Seismosoft | Earthquake Engineering Solutions

Hi,

I'm experimenting something strange. I have a accelerogram on which I applied a baseline correction and a filter via SeismoSignal. After saving the Acc/Vel/Displ, I can see (via excel) that the displacement saved corresponds to the one seen in SeismoSignal however, if I reopen the saved file with SeismoSignal, the displacement obtained is decaling down. Is it normal to experience that?

thanks

Dominic Michaud

Dear Dominic Michaud,

Can you please send us the files in question, so that we look into this matter?

Many thanks.

Seismosoft Support

I think you would understand better my problem with a video rather than the files. Here is what I just recorded. If you still need the files, I could send them but I think the video is enough.

http://www.youtube.com/watch?v=9Q-d_O7M0DE

Dear Dominic,

Thanks for the excellent video, very well done.

Unfortunately we could not reproduce the behaviour you describe, when using accelerograms different from yours, hence, if you send us your seemingly problematic time-history, it would be very useful.

Seismosoft Support

Dear Dominic,

As far as I understood, the acceleration time-history that you are opening the first time is not the same as the one you are opening the second time, the former being the uncorrected and the latter being the corrected one.

In other words, in the first case you applied one time baseline correction+filtering, whereas in the second time you applied it twice.

Dear Santon,

I made sure I did not apply the corrections twice, I don't think that is the problem.

The file I use is the following:

//////////////////////////////////////////

Temps Accélération
(sec) (g)
0.000000 -0.018168
0.005000 -0.002439

[data points removed - better to use support@seismosoft.com to submit files]

Hi Dominic, you are right. There is something not quite right there. I will investigate the issue a bit further and let you know. Probably, there is a bug somewhere there. Thanks for the detailed description.

Seismosoft Developer

I think I found it. The differences seem to be due to the round-off error for the calculations. In brief, the initial accelerogram is opened with 6 decimal points, whereas the second one (the one saved by us) features only 3 decimal points (default program setting).

If you change that settings to say 7, the results will be very close, although not exactly identical - integration intime can be a very sneaky thing

Seismosoft Developer

Thank you very much santon, you found it.

It would appear that this will often be the case. Using the default number of decimal places of 3, I corrected and filtered several motions and saved them. After opening the corrected data, it does not look the same as it did upon initial correction. This sems to be happening a lot, not just every now and then.