FEM Non-Jacobian Nodes

[Vision78]

I'll re-run the sim and check it, maybe i had a typo somewhere or smth because i rewrote the constraints for the model

[Vision78]

I checked it, although I havent run a mesh convergence study on it I would assume finer mesh would result in a more accurate depiction of the results.

EDIT: Ran it again with the top part having maximum mesh size of 20mm and the deforming side at 6mm, seems the deformation is mostly not happening on the compression but warp, also have to keep in mind that it is at a scale of 0,1% and the load is lopsided, not sure if that was inteded?

There is some rise in stress as well when you refine the mesh but its concentrated at the top part on the edge of the constraint so it would depend on the use case if I would get worried.

Either way, take these results with a handful of salt, I'm not that familiar with projects of this size and just been dabbling in FEM past few months.

Also note, the deformation factor is 50 to see where the material is going because at the scale of 1 you wouldnt be able to even notice it visually.

[ishnik]

Thanks Vision78, I really appreciate your work cross-checking this analysis.

With a finer Netgen 1st order mesh, I was roughly able to duplicate this - I need to read up the documentation and do an example or two in order to apply refinements to the mesh at specific areas though.

Some thoughts on this as related to the real life example are:

• * Buckling of the wall is the failure mode I'm concerned about - a more rigorous analysis would be to find the safety factor for buckling as the next step
  * The real life application will have a bolted connection (the holes are obviously for this) and I will design the connection to be slip-critical (in American) or Friction type (Aus / EU). With a bolted faceplate transferring the load across the whole faying surface, the hole area becomes less of a concern. AISC requires to design the connection to be safe in bearing anyway, but that's not a big deal and the detail will be nicer in real life.
  * Also in real life the whole 50t stack-up sits nicely ontop of the loadring so the asymmetric loading in the model is definitely conservative - the thing is going offshore so design is for worst case of "what if load shifts and only a single connection is supporting it at a time?"

Long story short, pretty sure "she'll be right mate", but I've still got a bit of work

[Vision78]

I had some concerns over the holes for the bolts as if I have understood correctly, using the inner surface does not load it realistically as it pulls the opposite side of the hole which a bolt would not do, however it might just be nitpicking.

I dont think there would be buckling but the simulation with fused parts did show the upper edge moving with the shell in the Z axis, which also explains the stress concentration on the top plate as it is basically working as a lever arm, parts in real world being separate however would behave differently which is a thing to consider.

I'd always trust real life testing over the simulations of course, the simulations of even the best and most expensive programs cant fail at some tasks and people running them do make mistakes and assumptions that do not apply in real life.

Edit: Also to add, region meshes only seem to be available for Gmsh, (selecting a face or faces for a finer mesh)

https://wiki.freecad.org/FEM_MeshRegion