Mesh Problem during Analysis


I am trying to use Karamba3D for my research, but I want to validate the results, so I constructed a simply supported beam (analyzed as shell for now) and got its deflection.

The theoretical solution is 170.97593 mm (using beam theory), and the following are the results from Karamba3D:

  • x axis: mesh size in m
  • y axis: deflection

Why am I obtaining weird wiggly shape in the convergence study and how can I solve that?

Simple shell (33.6 KB)

depending on the mesh resolution the error between the maximum displacement in the model and the calculation ranges between 0.75% (size of mesh faces ~1m) and 0.08% (size of mesh faces ~0.06m). The convergence is not uniform over the mesh-face sizes since the shell elements have non-conforming shape functions at their boundaries (see e.g. here for details).
The displacements across the middle section of the plate strip are not constant. When taking the average Z-displacement, the error at mesh-size 0.06m is 0.02% (see this definition: Simple shell (48.3 KB)

here the ShellSection-component of Karamba3D 2.2.0 WIP is used to get the displacements in the middle: Simple shell (48.1 KB)
– Clemens