problem in equivalent nodal forces and moments

wunico · Post by **wunico** » Sat Mar 26, 2005 2:18 am

when I use nonlinear element or beamwithhinges element to define a portal frame, and add nodal forces and moments to the end points of beam for representing the vertical distributed load along the beam. The beam's end section moment-curvature time–history curve is very second too.I think this curve is not right. Normally,the section moment-curvature time–history curve surrounds the origin .what cause this problem? And how can I resolve it?

springtea · Post by **springtea** » Tue Apr 05, 2005 9:36 pm

I meet the same problem too! I think it is because of the nodal moment . anyone know it?

silvia · Post by **silvia** » Wed Apr 06, 2005 8:44 am

i don't quite understand the problem, but here is my 2-cents:
are you first imposing gravity and then cyclic loads? isn't gravity imposing an initial curvature and then the cyclic revolves around that curvature, not the zero curvature?

Araj · Post by **Araj** » Wed Apr 06, 2005 9:49 am

Sorry to jump out in middle. But I am also facing the same problem.

I had also thought that way initially as Silvia has put on. But if we put the command "wipeAnalysis" before performing the Transient analysis shouldn't that wipe out any initial curvature from the Gravity Load Analysis ?

So ultimately when we start the Transient Analysis, both Gravity Load and Acceleration history will start from Time = 0 which means (if I understand correctly) at Time = 0 (start of Time History Analysis) there should not be any initial curvature. Please correct me if I am wrong in this.

Thank you.

With regards,
Aloke

silvia · Post by **silvia** » Wed Apr 06, 2005 9:54 am

Frank will have to help us out on this one.
However, I think the wipeanalysis object, according to MY understanding of the manual eliminates all Analysis objects, not the domain objects. And the current state is in the domain, not in the analysis objects.

here is what I got from the manual:

wipeAnalysis Command

This command is used to destroy all objects constructed for the analysis.

wipeAnalysis

This command is used to start a new type of analysis. This command does not destroy the elements, nodes, materials (nDMaterial Command, uniaxialMaterial Command), etc. It does destroy the solution strategies: the algorithm, analysis, equation solver, constraint handler, etc.

Araj · Post by **Araj** » Wed Apr 06, 2005 9:57 am

Thank you for your prompt reply Silvia.

I shall try to get the solution too. Please post it if any new developments.

With regards,

Aloke

Post by **fmk** » Wed Apr 06, 2005 10:43 am

there is a problem with element loads on the nonlinear beam column .. there is a bug report on it .. i am waiting to see if the person who wrote the code can fix it.

as for what happens after a wipeAnalysis .. nothing happens to the elements/sections/materials/nodes/loads/time .. they remain in the state they were at the end of the last analysis for the start of the next analysis.

the things to watch out for is what you do with the loads and the time in the domain for the start of the next analysis .. they also stay in the state they were at the end of the last analysis .. if you start a transient analysis, it starts at the time in the domain plus the dT (if a uniform excitation, it looks for the accel at this time and not dT, which is why in Example3.3 we use loadConst -time 0.0 .. the loadConst because we don't want the gravity loads to keep increasing) if you want to change these then you need to use the loadConst <-time $newTime> and remove (if you are starting another transient analysis and want to remove the last support motion) commands.

again NOTHING in the model changes after a wipeAnalysis command is issued .

wunico · Post by **wunico** » Wed Apr 06, 2005 9:06 pm

are you first imposing gravity and then cyclic loads? isn't gravity imposing an initial curvature and then the cyclic revolves around that curvature, not the zero curvature?

yes,silvia,you are right.there is initial curvature.but the initial curvature is very small.(in my problem,it's about 0.00012) .the frame is almost in elastic state.when the frame is under high seismicity,it will be in deep nonlinear state.and the curvature is much big(about 0.03).so the initial curvature could not cause the momoent-curvature time–history curve asymmetric.
my problem is :when i add the nodal forces(-99KN) and moments (-99KN*M on left node and 99KN*M on right node)representing the vertical distributed load along the beam(33KN/m).and do Transient analysis .the positive curvature of beam's left end section reach about 0.05(positive moment is about 200KN*M),but the section's negative curvature reach only -0.005(negative moment is also about -200KN*M).i think the two direction's curvature should be same roughly .when i only add the nodal forces(not add nodal moment) ,do the same Transient analysis.the two direction's curvature is almost same,and the momoent-curvature time–history curve is right!

I meet the same problem too! I think it is because of the nodal moment .

sprintea ,i think so! when i use very small nodal moment or not add moment,the beam section's curvature is no problem .

silvia · Post by **silvia** » Thu Apr 07, 2005 11:05 am

let's draw a bending moment diagram for the gravity load and then for the lateral load. one is symmetric, one is antisymmetric! so you get large curvature at one end, where they add up, and small at the other, where they subtract....
so the curvatures at the two ends of the beam will be different, right????