Pump Natural Frequency 5

[badvibes]

I have a velocity/CPM peak vibration spectrum plot of an Aurora fire pump with 6 impellers. The pump operates at 3550rpm and weighs 2275lbs.
How can I determine the natural frequency from the spectrum plot? It shows consistent spikes at 2X,3X,4X,5X and 6X(BPF) and there is a high spike at 1xRPM. At approximately 2500CPM there is what I think is the system's natural frequency. Is this the natural frequency or can it just be cavitation? Is it possible to accurately determine the natural frequency from this plot?
Thanks in advance.

~Bob

 

[electricpete]

You have a peak of the vibration with the unit running. It is not particularly useful for determining natural frequency.

An ideal test is with the unit shut down. Impact it with a hammer and measure the response. Try impacting at different points and orientations to excite each mode of resonance you are concerned about.

If the pump cannot be shut down it might also be possible to get a running test while impacting and a running test without impacting and compare the difference.

Worst and most unreliable test is a running test without impacting which I think you have. It is often true that on a log scale the noise floor will rise at the point of resonance, but not always.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[vanstoja]

As suggested by Electricpete, the chances of identifying pump natural frequencies in frequency-amplitude spectra of a running pump are minimal unless you happen to have a major excitation source right at the natural frequency and the amplitude increase can be attributed to the coincidence of the exciter and the responder. Say for instance that 3X is abnormally high compared to 2X and 4X in the spectra, then it is possibly being amplified by a natural frequency. You can do this sort of resonance searching if you can vary the speed of the pump continuously and move the X-harmonics around in the frequency range. Otherwise, a hammer test with accelerometers on the non-running pump monitored by a noise/vibration analyzer could uncover generally low frequency (less than 1X) natural frequencies of the whole pump-piping complex. The 2500 cpm (0.704X) peak you see is possibly a stalled flow excitation source that has nothing to do with structural responses of the pump.

 

[Tobalcane]

Hi badvibes,

Just wanted to ask why are you trying to find the Fn of the pump? Is the pump bouncing around and you are trying to dampen it or are you going to attached a unit onto the pump and you wanted to see what if the Fn of the pump is the same as the attached unit?



Go Mechanical Engineering
Tobalcane

 

[CESSNA1]

BADVIBES: I do not know how your vibration system is set up, but it looks like you are picking up the fundamental rotaitnal frequency and then harmonics. I agree with TOBALCANE, what is the basic reason?

Regards
Dave

 

[Tobalcane]

I agree with CESSNA1 that you should be more concerned with the forcing frequencies caused by the pump than the Fn. Also, is your spectrum a PSD for acceleration G^2/Hz?

Go Mechanical Engineering
Tobalcane

 

[dasa]

Hi

Try this, you need a key-Phaser and a acc or vel meter. Track 1XN Amplitud and Phase against the key_phaser during run up/down. From this data you can se if you have a resonace pass att 2500rpm Check also phase information if you have a 180degres phase change. From this data it can be hard to know if it a rotor or a stator reconas. A bently proximeter can give you answerto determ if its a rotor resonance.

 

[badvibes]

Thank you all for your replies!

The purpose of the test was to see what has been causing a low grinding sort of noise when the pump is operating at 3550 rpm. We mounted the accelerometer on the spindle bearing and plotted the peak vibration spectrum. Nothing looked out of the ordinary except for what I thought was the natural frequency at approximately 2500CPM. Can the "disturbance" in this area be an electrical problem because it is around the 60Hz (3600CPM) area? Can the 60Hz be considered a forcing frequency that might show up like that on the spectrum?
Please excuse my inexperience with this, I am relatively new to this and am working with a co-worker who is also relatively new.

~Bob

 

[vanstoja]

An airborne audible grinding noise suggests cavitation in the pump or piping and fits with my previous mention of a possible rotating stall phenomenon causing the 0.704X (2500CPM)peak. At 3550 RPM (59.17Hz) running speed you may be operating off-design at a low flow condition which is conducive to rotating stall or rotating cavitation if NPSH margin is small. 60 Hz line frequency(LF) is an unlikely noise source though multiple odd harmonics of LF often indicate ground loops in the noise instrumentation. Even harmonics of LF (2,4,6LF,etc.)are more plausible electrical noise sources in pumps with motor drivers associated with torque development. You need relatively high resolution on your noise analyzer to even separate 1X and 1LF since they're only 0.83 Hz apart. When trying to determine if a peaks in noise spectra are discrete, narrowband or wideband it is useful to compare the 6 dB down peak width against the 6 dB down width of a known discrete peak like 1X or 1LF in the same frequency range. Any peak 3-9 times wider than the discrete peak width is likely a narrowband source, possibly signifying a relatively intense hydroacoustic noise source (flow-noise) and >9 times wider is wideband, possibly signifying a structural resonance response or a relatively moderate hydroacoustic source.

 

[electricpete]

"Can the "disturbance" in this area be an electrical problem because it is around the 60Hz (3600CPM) "

I don't think your 2500cpm vibration is likely caused by electrical origin.

great comments vanstoja. It does sound like the most likely cause would appear to be cavitation, especially if your 2500cpm vibration is spread out over a broad range of frequencies a few 100 cpm wide (broadband).

Maybe you can try analysing the available NPSH compared to required on your pump curve. Also try changing fluid system conditions and check for change in noise. For example slightly throttling a suction valve and look for increase in noise. Also increasing flow through reducing system flow resistance or increasing pump speed should increase the suction pressure drop and increase cavitation.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.