On 01/29/2013 06:19 PM, Derek Gaston wrote:
On Tue, Jan 29, 2013 at 2:47 PM, Paul T. Bauman <ptbauman@gmail.com> wrote:
I'm sure it's on everyones wish list, but the only discussion I recall is that mixed-dimension problems are going to break. That is, if you want to couple a shell element (2-D in 3-D space) with a 3-D element, there's a fair number of places that will break. AFAIK, that should be the only limitation, but others will know better than I.

It depends on how you do it... they can't directly share DoFs (nodes)... but you can form constraints between them (like with Lagrange multipliers or other types of "transfers" like contact forces).  This is something we've been doing quite a bit.

OK, that's a good idea. But anyway, first step is a plate-only structure. I need to read up a bit on the literature on plates and shells; there's lots of options (Kirchoff-Love, Reissner-Mindlin, MITC)...