Instability with member end release

[kmarsh]

for a typical concentric brace to column/beam intersection, if I release the bending moments in the brace member, the node becomes instable (type 3) in the UX direction (of all directions)

Would anyone care to weigh in on the issue here? If I leave the brace fixed, this is fine. I can even release bending moments in both beams framing into the column and it's fine, it's only if i release the brace bending moments. I expect that, with the top of the column not released coming into the joint, there shouldn't be any problems.
Adding DSC option fixes the issue... but why?
Sparse Solver has this issue
Sparse M does not
Frontal and Skyline fail silently and do not solve.(may be a different problem)
relase instability.png

[pawel.pulak]

Hi Ken,
can you attach the file?
The reason of the instability may be more global - related to parts of the structure outside of the screen capture.
Regards,
Pawel

[kmarsh]

Hi Pawel,
Here is the model. I have removed all releases except the one on this brace and confirmed that removing the release on this brace will allow it to run (without DSC) but adding the release back to this one brace causes type 3 in UX. I have also tried selecting all and intersecting to find short members and fix them but don't have too many other tricks for now to find the issue. Would love to know if you all have any tips and tricks for resolving instabilities (a checklist maybe of troubleshooting procedures?)
Cheers,
-Ken

[pawel.pulak]

It is an unjustified instability reported by the Sparse solver.
Not observed in the Sparse M (multilevel) or in the Multi-threaded solver.
Reported to the DEV team to be analyzed.

[kmarsh]

Hi Pawel,
Thanks for the information, I'm glad it wasn't just somethign I was doing wrong. Also of note, considering that there might be an issue: try running Frontal or Skyline on that model. Robot seems to start running calculations but then just exits suddenly with no error or failure message and results remain out of date. Perhaps this is related.

[kmarsh]

As a side note, when i re-created a similar model from scratch to test this behavior, everything worked fine. Must be something funky with that model. It originally came from Revit and I can send you the Revit file offline if necessary.
Thanks again,
-Ken

[pawel.pulak]

Also of note, considering that there might be an issue: try running Frontal or Skyline on that model. Robot seems to start running calculations but then just exits suddenly with no error or failure message and results remain out of date. Perhaps this is related.

It is not related to the effect discussed before. This time it is a behavior fixed in RSAPro 2014 where the following message is displayed when skyline or frontal sover is used for your model:


There is no need to send Revit file. The unjustified instability message in Sparse is related to numerical properties of the model - it can be caused by for instance the density and shape of FE mesh.

Regards,

[jpo82]

for a typical concentric brace to column/beam intersection, if I release the bending moments in the brace member, the node becomes instable (type 3) in the UX direction (of all directions)

Just to make sure, have you double checked all releases you have made? I have found that instability type 3 usually occurs when model has too many releases. For instance, a beam is released in such way that it's possible to rotate about it's own axle => Robot can't decide which gamma angle should be used for calclulations and gives instability error.

[kmarsh]

I thought similarly so I went back and removed all releases from the model except this one brace. I've run into the member spinning about its own axis before and am usually careful to fix rotation about local x for one end of the release pair. However, in this case, having removed all releases except this one, it's still unstable (Ux) so unstable in translation in the global X direction; very strange. If I remove the release on this one member, (so that there are no relases in the model) then it will calculate. If i put it back in, then it fails again.
I just re-read Pawel's response and it sounds like an issue with the mesh on the roof-deck. I'll go back and see if i can tweak it to run. Also need to figure out why there is a rigid link in there (based on the new error message that should pop up).