Hi all, I would like to simulate a shell (in the shell model) using Fenics_Shells. But after reading about the library, I discovered that the shell should be defined with an analytic expression, can I simulate the shell that I have in STL file without any analytical expression? Is it possible or not?
I am also interested in knowing the answer.
When you say “simulate a shell”, what do you actually mean. It is important when working with finite element analysis to rigorously write down the problem you would like to solve. What sort of structure is this? What sort of boundary conditions? What sorts of forces are applied to the structure? What material model will you utilize? etc… If you provide more detail or an initial attempt at a problem, it might be possible to provide you with specific help, not just vague generalities.
After you have thought a bit further about the above question, you can narrow down what sort of shell model you will need. For example, Reisner-Mindlin (RM)and Kirchoff-Love (KL) shells have very different kinematic assumptions. Generally for thin shells one would use KL, but in a classical finite element context KL shell formulations are difficult to utilize as they require C-1 continuity. RM shells are often used for “thick shells”, but can recover the KL shell solutions at the thin shell limit. I don’t know much about the fenics_shells package, unfortunately. Howver, for RM shells could take a look at the shell_analysis_fenicsx repository written/managed by my colleage @Ru_Xiang here: GitHub - RuruX/shell_analysis_fenicsx
Additionally, .stl files are often a terrible format to represent your original geometry. If possible, use a different file format based on high quality CAD model (something like STEP/IGES) and then use a mesh generator like gmsh to generate a mesh that is of sufficient quality for your usage. If you absolutely only have an .stl, you can utilize meshio to read an stl and convert it to an .xdmf