Hysteresis graphs

[Ralazem]

Hello,

I am doing some hysteresis verification of my own, and I notice that a constant problem is the peak forces don't match up with experimental results. The software always produces higher peaks. What could be causing that?

[seismosoft]

There may be numerous modelling reasons why the numerical and the experiemental results do not match, e.g. material strengths, modelling of confinement, modelling of shear failures, modelling of the second order effects. If you are using the infrmDB frame element type anf you have not subdivided the members to an adequate number of elements, this can lead to increased numberical strength.
Seismosoft Support

[Ralazem]

I've ruled out these factors " material strengths, modelling of confinement, modelling of shear failures,"
Modelling of second order effects is done automatically if I've ticked that box, isn't it? I am also using FBPH, so that should not be causing it.

[seismosoft]

Not sure what the reason could be. You can send the model to support@seismosoft.com so that we have a check, however without the full details of the experimental test, we might not be able to spot the problem.
Seismosoft Support

[seismosoft]

We ran your model, it seems that the response is dominated by the rotational response of the link elements, which also exhibit very large inelasticity. It is likely this is what it causes the discrepancies.
Seismosoft Support

[seismosoft]

Also the plastic hinge lengths (50% of the member length) seem to be very large (though we checked and this does not affect the max. capacity).
Seismosoft Support

[Ralazem]

So you suggest if I change the rotational response properties, I may get a lower peak?
Also I have set the plastic hinge at 50%, because the calculated PH ( originally 15%, and 10%)was causing higher peaks.

[seismosoft]

Not sure. We are not aware of the details of the experiment, and there are dozens of reasons for the differences.
We only highlighted the factor that dominates the response.
Secondly, note that employing such an extreme value for the plastic hinge length, because it provides a better fit to the test results is not necessarily the best approach, since it somehow hides the reason for the differences that you get.
Seismosoft Support

[rbsce.ph]

Hi,

I tried modelling a squat wall with slip shear and compared it to the results of actual testing. However, upon analysis we were able to obtain the same Forces while the Displacements differ in results.
Additionally, I modeled the slip shear (link) using both Bi-linear (MIMK_bilin) and Quadrilinear asymmetric curve (quad_asm) and compared their results, however, nothing changed. Do you have any recommendation in modelling slip shear for walls?

Your response will be highly appreciated. Thank you.

[seismosoft]

Once a horizontal crack develops at the base of the wall throughout its entire width, which I assume is what happened in the experimental test that are trying to reproduce, the level of slip displacements should depend mainly on concrete-to-concrete friction and dowel action. Hence, if you wish to adopt a somewhat analytical approach to calibrate the hysteretic response curve of the link element that you are employing to capture the slip at the base , then you should search the literature for data/models on such phenomena. Alternatively, a trial&error empirical calibration should also do the trick.

For what concerns the issue of the same response being obtained with the bilinear and quadrilinear response curves, I guess it is a result of the fact that the level of displacement being developed in the link is not sufficient to go past the bilinear stage.
Seismosoft Support