IDA analyses on steel MRF

[Ana Santos]

Hi everyone,

I am doing IDA (incrementa dynamic analyses) on the same MRF frame but with different beam-to-column connections typologies... and for each of the connections typologies the MRF presents a different fudamental period... my problem is, when I check the maximum drift ratio (0.04 rad in my case) versus the spectral acceleration the MRf with the higher fundamental period, achieved the limit drift ratio for a higher spectral acceleration and I think that should be the opposite because a higher fundamental period means a lower stiffness of the structure..

I am taking the drifts from the relatives displacements in the IDA envelope...

Anyone knows what could be the problem here?

Thank you,
Ana Santos

[huffte]

Did you overlay the fundamental periods of the various frames on the mean acceleration spectrum of the ground motion set (or the spectrum of the single accelerogram if you are doing a single record analysis)? Just wondering where the frequency content of the ground motion is compared to the various structures. Are the analyses all linear or are you getting nonlinear behavior in the connection elements? In the beams and columns? Which drift is controlling in each case - top story, bottom story, different for each frame? Maybe some more details on your interesting results will elicit some ideas, Ana. Best of luck as always.

[Ana Santos]

Hey huffle,

Thank you a lot for your answer... I am scaling the spectrums of the different ground motion in order to have a similar behaviour around the fundamental period of the strucuture... and I have introduce scale factor until at least one of the connection os the structure achieved the ultimate rotation (0.04rad).
And in the damping properties in pre-processor, I have introduced the 1st and 2nd period of the structure...

All the connections have been modelling by link element with a smooth curve in the M2 dregree of freedoom and the beams /columns with displacement-based inelastic elements?

REgarding the analyses I am doing nonlinear dynamic analyses with the nonlinear propoerties of the structure...

I think that my problem might be due to the fact that you a MRF with a lower period (higher stiffness) the spectral acceleration that it will be faced are much higher than the one with a higher period (less stiffness) , so the stiffnest one will be acheved the ratio drift first than the other... makes sense?

Thank you,
Ana SAntos

[Ana Santos]

I have another question also... to see the beam rotation on SeismoStrucut...is in the IDA envelope in the dl parameter of elements deformation ? or in the R3(A) ? I do not know what those parameters mean and I can not find the definition anywhere ...

thank you, Ana Santos

[huffte]

Ok, Ana. Your reasoning seems sound. It is certainly possible that a longer period structure would require a higher acceleration level in order to reach a specified chord rotation value. Just a couple more clarifications. You are apparently using Rayleigh damping. Are you employing initial-stiffness based or tangent-stiffness based damping? I would at least try tangent-stiffness based damping. There is no consensus, but after having studied the issue a bit, I believe it is more accurate. Clearly, it can be difficult to converge with a tangent-stiffness based solution where large inelastic behavior is occurring. But it is at least worth a try to see how the results are impacted. And even tangent-stiffness based Rayleigh damping can result in essentially mass-proportional damping depending on the ratio of the two specified modes. So I am a little concerned your damping could vary significantly among the various frames. A good reference for these issues is:

'Viscous Damping in Seismic Analysis and Design', M.J.N. Priestley and Damian N. Grant, Journal of Earthquake Engineering, Imperial College Press, 2005.

Similar discussion on damping can be found in the fantastic, must-have book 'Displacement Based Seismic Design of Structures', Priestley, Calvi, and Kowalsky, IUSS Press, 2007.

Regarding the R3(a) parameter. It is my understanding that these are chord rotations, not rotations at the end of the beam, which I am not sure are available. Note that for infrmFB elements which are not subdivided, the code-based-criteria chord rotation is the direct output chord rotation from SeismoStruct. For elements which are subdivided, the user must post-process the resulting displacements to get a chord rotation to be checked against code criteria.

For a deeper discussion of element action parameters, I 'll recommend you look over the excellent SeismoStruct Help in the section:
Contents -> Running an Analysis and Processing the Results -> Post-Processor -> Element Action Effects

Best of luck, Ana.

[ruipinho]

Ana, hi.

For what it is worth, just a quick, and somewhat philosophical, note to say that:

- your attempt to try to understand the reasons why certain results are the way they are is praiseworthy and is something that analysts should always do (rather than simply accepting results at face value)

- interpreting nonlinear dynamic analysis of MDOF systems is, however, not always straightforward, given the intrinsicacies involved in such type of analyses (some of which have been touched already by you and huffte), hence at some stage one just has to accept the results as they are

Best,

Rui