Damping for user defined spectrum

02-Analytical capabilities
Post Reply
zbadar
Posts: 9
Joined: 18 Mar 2014, 04:57

Damping for user defined spectrum

Post by zbadar »

Dear Earthquake Experts,

I am currently dealing with Indian Standard on earthquake (IS1893). In this standard default response spectrum is given for 5% damping ratio and additionally multiplication factors are given to correct the response spectrum for other damping ratios. In my case I was supposed to use the lower damping values in the range of 0.2 to 2% which gives the multiplication factors in the range of 2.6 to 1.4 for the respective damping values. By applying these multiplication factors the response spectrums were exaggerated with higher accelerations. My next step was to use SeismoArtif to generate the artificial timeseries using these response spectrums. I chose the option of user defined response spectrum and as a result I had an additional input field 'Damping' to specify the damping (%) in 'Accelogram Generation' window. In this window I got a bit confused as giving a lower damping value as input here I got timeseries matching to response spectrum with lower PGA (and also other quantities lower such as PGV, PGD, AI, Cum. abs. velocity etc). I increased the damping (%) and the PGA was increased in order to match the same response spectra. Actually I had expected the result other way round that with lower damping the ground acceleration timeseries shall be more strong and with higher damping less strong but which was not the case here.

I further used these timeseries in our aeroelastic structural loads simulation tool for large wind turbines. Normally, I had expected higher structural loads with lower damping timeseries but that was not the case as the SeismoArtif generated timeseries was already a bit softer as explained above. Could you please give some guidance about use of Damping field in SeismoArtif in case the user uploads his custom response spectrum in order to generate artificial timeseries?

I have a second related question as well:

For artificial timeseries generation I preferred to use following default settings and generated the same successfully:
Smallest period of desired response spectrum: 0.02s
Largest period of desired response spectrum: 4.0s
Time-step: 0.01 sec

Now as a next step when I want to use the artificial generated accelograms in our aeroelastic structural loads simulation tool as input what time-step should I preferably use for my load simulation tool? Same like accelogram (i.e., 0.01s) or smaller or larger and by what factor larger or smaller. Normally for your information our interesting frequency range for different vibration modes in large wind turbine are roughly 0.25Hz to 10Hz (or say maximum upto 20Hz). Considering the same is it already appropriate to use default settings of 0.02 and 4.0s for smallest and largest periods of desired response spectrum or the range could also be curtailed?

Thank you very much to all experts for providing your insights on these topics.
huffte
Posts: 978
Joined: 22 Jul 2011, 10:19
Location: Cookeville, Tennessee, USA
Contact:

Re: Damping for user defined spectrum

Post by huffte »

zbadar, I'll attempt to address the time step question. One rule of thumb is to make the time step no larger than one-tenth of the smallest period of interest. For your case, the largest frequency of interest is 10Hz, so the smallest period of interest is 0.10 seconds. One-tenth of 0.10 is 0.01 seconds, which is the same as your specified time step. And certainly, the time step should be taken no less than the accelerogram time step.

I'll quote from page 124 of "Structural Dynamics" by Clough and Penzien, 1975:

"In general, an increment-period ratio dT/T less than or equal to 1/10 is a good rule of thumb for obtaining reliable results. If there is any doubt about the adequacy of a given solution, a second analysis can be made halving the time increment; if the response is not changed appreciably in the second analysis, it may be assumed that the errors introduced by the numerical integration are negligible."

I can only guess about the damping issue. It seems to me that it takes less shaking to damage a structure which responds at 2% damping than if the structure responded at 5% damping.
Tim Huff
huffte
Posts: 978
Joined: 22 Jul 2011, 10:19
Location: Cookeville, Tennessee, USA
Contact:

Re: Damping for user defined spectrum

Post by huffte »

As a follow-up, note that SeismoSignal initially uses a time step rqual to that of the accelerogram in generating inelastic spectra, which amounts to solving a series of SDOF systems. In the event that the solution doe snot converge, the time step is reduced to T/50, where T is the period being solved.

Admittedly, a time step of T/50 could be sluggish in solving a MDOF system. But I suppose it is possible that a time step this small could be required.

I would say that when in doubt, a sensitivity study should be done as recommended by SeismoSoft. Try dT equal to that of the accelerogram. Halve that and if there is no significant change, stop. If there is a significant change, further reduce the time step.
Tim Huff
Post Reply

Return to “02-Analytical capabilities”