Quadrilateral Plate Element Questions

[DCBII]

I am relatively new to finite element analysis. I have done a lot of beam element modeling but not much with plate elements. I'd like to use plate elements to model pressure vessels similar to elevated water towers (only shorter and larger in diameter). I have a few questions regarding quadrilateral plate elements.

I understand that singularities can be produced by corners of rectangular openings. Are there any other locations that I need to watch for these, particularly for the types of structures I'm designing? How about where the plate geometry suddenly changes, such as a 90 to 120 degree intersection where the tank bottom meets the sidewall? How about for concentrated forces (in-plane or out-of-plane) delivered by the supporting columns & bracing below?

 

[rb1957]

i don't think you'll have singularity problems with cut-outs ... you'll see stress peaks at a zero radius corner which are not real (for a multitude of reasons).

i think you can model the corners (base with sides) easily enough.

two things i'd mention ...
1) model a simple tube under pressure first, you can get the theoretical results (stresses and deflections) but it can be a bit "fiddly"; but it's well worth the effort. then add the end caps ... this is "much" more fiddly to get right.
2) i'd had real trouble with pressure shells modelled with plates and stiffeners modelled with beams ... the pressure shell "forgets" that it's a tube and reacts the pressure as a set of almost flat plates. modelling the stiffeners as rods returns to the "corrrect" pressure shell results ... got a call into the help desk to try to explain this

 

[rb1957]

meant to add ... mesh refinement around concentrated loads

 

[corus]

In-plane discontinuities will give singularities, such as square holes, but intersections between the shell and tank bottom won't as the shell will give only the direct (membrane) + bending stresses and not the non-linear stress through the thickness that you would normally get with a 3D model.

You have to be careful that the stresses you see at these connections aren't averaged across different thicknesses so plot the stresses for each thickness separately.

Tara