One-half times line frequency vibration on motor 5

[electricpete]

I have a 2-pole 665hp 60hz horizontal compressor motor (sleeve bearing) which has increasing amplitude at exactly 1800cpm... currently at 0.27ips.
Here are the measurements in inches per second:
POS 1A 0.07
POS 1H 0.09
POS 1V 0.28
POS 2H 0.08
POS 2V 0.21
POS 3H 0.11
POS 3V 0.08
POS 4A 0.11
POS 4H 0.09
POS 4V 0.08

Here is the spectrum which shows high peak at 1800, lower peak at 1x (3584), and even lower peak at 7200:

http://209.15.74.79/pub/mab_chiller_24.jpg

Here is log spectrum which also shows some slight harmonics of 7200:

http://209.15.74.79/pub/mab_chiller_24_log_scale.jpg

Here is the historical magnitude of the 1800cpm spike at the 1V position:
2/6/01 - 0.01
6/15/01 - 0.12
8/13/01 - 0.13
11/26/01 - 0.20
1/23/02 - 0.27

Spike energy measurements are very low. I don't have time waveform yet.

Note the the 1800 vibration which is driving the high overall amplitudes is very directional (much higher in vertical than horizontal). This would normally suggest that resonance is a factor, although I suspect the excitation is also changing for some reasons.

I'm not sure if this is a hermetically sealed motor (I will check).

Has anyone every seen high 1/2 * Line frequency? What is the significance? Any suggestions?

 

[GregLocock]

I know that you have effectievly told us, but for the non-electrical amongst us what is the rotational speed of the motor? Cheers

Greg Locock

 

[electricpete]

Syncronous (no-load) speed would be 3600 rpm. (2*line frequency/2poles=2*3600cpm/2.)

The actual running speed under loaded conditions during measurement is 3584cpm (from the spectra).

 

[GregLocock]

Thanks. Half order spikes in pure rotating machines can be the result of an interestign phenomenon, called half speed whirl. I've never had to deal with it, but a web search produces many hits.

It could be related to changes in your oil or bearing clearances.

Cheers

Greg Locock

 

[FredGarvin]

Just to back up Greg's suggestion, usually oil whirl or cage defects will appear as subharmonics, usually around .45 of operating speed.

 

[electricpete]

Thanks Fred. That fact would seem to push the fault frequency associated with oil whirl below 0.5 times running speed... the opposite direction of where I'm seeing my spike (slightly above 0.5 times running at 0.5 time line frequency). But it is a relevant fact for me to keep in mind.

 

[electricpete]

Fred - when I started looking a little closer I see that whirl can occur over a range of frequencies.... most often slightly below 0.5 times... but can extend up to 65%. So once again I'm glad that you have focused me on the fact that it doesn't have to be exactly 0.5 times.

I am thinking that perhaps electrical forces may be affecting the motion of the whirl. (hard to believe the coincidence of exactly 1/2 line frequency).

Does whirl typically give higher vertical than horizontal?

 

[electricpete]

I read that oil whirl can be influenced by forces on the rotor near 1/2 times running speed.
I can conjure up an (apparently) logical explanation why the magnetic field force will sustain a highly-directional oil whirl orbit at 1800cpm (I can't explain how it starts, but once started it will continue).
I have a written explanation below and graphical explanation here ->

http://reliability-magazine.com/pub/mab_chiller_whirl.jpg

ASSUMPTIONS:
* Shaft is rotating at 3585 rpm (slightly less than 60hz)
* Field is rotating at 3600 rpm (60hz)
* ASSUME that shaft is orbiting within bearing at 1800 rpm (30hz).
* Assume that at t=0 North pole of field is at bottom (6:00) and “high spot” of orbit is at bottom (6:00).

ANALYSIS:
* Attractive magnetic force is highest when the high spot aligns with a pole.
* Maximum downward force occurs at t=0.
* Maximum upward force occurs at t=(1/60) sec when high spot is at top (12:00) and south pole is at top (12:00).
* Maximum downward force occurs again at t=(2/60) sec when high spot is again at bottom and N pole is again at bottom.
* Maximum force never occurs in horizontal direction because high spot and pole never reach there at the same time.

CONCLUSIONS:
* The vertical force is periodic with a period of 2/60 (1800cpm), and will sustain the shaft orbit at 1800cpm.
* The horizontal force is smaller.
* Vibration will be directional.

There are a few problems with my analysis:
#1 - The flux has some tangential component and the force will also have some tangential component.
#2 - This model predicts force in-phase with motion. That may be characteristic of a system well below resonance (dominated by spring constant, mass negligible), but I would expect that high magnitudes would occur under resonant type conditions where force leads motion by 90 degrees).
#3 - I have not said anything about oil forces. I don't know much about them other than that they will sustain vibration at approx 1/2 speed when excited at approx half speed. Any more comments about how oil forces would fit into this picture?

I suspect that better modeling of item #1 will help to resolve item #2 (any comments?). At any rate, an exact model of magnetic forces may not required... I think it is enough to show that a 2-pole field rotating at 3600cpm (7200 poles passing given point per minute) CAN give a directional force at 1800cpm if the shaft is already whirling at 1800cpm.

Lots of speculation on my part isn't really productive, but it's kind of interesting. I'd like to hear any comments on my attempt at an explanation for what it's worth. We'll probably start trying some experimentation shortly (change to more viscous oil and monitor bearing temperatures).

Does anyone have any links to info on oil whirl?

 

[GregLocock]

I've a feeling that you are going to find out far more practical information than is available on the web.

Anyway, to answer the little bit I do know - no, there is no particular reason why the V readings should be higher than H. It pretty much depends on the dynamic stiffness of the bearing housing in those two directions at that frequency. A modal test would tell you if that is the correct explanation.

http://www.bkscms.com/Bks_cms/Applications/VibroDiagnose/GbdiagnoOil_Whirl.htm

Gives a simple story for oil film whirl. I wouldn't get too hung up on whether it is 49% or 51% or whatever, I suspect you are delving into the interaction between two non-linear systems, where rules of thumb are likely to be more fruitful than deep analysis.

Good luck
Cheers

Greg Locock

 

[sms]

For good information on oil whirl, look at

http://www.bently.com

there are lots of articles in their orbit magazine archives on the subject. In my opinion it would be a tremendous coincidence if the oil whirl frequency were exactly 1800 cpm. Whirl in general is 40% to as much as 60 to 70% of running speed, but it is running speed related. The oil force is due to circumfrential flow of oil around the shaft. The reason that oil whirl is generally in the 42% range is due to fluid velocity. If you look at the journal bearing in terms of two plates with oil between them, the top plate is moving at shaft speed and the bottom, or bearing surface is fixed. So the oil velocity at the shaft surface is shaft velocity, and the oil velocity at the bearing is zero. Simply averaging the velocities would give you a 50% of running speed vibration with oil whirl, but if you go back and look at your college fluid dynamics book, you’ll see that using the two plate model the average fluid velocity is actually around 0.43 X the velocity of the top plate. In order for the velocity to be above 50% something besides the shaft has to be adding velocity to the fluid. Oil whirl orbits by definition have to circular as the shaft is circular so the forcing function is not directional. Also if the bearings are properly loaded, oil whirl can’t really happen, as there is little circumfrential oil velocity. The best way to stop whirl is proper bearing load. Due to resonance’s or asymmetric stiffness the response at the bearing cap might be directional. It is also not unusual to see the oil whirl lock into a rotor resonance, so if you had a rotor resonance at exactly 1800 cpm, you might see what you are seeing, but that would be a tremendous coincedence. Typically if a 3600 rpm motor is running above the first critical, the first is around 2700 cpm.

If you don’t have proximity probes on this motor, I would install them. I would then gather spectrum data versus speed, as well as 1x amplitude and phase vs speed. The cascade or waterfall plot in combination with the bode plot would go along way toward sorting this out. I don’t kow what it is, but my gut is telling me it is either electical or magnetic.

 

[electricpete]

I agree that exactly 1800cpm (within 1/2cpm) can't be a coincidence. But isn't it possible that the electrical force (shown in my link) is helping to excite the oil whirl?

Here is a laundry list of troubleshooting actions we are considering:
- High resolution spectrum to make absolutely sure the peak is at 50% of line frequency (it has been repeatably near that
frequency with our low-resolution spectrum).
- check oil condition (the last sample was Fall 2000 before vib started increasing)
- phase check - 90 degrees between horizontal and vertical vibrations would be one characteristic of oil whirl
- also can help to rule out misalignment
- prox probe monitoring to determine shaft orbit? (YOUR IDEA)
- perform motor winding resistance and insulation resistance checks to look for motor anomaly
- change oil
- check whether vibration changes with compressor load.
- Alignment and soft-foot check?
- Inspect bearing condition (and air gap clearances).
- check for base looseness which can cause high ½ running speed predominantly vertical.

Does anyone have any more thoughts or comments?

 

[AGBurd]

I have been out of the vibration analysis business for a little over three years now so I am working with a fading memory here. A simple test to eliminate electrical induced vibration problems is to watch the spectrum during a coast down (if operations will allow a momentary shutdown). An electrically induced spectral component will dissappear as soon as the current to the motor is off. A mechanically induced spectral component will decrease over time during the coast down. Also, for the electrical faults that I have seen, there was an audible hum in the motor. Good luck!

 

[dgallagher]

I would agree that it is 1/2 line freq but what catches my attention first is that the vertical measurements are higher than the horizontal. This is not normal and is usually caused by a problem with the base, grouting, or hold down bolts. Also the vibration increases in two steps, sort of like something gave way or snapped. If possible, shut it down and do a bump test.

 

vibration at half or near half running frequency is known to be oil whirl or oil whip a charactaristic of sleeve bearing instability , please check if this bearing is of offset or lobe or lemon type , these types of bearings are madw to solve oil whip oroil whirl problems that occur at such frequency.
please make sure that these bearings are correctlly installed as the direction of offset is very important.
Spike energy measurement would not help in case of sleeve bearings , it is useful in case of antifriction bearings like ball or roller bearings where bearings with considerable damage do not show much of high vibration on bearing housing and hence can not be detected by normal vibration measurements so in this case spike energy measuremnts would be the answer
Best regards
Sherif Abdalla
Senior Support Engineer
ADCO oil Company

 

[electricpete]

Thanks Sherif. I'd offer a start but there's no way to do that for a visitor.

I am terribly embarassed to say that we now have identified 1800cpm present at 0.3 ips on the motor bearing when the machine is not running! At first guess it seems most likely originating from 1x vibration of our syncronous 4-pole turbine-generator located 50 feet above and several hundred feet away horizontally.... apparently traveling through the building structure. It's the only syncrounous machine in the plant.

Not quite as exotic as oil whirl :-(

 

[GregLocock]

Well that made me laugh. Interesting thread in any case! Cheers

Greg Locock