node rotation, eigen

pbeng · Post by **pbeng** » Sun Oct 04, 2009 4:17 am

Hi,

On a few projects now I've been getting unexpected modal analysis behavior at rigid type connections (rigidLink and equalDof). For example, I've two perpindicular lines of nodes joined as a T. On the vertical line are elastic beam elements, the horizontal nodes are slaved to the intersection node with rigidLinks. I've fixed the joining node against rotation (fix), but modal results have the horizontal line clearly rotating about the intersection node. This is happening at a pretty high frequency (52 Hz) relative to the lower modes (1 Hz). Is there something in the eigen solver process that would free up rotational constraints and perhaps explain some of the rigid connnection anomlies I've been seeing?

Thanks,

Paul Bradford

fran · Post by **fran** » Sun Oct 04, 2009 8:16 am

The joints are moving without rotating and this changes the orientation of the element. May it be because of the definition of the rigid link which may not be sufficiently rigid?

silvia · Post by **silvia** » Mon Oct 05, 2009 7:46 am

i do think it may be a model issue...

pbeng · Post by **pbeng** » Wed Oct 07, 2009 4:21 am

Thanks for the feedback Frank/Silvia. I agree Silvia, this sort of thing almost always turns out to be model error, I just haven't been able to find it (yet). And given I've noted what seem to me rigid element anomolies in the past, and that I've had mixed success matching OpenSees modes with those from another program, I'm a little wary.

The eigen output rotation is 0 for the retained node (as specified), output FEEx_c = 0.0339 for all the constrained nodes (discrepancy). Looking at the source code for the RigidBeam with y as the vertical coordinate, Uy_c = Uy_r - FEEx_r * dz + FEEz_r * dx. Eigen output verfies the fixed constraints at the retained node, FEEx_r = FEEz_r = 0, leaving the RigidBeam code with Uy_c = Uy_r as it should. So the rigidbeam "element" code checks out ok. But modal results show that FEEx_c <> FEEx_r and Uy_c <> Uy_r, the only things that come to mind are that something in my model is screwing things up or there is an error in the eigen solver process.

I'll continue exploring, thanks again...

Post by **fmk** » Wed Oct 07, 2009 10:56 am

can you post a small model with this happening so i can look into the matter in the source code.