Vibration Signatures of failed Sleeve Bearing 1

[LummyKem]

Hello All,

Can someone help with informations of Vibration Characteristics / Signatures of bad sleeve bearing such as wiped bearing, e.t.c.

Thanks for your contributions

 

[FeX32]

Vibration signatures of a faulty ball within a ball bearing are in general non-periodic in nature.
However, the other signature frequencies can be derived.
For example, if a defect occurs in the outer race then the characteristic frequency noticeable would become:

f_char=z*fr/2*(1-d/D*cos(alpha))

where, z is the number of teeth, fr is the rotational frequency of the shaft (and inner race), d id the diameter of the balls, and D is the pitch diameter of the ball bearing. (alpha is a correction factor taking into account the angle that the balls contact the outer race)



Fe

 

[LummyKem]

Thanks FeX32 for your response. I actually want informations on Vibration Characteristics / Signatures (waveform & spectral) of faulty SLEEVE BEARING and not roller type bearing which you tried to explain in your post. Hope you have info on Sleeve Bearing or Babbit Bearing?

 

[FeX32]

In regards to a sleeve bearing. Usually the lubricant within the bearing sleeves has a characteristic frequency that would effect the operating range or effectiveness of the operation.
From what I know this is called 'oil whirl'. This occurs at about 0.43*fr to 0.48*fr. At the critical speed of the shaft it is then called 'oil whip' (at resonance)
I have attached a figure of the basic spectrum of what is mentioned above.

Fe

 

[LummyKem]

FeX32,

Thanks for your response as regards oil whirl and oil whip, but what are the indications of bad/damaged sleeve bearing by looking through it's vibration spectrum and time waveform. For instance 0.41x to 0.45x TS (Turning Speed) indicates oil whirl/whip as the case may be. So what are the vibration characteristics of wipped sleeve bearing e.t.c? What characteristivs/signatures do one see in the vibration spectrum and time-waveform?

 

[FeX32]

That's a tough one. I have never actually done what you are describing. (I have mainly worked with ball/roller diagnosis)
However, a good general procedure would be to perform spectral analysis firstly. Then knowing the expected characteristic frequencies (a good start is the figure above), diagnosis the bearing based on the fact that if there exists any frequencies that are not expected, try and identify a culprit (possibly a bad sleeve ect.).
Of course if this is part of a large piece of machinery you would have to synchronize your signal or filter out the unwanted components.
By the way, if you can find a good method you should have no problems publishing your results. This is a hot subject nowadays.




Fe

 

[FeX32]

A good start would be to get a good and bad bearing and perform spectral analysis on both of them and then you can identify the general differences. (if you have both bearings)
I could imagine the main problem with these types of bearings is the fact that the mechanical impedance if very low.



Fe

 

[LummyKem]

I read that 2-channel orbit plotting analysis by connecting the vibration analyzer to the bearing using X-Y probes on the control panel can be of good assistant in this regards.

 

[FeX32]

Possibly, but I am not sure exactly what characteristic frequencies you should expect from that.
I would think that if there is any frequency component (relatively large one) that exists other then that of the expected, then there could be something wrong with the bearing.
Hope I could help,

Fe

 

[LummyKem]

My spectrum shows dominant 1x TS(turning speed) and 2x TS frequencies on both radial and axial vibration data; though with the highest amplitude recorded in the radial (i.e horizontal & vertical)direction. I need to distinguish these possible problems: wipped bearing, excessive bearing clearance and rub using the spectrum and time-waveform. This is my dilema

 

[electricpete]

I assume your are looking at housing accelerometer readings. I roughly expect to see increased running speed harmonics for increased clearances. Increased temperature if you are starting to wipe. Raised noise floor if you have a rub. As you mentioned 2-channel analysis with prox probes is preferred by many for large sleeve bearing machines.

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

 

[LummyKem]

electricpete,

Thanks for your response. On pulling out the bearing today, it was found out that the bearing was damaged as a result of wipe. This confirms the theory that "higher vertical vibration amplitude compared with horizontal amplitude with pronouced 1x TS in the spectrum" is a good indication of wiped sleeve bearing. In this case 1x TS vertical amplitude is 1.98mils (out of overall vertical amplitude of 2.3mils) while 1x TS horizontal amplitude is 0.96mils (out of overall horizontal amplitude of 1.1mils).

 

[electricpete]

Thanks for the feedback. I hadn't heard of that association between symptom and cause before. Did it show on bearing temperature?

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

 

[LummyKem]

electricpete,

Yes there was an increase in bearing temperature to shutdown level which automatically shuts down the equipment. All attempt to put the equipment back online prove abortive as it continue to shutdown on high bearing vibration and temperature.

 

[MachEng]

Most of my experience has been with diagnosing sleeve, journal or babbit bearings. I have used eddy current displacement probes comparing them to casing measurements.

The symptoms of the faulty bearing will change over time so you will notice differences in the vibration spectrum, bode and orbit plots.

Initially when the bearing first develops a fault through process abuse such as surging, cavitation or oil degredation a rub can develop. This will show up with an increased temperature, increased 1X RS radial vibration with a slight 2X RS harmonic showing looseness and the orbit will have a swing back or curl in the orbit indicating the rub. Phase angle may be unstable because of teh rub. The orbit may show a preload exhibited by an oval shape.

When the bearing fault progresses these symptoms may disappear because the rub may tend to disappear by the shaft smoothing the babbit out. I refer to this as the bearing healing itself. The primary symptom in the spectra now would be an increased amplitude at the 1X RS frequency with a more circular orbit. This can be confused as an unbalance problem. However, if you have a good base line at set process conditions one can pick the difference when there is no phase angle shift which usually supports unbalance. This is due to the increased clearance in the bearing hence stability is undermined. You may also start to see an oil whirl frequency at this stage. This healing of the bearing has been seen on steam turbines on the governor end beaing. The axial vibration may increase at the 1X RS freaquency due to the poor stiffness in the radial bearing.

Tilting pad bearings that are faulty may have an increased 1X RS frequency with sloping harmonics indicating bearing clearances opening up and a 1/2X RS with harmonics may be suttely present in the spectra as the pads will be rocking on their base.

Most of these symptoms are readily visible in the spectra with displacement probes. Accelerometers on the casing will have more noise and phase angle will not be very useful. So my preference is the displacement probe.

I trust this will be useful for you and if you have any further questions please post them.

Sorry I don't have any spectra to show you as I do not have access to any software containing history at the present.



DANIEL C. SMAISTRLA
judasmai@optusnet.com.au

Equipment: Centrifugal Pumps and compressors; Steam and Gas Turbines; Reciprocation Engines and
Compressors

Specialties: Machinery diagnostics, design application of compressors, FEED,Detail Design,construction,commissioning and maintenance.

 

[LummyKem]

MachEng,

Thanks for your contribution. What an insightful and wonderful post.

 

[spciesla]

On pulling out the bearing today, it was found out that the bearing was damaged as a result of wipe. This confirms the theory that "higher vertical vibration amplitude compared with horizontal amplitude with pronouced 1x TS in the spectrum" is a good indication of wiped sleeve bearing. In this case 1x TS vertical amplitude is 1.98mils (out of overall vertical amplitude of 2.3mils) while 1x TS horizontal amplitude is 0.96mils (out of overall horizontal amplitude of 1.1mils).

----------------------------------------------------------
I would not infer that the relationship you saw between horizontal and vertical measurements as a "good indication" of a wiped bearing. Typically there are significant differences in the directional stiffness of the bearing, i.e. for a horizontal machine the vertical stiffness is higher than the horizontal stiffness. Thus for the same force, you would expect to see higher vibration in the horizontal direction. This relationship will get more complicated when you introduce rotordynamics. Where is the machine operating in relationship to resonance? Is there a split-critical?

Be wary of drawing specific conclusions (i.e. wiped bearing) from limited data (ratio of vibration amplitudes). As other have already mentioned, utilize the oil data, temperature measurements, relative shaft vibration (proximity probes), bearing clearance information, etc.

 

[LummyKem]

@spciesla,

Many thanks for your input. Per your question, the machine was operated below it's critical speed. Will keep to your suggestions posted by you and others on the subject matter. But if I may ask you - what do you expect to see in your vibration spectrum & waveform when there is an associated problem with sleeve bearing? As in what trend or indications do one see in the vibration amplitude, spectrum and waveform before digging dip by utilizing the oil analysis data, temperature measurements, bearing clearances, proximity probes, e.t.c?

 

[TPL]

I agree with spciesla.

Unlike rolling element bearings which always have point contacts between the rolling elements and the races (allowing transmission of defect forces through to the casing surface where thay can be measure with an acclerometer), journal bearings support the shaft on an oil film - a reasonable degree of 'wipe' will still allow the supporting oil film to develop without producing a significant change in casing readings.

Depending on the machine in question, a startup from rest (with no oil film yet developed) will begin with rotor/bearing contact, allowing a (small?) amount of babbitt to be removed which does not affect the functional ability of the bearing to perform its job. Obviously with time, a significant accumulated loss of babbit will affect the geometry of the bearing surfaces and affect the ability of the oil film to do its job. It is possible that a wiped bearing will provide a greater radius and therefore greater area for area for supporting the rotor - this reduces the load per unit area and could give rise to oil whirl type phenomena.

In short, I don't think you can relate the wiped bearing to the difference between horizontal and vertical vibration.

If you don't have proximity probes, then early warning of journal bearing wear issues is best obtained through oil analysis looking for increased levels of babbitt components (lead + tin)

 

[electricpete]

I have a little bit of a followup question.

Our typical oil screen includes:
Spectroscopy by ICP or RDE (detects and characterizes particles up to 5 microns).
Wear Particle Concentration by Direct Reading Ferrography (reports large and small particle size range counts).
A bunch of other stuff not particularly relevant here (viscosity, water, etc).

Would either of the first two tests above be expected to detect babbit wear? I think the particles are too large for the Spectroscopy and non-magnetic so may be missed by WPC.

What type of oil tests can/have you detected sleeve bearing wear on?


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