Does doing FFT of numerically integrated velocity signal has some advantage over doing FFT of the acceleration signal?

[ishashukre]

Hi All

I have a very basic & fundamental question

As we know velocity measurement is independent of frequency with in the given frequency range (10 to 2000 Hz).

But, I have seen people mount accelerometer for measuring frequency with in 2000 Hz & then they do numerical integration to convert acceleration signal to velocity & do the FFT of the numerically integrated velocity signal.

Q1)
Is doing numerical integration of acceleration to velocity is equivalent to measuring the velocity signal directly. (I think its not the same, but not sure.)

Or

Does doing FFT of numerically integrated velocity signal has some advantage over doing FFT of the acceleration signal?

Q2)
If we use velocity as output measurement during the ramp test (i.e test in which input shaft speed is varied from min to rated RPM, as in case of measuring motor or engine performance (generally we measure as acceleration)). In this case when we draw the waterfall plot. The order lines will be straight, instead of inclined as velocity measurement is independent of frequency. Velocity output signal wont change no matter what is the excitation frequency between 10 to 2000 Hz. Is that correct?


Appreciate if some one can clarify above questions.

Thanks in advance

ISHA

[preston_johnson]

Thanks for your post ISHA,

Looking at the FFT in velocity has some advantages.  The aspect of integration inherently removes some of the "noise" in the FFT.  Also, the ISO vibration severity standard is listed in velocity:  http://www.reliabilitydirect.com/vibrationmeterproducts/ISO_10816.htm.

Accelerometers typically have a flat frequency response down to 10s of hertz, so you do get a solid and trustworthy measurement of vibration (in acceleration) from the machine.  Since the machine is vibrating at lower frequencies at slower speeds, it is more common to view the FFT in velocity as compose to acceleration.  At higher speeds (> 2k or 5k Hz), acceleration is a better gauge of machine vibration than velocity of displacement. 

I would use the accelerometer you have, and compare FFTs with acceleration, and with velocity to see which helps you best identify and measure the frequency of vibration you are looking for.  The NI Sound and Vibration Measurement Suites: https://www.ni.com/en-us/shop/product/labview-sound-and-vibration-toolkit.html has integration functions built in for integrating from acceleration to velocity.  These include the necessary high pass filters to remove any DC components in the signal prior to the integration step. 

Hope this helps. 

Preston