Hi folks,the problem high vibration on centrifugal chiller motor outboard end about 10 mm/sec rms, inboard end radial levels normal but axial levels comparitively high on all parts of machine.

Abnormal noise can be heard from motor, spectrum shows high 1X with what appears to be pole pass sidebands also peakvue shows a clear fault.

i am wondering wheter it is rotor bar fault or a bearing problem. The 1X ppf sidebands and high slip seem to indicate a rotor bar prob but i see no RBPF pattern nor a 2 Lf.To support the bearing problem i dont yet have the bearing info nor can i see any bearing fault pattern in the normal spectrum.
No history of machine availible i am planning to do current signature measurement later

Chiller_Motor.doc (43 Kb, 88 downloads) Spectrum

Looking at the sidebands of slip times # of poles in the velocity spectrum, it does look like a rotor bar problem. But on the other hand, it is displayed in a log scale. Is this same type data showing in a linear scale as clearly?

Looking at the Peakvue data, the 2.155 order peak sort of favors possible ballspin. Which seems to be also in the log scale just to the right of 100 Hz. What does this 2.155 orders look like in linear scale and acceleration units and not in Peakvue units?

What does this abnormal noise sound like, a beat sort of like going and coming in a regular countable frequency (like once or twice per second) or just a constant noise like a roar?

Thanks Ralph,
The sidebands are not apparent in the linear scale as the 1x is very dominant.Peakvue is in linear scale and have posted an acceleration speactrum i understand that is what u ar asking.i have not found a 2.155 order peak nor any of its harmonics in any of the spectrums in log or linear scale.

if it is a rotor bar problem why the motor inboard levels are low (around 1.5 mm/sec), one possibility is the motor is horizontally flange mounted on the compressor casing and the outboard end is basically overhung.

The abnormal sound appears to be a constant grinding noise

Chiller_Spectrum_2.doc (44 Kb, 50 downloads) spectrum 2

The problem is probably of damaged rotor bars. RBPF points more towards bar looseness & eccentricity. 2FL too would show up for coil looseness or eccentricity.

I saw a case recently where overall vibrations were under 2 mm/sec, RMS; but lots of PPF sidebands areound the 1X harmonics. On inspection, 19 bars were broken, out of 50. This was a 400 KW, 4-pole machine.

The grinding noise could well be due to broken bars causing a rub.

Regards,

Aditya

quote:

On inspection, 19 bars were broken, out of 50. This was a 400 KW, 4-pole machine.

WOW!Aditya,

How was it still even running?

Thanks for the data viki,

quote:

horizontally flange mounted on the compressor casing

Is it possible the 2.1 orders is from the compressor?
Is the noise in the motor only or in the compressor also?
Are there any gears to cause a speed increase that might be related to the 2.155 orders?

It is stil a little hard to read the spectrums. With the high vertical scale of 14. If you could look at it with a vertical scale of somewhere around 2 or 4, we might could see if there is anything there at the 2.1 orders. Of course you might have already done that. Also the sidebands might show a little better with a lower scale in velocity.

I see some noise in the 3K Hz spectrum in acceleration . This may be where Peakvue is getting its data from. Is there anything on past 3000 Hz?

What are the small peaks on the low end in your Peakvue spectrum, .4 orders for cage?

The log scale realy shows good sidebands of #P times S. Might well be rotor problems.

What version of RBMware do you use? You could send me the machine in a small database where I could look at it a little closer. Just a thought.

Help us out here EPete!!!

I'm flattered that you consider my opinion worth asking. So I spent some extra time to give a good answer, and here's what I came up with: Ummm. Beats me.

Looks like a few things going on
1 - High 1x (10mm/sec)
2 - pole pass sidebands around 1x and harmonics.
3 - impacting at 2.1x. It shows up as 10 g's true pk/0 on the peakvue but much lower on the normal acceleration... I guess the sample rate is too low to catch it on the normal acceleration.
4- abnormal noise

(I didn't think the raised noise floor in g's was that unusual considering the low magnitude and the high frequency, but I might be wrong.)

Which are related and how? I liked the theory about the rotor bars causing pole pass sidebands and high 1X and noise until Ralph had to go and spoil everything and find that 2.1x impacting.

Ralph mentioned investigating what might be coming from the chiller and I agree - it should be easy to check by looking for the frequency on the compressor and the base. Easy to imagine that there might be transmission if they are mounted on a platform on a tank (receiver? condensor? I forget what you call it). And I know centrifugal compressors could be a good candidate for stuff non-sync to motor due to internal gearing or due to suction conditions at low load causing something like a rotating stall.

What kind of bearings (sleeve, rolling)? How many horsepower? Is compressor the same speed and what type of bearings does it have?

Even if the 2.1 remains a mystery, checking the current signature is the next logical step as everyone else already mentioned.

This message has been edited. Last edited by: electricpete,

One thing I just noticed is we're looking at a log scale and the pole pass sidebands around 1x are only perhaps 1-2% of the fundamental. So those may not mean much (but current signature will tell).

What is line frequency in India.....50 Hz? If so, the mystery 2.1X is 2X line frequency.

David Eason

Good point. That's something I never thought of.

The frequencies don't seem to quite line up. Running speed was 48.8 and 2*LF is 100.
100/48.8 ~ 2.05. But maybe the 2.1 was rounded up or there was a change in load between readings and the speed changed?

David,

May be 2X LF but the Peakvue spectrum has it listed as 105.17 Hz. HMMMMM! Could 2X LF be that far off?

You have gotten awful quite viki!!!! Help us out here.

Ok, look at the attachment. Notice that the top plot is the exact same plot as the bottom plot. They have the same name and same time stamp. The cursor in the top plot is on 5625.0 cpm and the cursor in the bottom plot is on 5632.7 cpm. How can that be if they are the same plot? In the top plot I pressed the "X" key to toggle through the harmonic families. In the bottom plot I pressed the "Z" key to locate the peak. I still believe that viki's peak is 2xlf; the pattern of that spectrum is very common to 2xlf in peakvue.

David Eason

2xlf.doc (132 Kb, 27 downloads)

Sorry guys was busy had no time to get back.Still unable to get any data on the machine but to eloberate on the machine description it is a sealed motor cooled by the refrigirent single shaft no coupling, the impellers are mounted on the motor shaft and the motor is flanged to the compressor casing, so the complete assembly is at the same speed, no gears.appears to have roller bearings at the outboard end but am not too sure. its a 1000 ton chiller

The radial vibs at the motor inboard and compressor are very low(~1.5 mm/sec) but axials at all points higher (~5 mm/sec) all 1x dominating. peakvue plots on compressor casing show nothing, motor inboard shows a similar pattern to what i posted for the outboard end but amplitudes are much smaller(maybe picking up from the outboard)

Thanks Aditya

Ralph the plots are in mastertrend i will try to convert to RBMWare 4.8 and mail it to u.

David the peakvue at 105.5 Hz is too far away from the 2Lf ie 100 Hz(just my opinion)

Pete this is supposed to be a low slip motor, so the present motor speed worries me.Im waiting for the current spectrum

quote:

Ralph the plots are in mastertrend i will try to convert to RBMWare 4.8 and mail it to u.

Thanks viki,

Mastertrend is ok.I have access to MT also.

Mail to:
ralph_stewart@alertanalytical.com

quote:

Pete this is supposed to be a low slip motor, so the present motor speed worries me.Im waiting for the current spectrum

The speed is awfully low if it a "low slip" motor (72 rpm slip) or there abouts. What is the running amp load? What is the amp load on the manufacture's tag?

This message has been edited. Last edited by: Ralph Stewart,

David,

I am not disagreeing with what you are showing us in your plots. Please do not be angry. I have seen some things that are really strange in the software over the years also. Especially in the Mastertrend when I have multi plots displayed. The cursor is no where close to what the reading is on a single plot of the same data.

But in your example you are using cpms and Aditya is using Hz.
The differences in your example is somewhere around 7 cpms (.128 Hz) for the same peak used. Where as viki's is 5.17 Hz above 100 Hz which is like 310 cpms.


Only my thoughts and I could be totally wrong. Maybe when viki sends me the data we can help him confirm which is which, huh?

This message has been edited. Last edited by: Ralph Stewart,

Ralph,
Being angry never entered my mind. We're all exploring possibilities.

David Eason

quote:

Ralph,
Being angry never entered my mind. We're all exploring possibilities.

David Eason

Thanks David. Some get ruffled easily. And I surely don't want to make anyone get that way. Not on purpose or accidently.

Viki,

interesting data and discussion. I have some ideas but I'll reserve opinion until I see the data. Any more info on the chiller? Will you please email the database to me if you get a chance?
bill.kilbey@mobiusintitute.com

This message has been edited. Last edited by: Bill Kilbey-Mobius,

Viki
Your first plot - the log scale with pole pass side band, is obvious: there is a problem in rotor of your motor. Maybe broken bar or blow shaft.

100 or 105 Hz component is lower and no means problem.

Current analysis will help but is not necessary.


Ricardo Góz from Brazil

Viki,
Broken rotor bars, in a log scale display will usually have multiple sidebands (4 or more on both sides). I don't think you have a bad rotor. It does look like resonance under the 1x peak in the log display. This would cause deflection of the rotor in the airgap and that would cause the slight modulation you see. Historically, has this been present? If not, then you need to consider that your 1x has increased, evaluate what problem most fits this increase in 1x and then take action. I have also seen sidebands that do not match slip that were due to a bounce in the bearing cage.