wrong harmonic motion calculation

[kvolkan]

I generate simple harmonic motion and when I give the analytically obtained acceleration record (although the software reads the acceleration record correctly) I can not get the expected displacement and velocity time histories for the given data.

I use SeismoSignal Version 4.1.2 on Windows XP.

Similar problem occurs when acceleration time history of earthquake experiments is given to the software.

[kvolkan]

Additionally, the same problem occurs for SeismoSignal version 4.3.0 (again on Windows XP)

[huffte]

Hi kvolkan.

I tried this same exercise. I created an acceleration time history of amplitude 1.022524 g's, a period of 1 second, and zero phase angle.

This gives a theoretical velocity amplitude of 62.83 in/sec and a displacement amplitude of 10 inches.

I used a time step of 0.005 seconds and I changed the default number of decimal places to 10.

After initially loading the created harmonic accelerogram, I baseline corrected and filtered the data using defaults for those processes. The velocity and displacement do show some transient behavior at start-up, but the teady state is reached after about 18 seconds of my 30 second record.

The velocity and displacement appear to have converged to the theoretical values listed above.

You may need to simply increase the number of decimal places to minimize round-off or you may simply need to perform the baseline correction and filtering of the record.

I hope this helps kvolkan.

[kvolkan]

Hi huffte,

Thanks for your help, but let me give more detail:

I created considerable amount of harmonic motions with variable frequencies and amplitudes. However neither of them converged to the expected displacement and velocity time histories (records created with 0.01 second time step and 20 seconds long)

When baseline correction and filtering is applied I observe the same behavior with you but the maximum accelaration value changes.I found it weird to face such a problem in even simple harmonic motion.

By the way, increasing number of decimal places did not change the results considerably in my applications.

Nice to meet you

[huffte]

I see kvolkan. I'm not sure what is happening internally with your problem.

You might try NONLIN fro solution of SDOF dynamics problems if you cannot get comfortable with resolution of the discrepancy here.

NONLIN is available at:

http://filebox.vt.edu/users/fcharney/CE ... loads.html

Best of luck with your work kvolkan.

[kvolkan]

Thanks huffte,

I think I will use NonLin for the rest of my work.

Best regards...

[seismosoft]

Dear kvolkan,

We confirm the suggestions and comments of huffte, noting that double integration of acceleration time-histories may indeed introduce residuals in their displacement counterparts that one should then be able to eliminate through either:
- changes in the integration parameters, such as the use of smaller value of integration time-step (see Help System)
- baseline-correction and/or filtering (adequately set to avoid changing the characteristics of the initial time-history)

Hope the above is useful.

Seismosoft Support