Model fails to converge past certain drift.

[Joesh]

Does anyone have any ideas as to why a model would stop converging around a certain drift limit? I am modelling a 3 story SCBF using the 2D technique, and in two different failure modes (weld fracture, brace buckling), the model stops converging around a first story drift of 2.5%. In first case, base shear is almost 0 at the end but in the second case, base shear is still 50 kips. I've tried a few different things, ranging from:

-Different numbers of fibers
-Different numbers of integration points
-Pumping up -iter on my fBCs

I'm running my analysis in a loop, so I reduce step size if it doesn't converge, in additon to relaxing tolerances some and using KrylovNewton and Newton -initial and NewtonLineSearch.

Main error I am getting is failing to get compatible element forces for many elements at that particular point.

Same thing happens in my 3D model of the same problem.

[Joesh]

Actually I may have fixed the problem by reducing how discretized my columns and beams were from 2 elements to 1 element, and reducing brace elements from 12 to 8.

Vesna/Silva/anyone, can you comment on the SCBF webinar given the other day. A slide mentions that 20 elements may be needed to correctly capture brace fracture. However, I've read that 8-12 elements per brace is typically the recommended amount for force beam columns. Should we switch to dispBC if we want to model fracture of braces?

[vesna]

If you use fatigue material to capture the failure of the brace Uriz (PEER 2008/08) suggests 10-20 FORCE-based BC elements per brace. Displacement-based elements are not recommended for this purpose.