Acceleration time history

[anbazhagan]

I have the acceleration time history record with almost 35410 data points and I want to get the response spectrum but in seismosignal its showing out of memory while I tried to find spectral parameters. If I consider acceleration time history leaving initial or last portion of acceleration time history , ground motion parameters that I am getting is also getting affected. So what points I should consider to reduce the acceleration time history or which portion I can omit from the data record so that I get the spectral response for different time periods without variation in the ground motion parameters?

[huffte]

One option is to use a text processor or spreadsheet software to eliminate every alternate point and then double the specified time step. Another option is to load the original file and observe leading and trailing regions which are negligible and to specify beginning and ending line numbers when you load the file into Seismosignal corresponding to those negligible regions.

Best of luck anbazhagan.

[ruipinho]

huffte,

Just a quick note on the fact that it should be possible to carry out your "Option 1" from within SeismoSignal itself, by choosing to read-in both 'Time' and 'Acceleration' values, whilst setting a 'Time-step' that is double that of the original record.

Best,

Rui

[huffte]

I am not certain that you could instruct SeismoSignal to skip every other line though? Or perhaps I misunderstood. Thanks very much, Rui. All the best. - Tim

[seismosoft]

Tim and Rui hello,

Indeed, it is possible to skip every other line with SeismoSignal. This can be done by employing option 3 in the opening SeismoSignal dialog (i.e. Multiple Acceleration values per line) with an Initial values skipped=1 and a Frequency=4.

Regards,

[seismosoft]

Note also that 35410 data points should not give an out-of-memory message. I note that 35410 real numbers are (4 bytes per real): 35410*4/1024=138kB. I would assume that 10-20 series of data points are stored per accelerogram (acceleration, velocity, displacement, corrected uncorrected etc), which means that the 35410 data points are stored in 2-3MB. This could not give an out-of-memory error. Something other is happening. Some tips: (1) the text file has 35410 lines but the time-step specified is extremely small. (2) The period step is extremely small when calculating the spectra.

If nothing works, anbazhagan please send us the input file with the input settings at support@seismosoft.com

Seismosoft Support

[ruipinho]

I still believe that what I proposed works (but perhaps I am not understanding what the others are saying here?):

- let us say that one has an accelerogram featuring 10.000 pairs of time-acceleration points, defined at every 0.01 seconds

- from within SeismoSignal, one can read-in such accelerogram (using option 2, which reads both time and acceleration), specifying a 'Time-step' of 0.02 seconds

- SeismoSignal will thus create an accelerogram that features 5.000 points, which is equivalent to having skipped every other line, not?

Cheers,

Rui

[huffte]

I stand corrected, Rui. You are correct. I misunderstood how this feature was working, but tried your suggestion and it appears to work perfectly. This is a very nice feature and I am sure I will now find it useful. It is also possible to skip 2 lines or three or any number of lines by setting the appropriate time step. All the best - Tim.

[seismosoft]

You are right Rui, this can also be done with option No.2. In fact it is a much more straingtforward approach, with respect to what we suggested (employing option 3).

Regards,