Fiber hinges - Pushover Analysis 1

[StereoCivil]

Dear All,

I am running a Pushover analysis for some 3D RC frame structures with OpenSees and trying to compare the results with SAP2000.
I am using nonlinear fiber sections (distributed plasticity) in OpenSees, so, to achieve a similar modelling in SAP2000, I used Section Designer to create the sections, set reinforcing bars and define fiber layout, and then assigned fiber hinges P-M-M ( with "Default from section" fiber definition option selected) to column & beam elements. This way, I think that definition of nonlinear properties of the structure approximates OpenSees fiber sections.

After some pushover analyses, for instance say a regular 1-floor - 1 bay building, I get quite similar pushover curves from OpenSees & SAP2000 ( an ascending bilinear elastic branch until ultimate displacement and then a descending one) , but thing is that these displacements are relatively large for this building (>25.3 in).

When I try a pushover with user-defined hinges properties (not fiber ones) in SAP, what I get is a much different curve with sudden strenth drops and much smaller ultimate displacements.

What I need to know is, whether the results I get from fiber hinges-sections are considered normal or not, because my experience in nonlinear analysis is limited.

Thank you for your time

 

[stressed]

Did you actually input all the P-M2-M3 data for each curve of each hinge for each member in order to try to match the fiber hinge properties? Because that would require a lot of input if done correctly. See attached

 

[StereoCivil]

Thanks a lot for your reply.

No, actually, I didn't. So I suppose it's all about the different hinge properties...
I also tried with FEMA auto hinge properties and I got a curve with sudden drops and values of the same order as with Fiber hinges.
Below are shown the images of the pushover curves. First one is from SAP with FEMA hinges, second is from SAP with Fiber hinges and last one form OpenSees.

Column sections: 50/40 cm w/ 4 rebars D20cm
Beam sections: 25/50 cm w/ 4.00 cm2 rebars amount
Gravity loads on beams: 50 kN/m

I just wanted to ask two things.
1) Is maximum floor displacement for this particular considered normal? It seems to me too large
2) Is Fiber hinges modelling for a Pushover analysis in SAP2000 considered reliable?

Any other tip or recommentation would also be appreciated

 

[stressed]

1) For pushover analysis, it's common to use nonlinear "Load application" to be "displacement control" instead of force

2) Fiber hinge modeling is generally considered more accurate than FEMA hinges because of the P-M-M interaction based on material stress-strain

You didn't mention the lengths of your columns and beams. My only other tip would be to change "Results saved" in the NL load case dialogue from default 'Final state' to 'Multiple states' using program defaults for Multiple states.

 

[StereoCivil]

Thanks again for your prompt reply!

1) I used the displacement-control option with a target displacement of 10% of total building elevation. I am sorry I neglected to mention the lengths; columns length= 4.73 m , beams length= 5.00 ~ 6.00 m along the two bays.
Now do you think the curve I get is 'normal'?

I also used 'Multiple states' with program defaults.
However, concerning the Warning messages I get ( maximum total & null steps reached), it seems that the analysis steps on the degrading branch of the Pushover curve fail to converge, so the procedure interrupts.

I tried to make a finer mesh of the Fiber Layout of the sections using 12x12 or more fibers & increase Maximum total & null steps of nonlinear parameters, but rarely with better results.

Is there anything else I can do to overcome the convergence problems and get the full Pushover curve until the target displacement I set?

 

[stressed]

Review of individual hinge results (moment vs rotation) would be helpful. Non-convergence = questionable results, even though SAP will report results from the last step. One of your screen shots showed a lot of intermediate hinges assigned along the length of the columns and beams. Try assigning only at the near-end points (.05 and .95 relative distance)and see if that solves your convergence problem. So many hinges along the length may cause convergence issues. If you don't have any luck with that idea, send to CSI's tech support and let us know what they say. Without seeing the loading details and output, it's difficult to give more advice.