IDA Parameters - Scale Factors

[ucesmin]

Dear All,
I have modelled a 3 storey RC frame in Seismostruct and I ran a Dynamic Time-History Analysis for specific seismic event (scale=1.0, completed successfully). Next, I ran an Incremental Dynamic Analysis for the same frame using the same seismic input applying the factors 0.1, 1.3, 0.1 for start, end and step scaling factors respectively, and the model completed the first 7 steps of analysis (up to scale 0.7) and then stopped, as it was “unable to apply the next step load”. Without changing the aforementioned model, I just applied the following scaling factors 0.5, 2.5, 0.5 (start, end and step respectively) and my model completed the first 3 step of analysis (reaching 1.5 scale), but was unable to continue.
I would like to ask why this thing happens, and whether I need to change any parameters in order to make my model run all the steps (i.e. convergence criteria, iterative strategy etc.).

Thank you in advance,
Regards

[huffte]

That certainly is a puzzler ucesmin. I'm wondering if you could establish the "failure" loading factor for the structure by examination of the results? Clearly, if everything is defined appropriately, the first IDA should not have stopped at a scale factor of 0.7.

Do results for the second phase of the second IDA, which would correspond to a scale factor of 1.0, match results for the original dynamic time history analysis?

Was the same file used and the analysis type simply changed from Dynamic Time History to Incremental Dynamic Analysis? Probably so, but if not, it could be something very subtle. Both cases have the same permanent loads? The same convergence criteria?

I'm just thinking out loud really, ucesmin. Trying to come up with a reason that the first IDA would fail at 0.7 and starting with the obvious.

I'll be interested in learning of the resolution of this one. Best of luck.

[ucesmin]

Hi huffte.

As you said, the model implemented is exactly the same in all cases, I just simply changed from Dynamic Time History(DTH) to Incremental Dynamic Analysis(IDA) and then played with the scaling factors.
Out of curiosity, I rerun the DTH several times scaling the seismic input myself (using 0.1 step, in total 13 DTH analyses!) and compared with the IDA(0.1 scale, which stopped at 0.8, instead of reaching 1.3 scale). I’ve noticed that the results are identical up to 0.6 scale in all 4 storeys displacements of the building(and also BS). However, in the 0.7 scale, even though displacements at first 3 floors are very similar, the displacement difference at roof level is huge. Subsequently, the analysis is continued in the IDA for the remaining 6 steps, but instead of using scaling of 0.8, 0.9, 1.0 etc the software is using a constant scaling factor of 0.002284(!). Interestingly, the output of the remaining steps is almost perfect for the first 3 floors, having again a great difference in the top floor and BS.

In conclusion, it seems that the software is carrying the whole range of the analysis, but it fails to calculate properly the top displacements and BS after a certain scale.

When comparing the two IDAs that use different scale steps(0.1 and 0.25 respectively), things are even worse. There is no similarity in the results at any common scale (i.e. 0.5 and 1.0), making the 0.25 step IDA model look inaccurate.

I have already sent copies of the above models to support team and I am expecting for their final deduction.

[seismosoft]

Dear ucesmin,

We confirm to have observed similar anomalies, and will now be looking into the possible causes for such (mis)behaviour, and the needed corresponding solutions.

Thank you for bringing this matter to our attention.

Seismosoft Support

[Stelios_Antoniou]

Dear Ucesmin,
We carried the analyses with your model, which helped us identify a bug, regarding the output of IDA results. The bug will be corrected in the upcoming v6.5 of SeismoStruct, which is due to be published during the next days.
Thank you for your contribution in improving our software.
Regards,
Stelios

Seismosoft Developer