 - (47.2 KB). NII is used here to control the transverse stiffness of the membrane.

“mesh load const” is probably the best option here.

Karamba3D currently does not have quad elements for shell/membrane calculations.

– Clemens

1 Like