Page 2 of 2
Re: moment in the end of brace pinned with link
Posted: 13 Nov 2013, 21:20
by jackrussel
Hi huffte,
1) The first model was 3D and I changed it to 2D(fema365). So I consider out of plane buckling.
2) I used effective duration of arias intensity from seismosignal.
3) As I said the model was 3D with rigid diaphragm so I used constraint to approximate the effect of rigid diaphragm.
you are right. the constraints at ground was a mistake that I don't have in other model.
The problem of braces remains.
thanks a lot
Re: moment in the end of brace pinned with link
Posted: 18 Nov 2013, 00:24
by Stelios_Antoniou
Can it be that you are not using a force-based convergence criterion? If you use it and it leads to divergence, it is most likely that you are employing a non-linear curve in the links. In such cases make sure to use a significant (5-10%) value of strain hardening.
Re: moment in the end of brace pinned with link
Posted: 18 Nov 2013, 16:35
by jackrussel
Dear Stelios_Antoniou,
Convergence criterion is Displacement/Rotation. Strain hardening is 0.004458
Re: moment in the end of brace pinned with link
Posted: 20 Nov 2013, 18:15
by seismosoft
Yes, but, as Stelios Antoniou suggests, have you tried using a force-based convergence criterion?
Also, is there distributed load on the frame elements? If yes, have you activated the stress-recovery option?
Seismosoft Support
Re: moment in the end of brace pinned with link
Posted: 21 Nov 2013, 20:13
by jackrussel
seismosoft,
1)There's no distributed load on elements.
2)When I use force-based convergence criterion the analysis diverges. because I have rigid elements in my model.
Re: moment in the end of brace pinned with link
Posted: 22 Nov 2013, 13:24
by Stelios_Antoniou
Convergence is generally difficult in such cases of frame structured with steel braces pinned at the beam-column joints, and a compromise has to be sought for. Either you use a displacement-based criterion and accept that you will have unequilibrated forces at the joint, or you impose a force-based criterion too, but then you need to change the constraints type to Lagrance Multipliers and increase the strain-hardening parameter to values up to 5-10% in order to allow the model to redistribute the forces after yielding.
Note that such convergence problems are not related to SeismoStruct and its inability to correctly redistribute the unbalanced forces (to our knowledge SeismoStruct is one of the most stable inelastic analysis packages worldwide), but rather to the type of modelling
Stelios Antoniou
SeismoSoft Developer