Ansys shell element bug
We were analyzing an excavator part for strength. We wanted to set up a global model to get all effects from the forces transferring through the hydraulic cylinders. The model was a combination of shell elements and solid bodies.
Contacts and joints were set up, and a mesh for checking the connectivity was created.
We added gravity, and everything seemed to be Ok.
But when we started to add movement to the hydraulic cylinders, things started to be strange. We started to have solution convergence problems and very high stresses. Stresses exceeded 20000MPa, and gravity was still the only load.
As any FEA users would do, we started troubleshooting. We checked everything, but it was set up correctly. We started to suppress parts to see where the problem started.
After just having the first component left, and still having this problem, we started to be suspect that we were not in control of the problem.
A small test analysis was created, and we finally have some idea of what was happening.
Gravity was added, and we got some logical results.
But as soon as we added a remote displacement. First a linear moment, then a rotation.
Stress was OK after 50mm travel:
.
But after 30degree rotation, thing started to happen:
After some more investigation, it turns out that ANSYS does not handle ordinary connections and rotations any good. Her we have an analysis with just two shell elements and an edge-edge connection. We are only displacing it in 3D space, and gravity is the only load:
We will probably use mesh connections to replace the contacts.
But remember! Be careful when analyzing in FEA. This bug is confirmed in ANSYS 19.2 and 2020R1
There is now information about this on the web, and we assume Stressman is the first to catch this bug.