I don't have an answer but point out that it is very possible that there are bugs in any software...
However, this thread and your examples do make me remember an old discussion from few years ago. In it, (in no small part due to how PTC handled it), I developed severe mistrust in Creo Simulate FEA capabilities:
Non-symmetric results for symmetric model and load?
Disclaimer: I don't have maintenance, so I can't track if the SPR 2258467 has been closed. I have to say that it seems to me that PTC has gotten better lately, so I'd be curious if there was a resolution to a very basic issue that somehow got past PTC QA and made it to a released product costing $$$