Large Deformations/Nonlinear Analysis Possible in RISA-3D? 1

[b-clark]

I know RISA-3D can take P-Delta into account, but it doesn't seem to take other large deformation effects into account. Or maybe I just don't know how.

For instance, if you fix a cantilever beam to one node and load the opposite such that there is a lot of deflection, the second node has no lateral displacement. This is obviously not correct. Is there a way to get RISA to account for this?

I know that these scenarios are rare in structural engineering practice, but I want to have a good handle on everything it does and doesn't take into account.

 

[JoshPlumSE]

No, the RISA is based on small deflection theory. The only attempt to move beyond this is with the P-Delta analysis. But, that's meant to be a correction for normal moment frame or braced frame buildings.... not a true large deflection theory.

Even many programs geared towards a more advanced analysis will have difficulty with what you're doing. You might test it out with SAP2K or GTStrudl. But, even in those programs you might have to use a cable element in order to get the type of behavior you describe (where a transverse load results in axial tension and displacement).

 

[b-clark]

Josh, thanks for the reply. I thought this might be the case.

I guess I don't fully understand why this would be hard. The elements used by RISA are capable of t/c and bending simultaneously, so it seems that if they just implemented a nonlinear iterative fashion of solving then this behavior could be accounted for.

Am I missing something? is it just that the demand for this type of analysis with line elements isn't there so they haven't included it?

 

[JoshPlumSE]

There are multiple aspects of the issue.
1) The type of solution: Linear vs Non-Linear.
An iterative non-linear solver isn't as simple as it sounds. Theoretically it isn't all that difficult, just apply your load incrementally and modify the stiffness of the structure as it deflects. But, there are so many convergence / divergence issues when you start doing this that you have to add in all kinds of controls and checks on the solution and such.

2) The elements themselves. This is why I mentioned cable elements. Most non-linear solutions still don't account for the case of transverse loading causing axial strain. I believe this is called "Green's Strain" though I cannot say that I'm an expert in the subject matter. Depending on the way a program derives their Cable elements, they may be able to handle this. However, it is element dependent.

All that being said, RISA/Nemetschek still has at least one person (Christine, a programmer left over from the original RISA days when I was there) that could totally handle this sort of thing. If you're really interested in RISA/ Nemetschek doing this sort of thing, you should contact them directly. If they get enough user demand for it, then they may decide it's worth the considerable development effort it would take. But, like most businesses, if they don't see a profit / revenue argument for it, it will not get done.

 

[b-clark]

Thanks again for this info. This sheds a lot of light on the issue for me. I'll look into your suggestions.

 

[IDS]

I don't use SAP2000 or GTStrudl, but my guess is that they probably would handle this, certainly Strand7 does (which I do use).

I don't see how a cable element would work in this context, since they can't handle bending, and I don't think that Green's strain would have a significant effect.

The important thing is that the geometry should be reset at each iteration, so the reduction in the lever arm and the vertical component of the reaction force is included in the analysis. Given that at least some of the competition does this, I should think that it would be well worthwhile to suggest to the RISA developers that they might want to have a look at it.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[WARose]

I've never used it but STAAD has a "LARGEDELTA" command that takes into account forces and stresses for geometric stiffness terms from large P-Delta effects. You can use it for frame members and plate elements only.

 

[JoshPlumSE]

I've never used it but STAAD has a "LARGEDELTA" command that takes into account forces and stresses for geometric stiffness terms from large P-Delta effects. You can use it for frame members and plate elements only.

Sometimes STAAD's features can be a little misleading. I have little doubt that this feature would do geometric stiffness adjustments and give you good P-Delta behavior. However, my guess (and it's a total guess) is that it doesn't do true large deflection theory. And, that you wouldn't see axial forces develop in the beam for this case the OP gave.

I could certainly be wrong because what little STAAD experience I have is from the 1990's! However, I do keep an eye out for STAAD's advertising and such. And, I don't recall ever seeing STAAD advertise an iterative non-linear solution with user defined convergence controls.


To IDS: I'm sure you're correct. In retrospect, my mentioning Green's strain seems downright silly.... I probably confused a conversation I had about cable elongation effects on non-linear cable analysis (which I didn't truly understand) with the much more mundane concept of axial force development in this guy's cantilever beam. I'll blame the RISA deflection diagram because it makes it look like the beam could have elongated significantly. LOL.

 

[WARose]

Sometimes STAAD's features can be a little misleading. I have little doubt that this feature would do geometric stiffness adjustments and give you good P-Delta behavior. However, my guess (and it's a total guess) is that it doesn't do true large deflection theory. And, that you would see axial forces develop in the beam for this case the OP gave.

I'm almost positive STAAD does small deflection theory (only) so you may be right (not sure if the command I mentioned above somehow compensates for that).

I might be able to try a test case tomorrow (if I can think of one; the OP doesn't have enough detail).