accelerated vibration test

[gio1]

Hello

I am planning an accelerated vibration test on a shaker rig, starting from time-history data of measured accelerations.

My approach is as follows:
- calculate gRMS of measured data
- calculate gRMS of profile to be used on rig for a (say) 3x accelerated testing time, using Coffin-Manson power law
- calculate corresponding acceleration spectral density (g2/Hz)

However I am a bit puzzled as to how to assign the correct PSD to individual frequency ranges.

The first problem is how to allocate PSD figures to individual ranges so to comply with rig limitations (max displacement at low freq) whilst maintaining the overall target gRMS.
Do you think I could filter the measured data through separate bandpass filters, then calculate the gRMS for each frequency range and apply the Coffin-Manson rule to each individual range? After this I could play with PSD values in each interval, as long as I make sure that the overall gRMS is still the same?

The second problem is that the rig is only capable of frequencies up to 2kHz and my measured data contains some non-negligible gRMS which falls above this threshold. How can I make use of this data on the rig profile?

Thanks

Gio1

 

[GregLocock]

Not exactly answering your question but are you satisfied that pseudo random will give you the same cycles to failure as your real signal?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[gio1]

Hi Greg

I don't have information on the failure of the part, only measured acceleration data on it. The part has never failed in service but I'm trying to achieve a failure on rig through accelerated vibration testing, and I think that using an amplified random vibration based on a measured profile is reasonable, but I'm not sure about my detailed approach, especially as the rig cannot reproduce the full width of the operating spectra

Cheers
Gio1

 

[floattuber]

For the second part of your question...the shaker can probably go a little higher than 2000 Hz, maybe 2500 or a little higher. Our Unholtz Dickie T2000 and smaller tables can do 2500 no problem.

However, much beyond 2000Hz and the displacements are very small and therefore much less destructive than lower frequencies. In my industry, we don't care about things much above 2000hz but maybe that's different for your industry or part.