This content has been marked as final. Show 3 replies
You are right but slide 25 shows gap should be 1/2 wall thickness. I attempted thickening the wall thickness definition on your shell model, but it still would not join. The best option I've used frequently is to make an assembly and insert an empty part, then (as you mentioned) create midsurfaces manually on the empty part, open it into its own window, and extend the surface to meet the midsurface of the other component, and then create a split line. It is a lot of extra CAD work, but worth it in the end. Good work. Sincerely, Tony.
there's probably something wrong or corrupted with my model.
I've cleaned up my registry and I've built another model with a smaller thickness (3mm) and the suggested procedure worked: I manually extracted mid-surfaces of those plates so that the gap between shell elements was only half thickness.
Contact was defined between those shells and when analysis was run everything worked fine and the model deformed as if shell elements were glued.
I think that the issue with my model was maybe the fact that it was too thick (5 mm) or simply the file was corrupted.
As you said in the previous post, we can therefore state that contact between shells works even if there is gap between them as long as that gap is half thickness.
This does not matter. Shell meshes are an idealization of reality. Whichever side you pick to create the shell should not matter, the results are not affected significantly since the solid is thin enough. This is the basic assumption of shells, if the results between selecting one side of the solid or another are big you shouldn't be using shells in the first place.
The bonded contact is a mathematical bond, not a physical one. Even if the shells are not bonded on the screen, if you defined the bonded contact correctly, those shells are bonded.