axial force problem on nonlinear beam element

[hibrhmdrn]

Hello everyone,

In a 3D model, I assign rigid diaphragm to the stories using rigidDiaphragm command. When I use elastic beam and column elements and run analysis, I get 0 axial force on beams as expected. But when I use nonlinear beam and column elements using nonlinearBeamColumn command with fiber sections, after analysis, I get huge axial force on beams due to shift of neutral axis.

To get rid of axial force on beams, I assign axial hinge to the nonlinear beams. I mean, I define an elastic uniaxial material with very small axial stiffness (10e-9) and I assign it to the beams using section aggregator. When I do this, I get reasonable results and there is 0 axial force on beams.

Can any expert comment on this problem? If this solution seems wrong, how can I get rid of axial force on beams?

[Allaaldeen]

I am not an expert, but I had the same issue.
The axial force in fiber elements is caused by restraining the elements from expansion. Concrete fiber elements expand when subjected to moment. that is because the axial displacement of the element is calculated by integrating the strains at midsection at each integration point. these strains are often in tension resulting in expansion. If you restrain that expansion, a considerable axial force develops in the beam and, more importantly, the yield moment increases.
The expansion of beams is real [1]. However, I think expansion would happen in bare beams not in beams with slabs. To solve the problem you may model the beam with part of the slab, or use plastic hinge model for the beams.

1- Pantazopoulou, S.J., and X. Qi (1991). “Response of RC Frame under Lateral Loads,” Journal of Structural Engineering, Vol. 117, No. 4, pp. 1167–1188.

[geoStru]

how can I resolve this problem for the beams(3d RC frame)?? any ideas?

[selimgunay]

The solution in the first post looks like a reasonable solution. Or instead of modeling a fiber section, you can compute the moment-curvature relation of the beam offline and assign it to your section.

[geoStru]

The solution in the first post looks like a reasonable solution. Or instead of modeling a fiber section, you can compute the moment-curvature relation of the beam offline and assign it to your section.

for example

uniaxialMaterial Elastic $matTag 10e-9 #material

at the moment I have got this section aggregator for my beams..
section Aggregator $BeamXSecTag $SecTagTorsion T -section $BeamXSecTagFiber

how should I introduce the uniaxial material?
section Aggregator $BeamXSecTag $matTag P $SecTagTorsion T -section $BeamXSecTagFiber
is it right?

[selimgunay]

Yes that looks right. Let's see how your results look like when you include that.

[geoStru]

it seems work
but if I want to compare the results with penalty method that I find in a lot of posts in this site

how could I use the value of alpha in penalty method?