hello specialists:
i am new member joined today.I have one headache want help.
in our mill,some 3000RPM centrifugal pump they have very high impact noisy ,at time domane can see very high G level,the distance of two impact
is rotation frequency,it is not bearing innering defect I think ,but except this what's problem exist

Presentation1.ppt (414 Kb, 163 downloads) use csi 2130

Is this the year of the monkey? That monkey was dated 24 AUG 05..........may he rest in peace!

The saveall's that I have worked on were in a very, very wet condition (too wet to monitor) and prone to having water displace the oil resulting in bearing degredation and complete disintegration of the geometry.

I never took any readings on the saveall pumps, but I would imagine that they would have had spectra like you show not long before they failed.

I am a Sulzer representative. You show only one spectrum with hi frequency range and no cursor, change fmax to 1000hz and post other axis data please. What is the speed of the pump? It would appear that the primary frequency in your data is vane pass with multiple harmonics of vane pass. # of vanes on impeller X rotating speed. The pump supplies a given amount of flow, pressure and head to a system. If any of these conditions are ever changed from the original installation, this will cause the pump to operate in an unstable area of the pumps hydraulic curve. IF the impeller size is changed this will affect the pump curve, the suction or discharge lines could be partially blocked, if this supplies a heat exchanger it could be fouled and causing back pressure. More data is needed to help difine root cause, the data you have supplied could be a bearing fault or it could be vane pass, with high "G" impacting present.

thanks your comments Dave:
1.because pump velocity overall vibration lever not so high .so I think vane pass frequency maybe is not main reason.
2.maybe it's bearing fail ,but the waveform impact period quite rough,no bearing fail symptom find clear.0nly innering defect I can think about,because the impact period looks like rotation frequency ,but no innering defect frequency find at spectrum,just noise.?
3.I find pump coupling side have severe noisy compare vane side ,so maybe it coupling problem?
or there have misalighment ,I do the phase measure ,a little misalighment.
4.if there have cavitation,is it possible high G lever can find?

br lu.

There is no maybe in vibration, just facts. The facts are you have over 20g's of impacting in time waveform with 4 being normal. The frequency range on your spectrum is so high that I can not dertermine if the vibration is coming from vane pass or 1 x rpm or both.
If the coupling were at fault the drive end of the motor would also show high vibration, do you see this?
Yes cavitation can cause elevated "G" readings but not at a specific frequency like your data is showing. This fault appears to be a stage 3-4 bearing fault (its about to fail)
How is the oil in the pump? Dark?

Considering clear high amplitude impacting at 1x and noise in the low frequency range there is a possibility of some looseness developed in rotor mounting likely on IB fixed bearing. It is either a cracked inner race or loose bearing in the housing/shaft. Impeller rub also can't be ruled out. Is there any historical data? Does impeller side bearing show something similar? ASAP inspection is warranted at this time IMO.

Keep us posted about the problem found, please.

I agree there have some loose David ,some add data list

spectrum.doc (196 Kb, 28 downloads)

Another possibility to consider - maybe this not impacting at 1x but modulating at 1x.

In the first file we have a very large 1x in peakvue spectrum (slide 3) with no harmonics in peakvue spectrum. If this were impacting at 1x we would expect to see harmonics of 1x as well in peakvue. Then look at the envelope of the time waveform in slide 2 - it has a very sinusoidal appearance varying at 1x. The FFT of that smooth twf envelope is the single-frequency 1x peak in the peakvue spectrum.

Perhaps bearing impacts occuring within that big envelope?

If it were my data I would zoom in on the TWF to look for repetitive impacts at a frequency higher than 1x. I would also put the velocity spectrum on log scale and look for families of non-sync harmonics or peaks spaced at 1x or any other recognizeable patterns.

The only thing that doesn't add up is why don't we see other frequencies (like bearing defect frequencies) in the peakvue spectrum. Maybe on log scale they would show up better if they really are there. Or maybe there is broadband flow noise for some reason varying (modulating) at 1x. ... that would be different.

sorry I totally wrong ,the file that I send first and second file I send they are different
pump,but actually they have same symptom,so I confuse something.
thanks very much if you have any pump vibration experience share for me

I always struggle on how to interpret TWFs shown in the first file. On one hand we can think of it as some high frequency modulated by 1x, on the other hand - as an impact similar to a bump test pattern. In real life we always have both noise and impacting on the top. Therefore, the TWF in this case is not as clean as in an bump test, which makes it hard to interpret.

Pete has concluded that from the Peakvue spectrum in the first file it is rather modulation then impact based on presence of only one 1x harmonic. My contention is that PeakVue spectrum is calculated based on PeakVue TWF ( not from the regular TWF) which saves only peak values in every 1/Fmax time period regardless whether it is due to a bump or modulation. So, is a multiple harmonics pattern an indicator of the above issue?

Actually, it would've been interesting to know proper interpretation of multiple harmonics in a PeakVue spectrum. May be it is due to multiple bumps?

Aside from the above this case represent IMHO looseness and bumping as a result of it, and not necessarily a result of a bearing defect.

If you see envelope spectra lines that don't have significant harmonics, these are usually the result of friction loads and not impacts. A 1x line envelope spectrum line without significant harmonics is usually caused by rotation of the load zone around the outer race and the most common cause is unbalance.