Here's some data on a Mill Supply Pump; 1X = 1800.
What's wrong?
And, life expectancy?

What_do_you_think.doc (32 Kb, 164 downloads) Vib data MS Pmp

Sam,

To me it looks like cavitation. Does the ouboard bearing look the same? There is also a possibility of bad lubrication or even cage destruction. What kind of bearings are in there?
What does PeakVue TWF say?

That data were typical - all A-F brgs. Goulds 3450 Pmp. Falk Coupling.

Any projection on failure time?

Seems like I recall someone posted, on the old board, that he can predict within a few days or maybe it was hours, when something will fail. I don't recall the name at this time, but maybe he will read this question and respond.

My questions are: what does the spectrum look like in velocity?

Why is Peak Acceleration used rather than the "standard" RMS, not that it makes that much difference except in amplitude?

Like David asked, what does Peakvue show with a 1K and 2K HP Filter?

Is this all the data you have on the pump? No history? Must have something with a trend from Nov 18, huh?

Looks like it might have been greased on day 10 where the trend went down or are they oil bath bearings or was the demand less? Can't tell why the trend went down on that day with the overall value trend. Hard to make a good call from an overall trend value. Anything could have caused an increase or decrease, ski slope, cavatation, etc. Being a water supply pump, some of the energy might be flow noise.

Naturally as you can see and have called - cavitation is a major factor. Those data are what's available. There are no trick questions.

I have already made my calls but am wanting what others think. What you think has nothing to do with what I think so I can't respond as it will mess-up what I seeking. It's an internal thing within my organization only.

It's a 30 Hz machine; electric motor driving a Goulds 3450 Pump grease lubed. I'm only curious as to what precentages it will be categorized in. Example Bearing Catergory 1, 2, 3 or 4 - Priority 1, 2, 3, or 4 etc............. How many class it as??????????? Maybe I should have made it a voting thing-a-ma-bob.

Anyway, I appreciate any contribution you can make.

Sam,

I really can't tell much from an acceleration spectrum, I just don't analyze them very often. I calculate the .428 g peak @150 hz to be .175 ips at 5 x running speed and would also guess cavitation. Velocity spectrum would be more helpful. As would PV, but I think you said that you didn't collect that.

The waveform looks like it's random and mostly due to cavitation. Unless there had been some drastic change since a few days ago, I couldn't tell anybody to shut anything down based on this. I recognize that it is probably cavitation and not right, but given only this and the multitude of real problems in a plant, I probably wouldn't recommend anything.

But then again I wouldn't make a call either way based on only this. Given the right circumstances, it might make me check further into the situation.

Sam,

The trend plot indicates that overall acceleration is not increasing, since 50-days ago it was about same level. The 5 to 10-g variation is either due to measurement issues or possibly change in pump operating state.

10-g is excessive if due to bearing defect, but no dominant fault frquency appears to be present. At 1800-rpm, bearing should be good for several months if faulty.
10-g is not excessive if Cavitation is the cause. Internal damage from Cavitation is probably not likely, assuming the measurement point was good for optimizing Cavitation measurement.

Recommendations:
1) Check when bearings were last greased and if correct lube and procedure were used. Lube as necessary.
2) Check pump operating point and adjust pressure/flow to get close to BEP.
3) Adjust measurement/analysis to improve bearing fault detection.

I generally do not make a beer-bet on machine life nor fault cause based solely on one data set, but that is my 2-cents on this topic.

Walt

The 19.34 pk-pk says somethings wrong, bearing? cavitation? need more data, resolution, etc.

Note: appears alarm levels are not correct, your fault is lower than your warning in the trend setup.

I have no idea what your looking for from the question asked, but thats my 2 cents.

Mike

Well all I can say is what I see.Your trend has doubled,P-P is high,crest factor is getting to the better check it out some more level,there is high floor noise,perodicity in the waveform is fairly random but there is impacting. the fact that the high end of the spectrum is dominant over the 1x is of concern but without the ability to play with the info it is hard to call anything.I would say this is on the list of things to dig deeper into.Sorry sam i am with the others not enough to go on

Thanks to all: I fell short but really appreciate the input. I should have done a better job but you've pointed to problems:

If voting would it be:

A: cavitation

B: Bearings

That's what I was really after.

Sam,

I can't believe you took this data.

The resolution is too low, no harmonic cursors showing, no bearing frequencies listed on the spectrum, waveform showing in time (may be your preference) rather than revs (my preference), revs not marked in waveform, no zoom in on the higher frequency area, no Peakvue starting at 1K Hz.

Let us know what it was.

Or does this quote prevent you from telling because of company policy or disagreement between you and someone else as to what the real problem MIGHT be?

quote:

. I have already made my calls but am wanting what others think. What you think has nothing to do with what I think so I can't respond as it will mess-up what I seeking. It's an internal thing within my organization only.

Ralph,
I'm sorry you don't like the data; it's an 1800 RPM AF brg type pmp - Goulds 3450. Time is in Seconds or that's what I think of as universal and Hertz; again my thinking as being universal. So 2,500 Hz is OK by me. I didn't post 10 spectra - just a snap-shot giving you a choice either 'to' or 'not' iterate an opinion regarding the machine condition. I'm not critical of anyone, just appreciate any input on what's presented. Obviously, it's cavitation and/or bearings but severe enough to say it has one month left or 2 or 3 or ???

Cavitation sticks out like a sore thumb. But also I see bearings as a problem. The trend data was from two vib techs w/the last two points by one person. Need a new tech! But all spectra were typical of that one - positions CH, CV (pmp Couping end or I/B), DH, DV & DA (pmp O/B).

I said basically one month left until brg failure and was told; you're crazy, it's cavitation. But the PdM data is all there is at this point - no massive analyses data to analyze. So, we'll know in a month or ???

quote:

Ralph,
I'm sorry you don't like the data; it's an 1800 RPM AF brg type pmp - Goulds 3450. Time is in Seconds or that's what I think of as universal and Hertz; again my thinking as being universal.

Sam,

I am sorry too. No harm intended. It is not that I don't "like" the data, it just is not enough, in my opinion and I could be wrong, for me to make a for sure call. But for someone to say you are crazy about your decision and this remark based solely on what you have posted here, is subjecting himself to an, "I told you so" pay back.I guess we'll know in a month.

If we (those who replied) had the data on a computer in front of us and this is all there was, we might could make a better guess by manipulating it in different ways, like expanding, measuring, pushing and shoving it around.

As far as time in the waveform rather than revs, that is purely a person's choice and really not having to do what everyone else does.

Again, Only my opinion and I could be totally wrong.

The file won't open on this computer.

Yeah! 1 month will tell. I wish I was really dead sure; but, I did make a commitment to one month.

Mid-DEC I gave this ~3 weeks. Today I got another look and told them I hoped it would run 24 hours and then made a wager.

What do you think?

Real_FUN.doc (32 Kb, 79 downloads) failing

Sam,

I would have to agree this one is on the way out. But I am not confident to wager 24 hours. I would recommend shutdown as soon as possible and monitor every day until shutdown was possible or until it failed and shut itself down.

It appears it is still acting like a bearing, seems like there might still be some cage sidebands around somewhere. Also some peaks that appear to be inner race.

Are they a run till fail type plant?

Sam,

After seeing your last file, I went back to original file you attached. The two data sets are obviously not from the same machine (pump and then fan). Is this a slight-of-hand contest?

I have seen fan bearings last longer than 1-month with higher vibration levels than you measured. So what's the point? Apparently the plant does not do the M-part of PDM. If no one is going to fix it soon, you might as well start an office pool on when it will fail, so at least someone can make some money from the correct GUESS!

Walt

Sam,
What other signs are showing that indicate this extreme condition? That broadband noise floor usually is accompanied by multiples of running speed and at least some elevation to 1x. If I had to base my decision solely on your spectrum, trend, and waveform, I would recommend bearing replacement ASAP.

PureCavit.doc (94 Kb, 29 downloads) SpectrumSig

I, like Walt, am a little confused... are we looking at a pump, or a fan. If it's a fan, then definitely bearings are bad.

And if it's a pump, with the trend indicated (but the graph says it's a fan), then it's also bearings.

As regards your 1st data (pump), it's hard to say. I look at lots of pumps that are very noisy, and my litmus test is to 'listen' to first the bearing housings and then the pump casing. If it's bearings, the housings will be louder with clearly metallic impacts. But if it's a flow problem, the pump casing will ALSO be noisy (bearings are noisy as well).

My opinion is that most 'noisy pump' problems are NOT cavitation... cavitation is a very specific problem that will either be seen when the unit is 1st placed in service (poor design) or after a significant change is made to the pump or system that alters the suction head available or causes the pump to run out on the curve. A more common problem, is what I call "flow noise" which can have a number of different causes, but which is not necessarily 'cavitation.'

Would like to see more data as well. From what we have been given, my guess would be flow noise or cavitation.