Jagged Pushover Curve and Period variation

[aydindaei]

Hi,
I do an adaptive pushover analysis for a RC building and I extract pushover curve.But after a specific displacement the pushover curve gets jagged and the period of the structure changes to either higher value or lower value in subsequent steps.So my question is,are the results of this analysis reliable or not?
It's better to say that I've done lots of other enhanced pushover analysis on the similar building but I didn't encounter this phenomenon.

[fmk]

did you cut the step size down and observe the same behaviour?

[aydindaei]

hi dear fmk,
To solve this problem, I write a script to add load gradully.In this case the load vector of each step is the "summation of previous load vector and a factor of new load vector" and I take this factor=0.1.When I do this, the pushover curve gets normal but again there is period variation in the analysis.I mean in post peak region the period of the structure rises and falls frequently and it reaches a big value then the eigen gets negative.I have this situation in all of pushover methods(e.g. mode-1 pushover).What is the reason of this rising and falling in the Period and the high values of it?Is it rational or not?The answers(disp,drift) are rational.
It's good to say that I used Concrete07 material.

[fmk]

as the initial stiffness is the largest, the fundamental period should be bounded by the initial period and inf. think of a sdof spining T=2*PI*sqrt(m/k), as k goes to 0 T goes to +inf .. the periods of very large period should be equating to instances where structure near inf flexible (at least for one of current modes)

[aydindaei]

Thank you for your prompt reply dear fmk,
I get part of my question but again I don't know what is the reason of rising and falling of period at post peak.For example the period rises to 10s at one step then falls to 5s at subsequent step and this procedure continiues until the eigen gets negative.
I really appriciate your answer.

[selimgunay]

During the analysis, period is computed using tangent stiffness, so the ups and downs in periods are consistent with the jaggedness in the pushover curve. Jagged part of the pushover curve could be due to some numerical issues. Could you post a picture of the pushover curve?

[aydindaei]

As I mentioned earlier the jaggedness of pushover curve solved by incremental update of load but again the period variation exists.what is the problem?

[selimgunay]

This means there is some local unloading in several members due to some sort of redistribution, which increases the tangent stiffness momentarily. If your pushover curve looks good, you should not worry about the periods.

[aydindaei]

Thanks a million Mr.selimgunay.