(To my followers on linkedin, this is a repeat of an article I post there on Oct 28, 2019. I’m re-posting it on my blog so it will appear in the table of contents)
I’ve written an article on calculating accurate stresses from FEA models that is available here. It describes manual mesh refinement, adaptive mesh refinement, and p-adaptivity and breakout modeling using stressRefine.
An exciting new alternative is Altair’s SimSolid, a completely meshless adaptive analysis tool. There is skepticism about SimSolid because it seems too good to be true, but it is for real. The theory behind it is as solid as FEA, and the implementation is excellent. Since it is a general-purpose tool that does various types of simulation, I was curious as types as to how well it did specifically on the challenging problem of accurate stresses. I was in a good position to investigate this because I had an extensive suite of models that I’d used to test stressRefine. I tried them all out in SimSolid and the results were amazing. It got extremely accurate results on all of them, with no user intervention. The validation manual I wrote is available here. I also submitted these to Ken Welch from Altair Simsolid and he describes them in a blog post here.