quad elements posing as triangles

[corus]

It's well known that 2D linear triangular elements are useless for stress analysis. If a model is meshed with triangular shaped elements but these elements are in fact quadrilateral elements with 2 nodes the same, ie. node numbers 1-2-3-3, will the reults obtained be just as useless?

corus

 

[johnhors]

Corus

The results will be even more useless (if that's possible), triangles and quadrilaterals use different shape functions, thus a quad with a degenerate side is not the same as a triangle element, at least the linear triangle will generate some form of results, whilst the quad (if the solver doesn't kick it out) will produce rubbish.

 

[myCAE]

"It's well known that 2D linear triangular elements are useless for stress analysis."

Do you talk about the fact that there is no continuity of the second derivatives , i.e., the stress field for linear triangular element, plus some stresses that cannot been accessed, despite mesh refinement ?

 

[jxc]

Corus,

I am sorry for this late reply & I hope your question is answered, but when this has occured in any of my analysis an element check will locate these elements &/or the analysis will fail with a message hinting (solver messages are never direct) the problem. I use FEMAP & CAEFEM currently, but NASTRAN & Patran have also notified me of these conditions.

JXC

 

[EVdave13]

Like jxc, I have encountered fatal errors where a degenerate element was detected by the solver. You can't get any results that way (normally, anyway. You can force Nastran to solve). The hard part is finding and deleting these bad elements with some pre/post processors! Sometimes you can't pick them. Degenerate triangles posing as line elements are also a problem. I've run into this with both Patran and Femap, usually with imported models.