Has anyone been able to see failure of babbit bearings through vibrations analysis? Unlike a roller bearing which has many parts that can give you signs of beginning defects there are none for a babbit bearing. We recently lost a motor which uses babbit bearings and the vibration reading just two weeks prior looked good. I was asked "How did you not see this?". I had no answer, can any one else help?

CMT,
Here are a couple of links to sleeve bearing faults and vibration. Oil analysis is your best defense for babbit health. With vibe you would look for rubs, excessive clearances, oil whip and whirl. I don't think you would see degrading babbit material.
http://www.vibanalysis.co.uk/vibanalysis/sleeve/sleeve.html
http://www.vibrationschool.com/vibeschool/mans/SpecInter/SpecInter18.htm
http://www.skfcm.com/service/support2/New%20Library/CM3...Bearings.pdf#search='vibration%20analysis%20of%20sleeve%20bearings'
try a google search for vibration analysis of sleeve bearings.
Regards,

Good comments.

I would add that temperature is an important parameter for monitoring sleeve bearings (more than for rolling element bearings).

Another interesting thing we noticed. We have six 500hp 3600rpm sleeve bearing horizontal motors driving centrifugal chillers. Two of the six had excessive oil leakage. We checked seals and bearing housings and mic'd clearances... could not correct the leakage. Then we changed the bearings and leakage went away in both cases. I think possibly the bearings had warn enough to put the shaft low within the seals leading to the leakage? (we never did do a plastigage check for bearing clearance as far as I know). In any event, when I see abnormal leakage, ONE of the things I think about is possible bearing wear. Has anyone else seen change in leakage possibly related to bearing wear?

If you are using an orthogonal pair of proximity probes then monitoring the DC gap voltage will often give warning. This can be done as simply as logging voltages once weekly or through options available in some monitoring systems such as the gap alarms in the Bently 3500 rack.

VENDOR WARNING

Vibration analysis using a combination of normal spectra and envelope spectra was described in a paper publish jointly by Exxon and Boeing in 1973. There have been a few users of this forum and its predecessor using other manufacturers systems for who have commented that they use it also. My company, Vibrotek, Inc., has an automatic diagnostic system called DREAM that includes the diagnostics of fluid film bearings as one of it's modules.

Two links to technical articles on this topic that are posted on our web site are:

http://vibrotek.com/article.php?article=articles/brn97/index.htm

and

http://vibrotek.com/article.php?article=articles/brnvi97/index.htm

Duncan

See also section 8.2 in:

http://vibrotek.com/article.php?article=articles/intelect-eng/index.htm

Thanks for all of the help. This has given me some things to think about and research. I will start adding more technologies to these machines.

CMT,
I would recommend that you monitor your sleeve bearings in all directions (horizontal, vertical, and axial) on a regular basis using velocity. I typically set my fmax ten to twelve times running speed just for the bearings with a resolution of 800 lines. Look primarily for an increase in running speed harmonics or an increase in 2x mostly in the vertical direction. Normally, I have a problem with sleeve bearings if the 2x is 50% or higher than running speed if all other problems (bent shaft, misalignment, resonance) has been verifiably discounted. If possible, I would include all of the above suggestions in the previous posts and also try using a shaft rider (stick) to measure a displacement comparison between the shaft and bearing housing (taken in 10 degree increments for 135 degrees). The comparison can be somewhat deceptive if the the exposed portion of the shaft is not perfectly round or is in rough shape. Sometimes, as the clearance of a babbitt bearing increases, the alignment of the machines is affected and your axial readings begin to increase in amplitude. Beware, babbitt bearings can be tricky due to displacement of the wiped babbitt. In other words, even with a wiped bearing, shaft displacement may remain at a minimum or as in the case of variable speeds, the harmonics may disappear with speed variances. Try to stick to the recommended bearing clearances in all cases as very small discrepancies can make the biggest differences in reliability.

The key to detecting failures and condition starts with the measurement. As John pointed out, using shaft relative transducers gives different and more useful information generally than casing readings. Casing vibration may be advised if the situation requires it.

Shaft sticks or fishtails are an old technology than can be used by hand. If you use these devices take care, because people can get seriously hurt using them.

With shaft relative probes, one can measure both vibration and position within the bearing. Use both for monitoring.

Pete mentioned bearing temperatures. Use bearing metal detectors as well as drain oil and inlet oil temperature measurements.

All this is said given that it will be worth detecting and preventing future deterioration that leads to failures. If the economics aren’t there, just keep repairing the equipment. With proper instrumentation, if just for a test, root cause may be found.

The design life for a fluid film bearing should be infinite on a motor. One doesn’t have a scheduled replacement.

Lubrication must be correct. The oil must have the proper properties, including cleanliness, supply temperature and pressure if delivered under pressure. However, if you detect bearing material in an oil sample this would mean the bearing has started to wear, which it should not – the bearing may be in the process of failing at this time. The goal is to prevent failures. With rolling element bearing, one can not always prevent failures, because they usually have a finite life expectancy, L10, etc.

Sometimes on a large motor using sleeve bearings, you will see a change in rotor slot pass vibration as the bearings wear. As the babbit wears, the rotor drops a little, which changes the air gap. This can lead to a change in slot pass vibration. I take a reading on the motor housing, at the outboard end, vertical. You have to set up the data point with a high enough Fmax to see it (50x rpm, as often recommended, usually misses it).

So, a big enough bearing drop can change the slot pass vibration. The inverse statement, a change in slot pass frequency means there is a big drop in the bearing, does not hold logically. Would you stop the motor to inspect the bearings if you had a change in slot pass vibration?

Why not measure the bearing centerline position to determine this? Bearing temperature may also show this very well if you have the temperature probes in the correct position, which for a horizontal machine would be close to the bottom depending upon bearing type.

quote:

Why not measure the bearing centerline position to determine this? Bearing temperature may also show this very well if you have the temperature probes in the correct position, which for a horizontal machine would be close to the bottom depending upon bearing type.

Bill, Would be this the load zone?
There is a characteristic behavior where is located the load zone.

Slot pass signals often show considerable variability for other reasons such as supply and load variations.

Here is an app note on journal bearing analysis:
http://www.reliabilitydirect.com/appnotes/jbfail.html

First attach of Roy Gariepi

quote:

http://www.vibanalysis.co.uk/vibanalysis/sleeve/sleeve.html

It tell :"oil whip may occur if a machine is operate at or above 2X rotor critical frequency. When the rotor is brought up to twice critical speed, whirl will be very close to critical speed"

My question is : Is that a rule?, or is associated with second mode shape (conical) characteristic for particular rotor (symetric and between bearing), or what ?

Thoughtful comments will be apreciated
Thanks in advance
Miguel Kovac

Oil whirl generally occurs at around 0.47x.

When machine speed gets near twice critical speed, the whirl frequency will be near rotor critical speed and will lock onto rotor critical speed as speed is increased further.

So I agree, oil whip cannot be expected until machine speed near or greater than 2x critical speed. (oil whirl can occur at lower speeds)

API gives good guidance with regard to placing temperature sensors.

quote:

6.1.8.1.1 Unless otherwise specified, temperature sensors
for sleeve journal bearings shall be arranged as follows:
a. Bearings whose length-to-diameter ratio is greater than 0.5
shall be provided with two axially collinear temperature sensors
located in the lower half of the bearing, 30 degrees (±10
degrees) from the vertical centerline in the normal direction
of rotation.
b. Bearings whose length-to-diameter ratio is less than or
equal to 0.5 shall be provided with a single sensor axially
located in the center of the bearing, 30 degrees (±10
degrees) from the vertical centerline in the normal direction
of rotation.
6.1.8.1.2 Unless otherwise specified, temperature sensors
for tilting-pad journal bearings shall be arranged as follows:
a. Bearings whose length-to-diameter ratio is greater than 0.5
shall be provided with two axially collinear embedded temperature
sensors located at the three-quarter arc length (75%
of the pad length from the leading edge). For pads with self-
aligning pivots, installation in accordance with 6.1.8.1.2.b is
acceptable.
b. Bearings whose length-to-diameter ratio is less than or
equal to 0.5 shall be provided with a single sensor axially
located in the center of the pad at the three-quarter arc length
(75% of the pad length from the leading edge).
c. For bearings with load-on-pad designs, the sensor or sensors
shall be located in the loaded pad (see Figure 16).
d. For bearings with load-between-pad designs, the sensor
or sensors shall be located in the pad trailing the load (A reference to a figure is given here.)

I would add that one would not be wrong to add sensors to each of the between load pads. Most likely each of these pads carry the majority of the load, and load direction may not be a 100% certainty.

Both oil whirl and oil whip occur at natural frequencies. So, why does oil whirl move in frequency? It moves because the rotor-bearing system is a parameter varying system with the bearing (and possibly other) stiffnesses and dampings depending upon rotor speed.

Oil whirl doesn't move so much in frequency because it has locked into a very flexible mode with essentially nodes at the bearing locations. Any further stiffness increase at the bearing does not significantly alter the frequency or mode shape; thus, the oil whirl frequency doesn't move much. Because this is a flexible mode (most often the first mode), this produces very high internal vibration, often damaging by exceeding internal clearances.

Some instabilities have occured at higher frequency ratios to running speed.

This message has been edited. Last edited by: William_C._Foiles,

Hi Guys,

I would personaly be inclined to use some temperature detectors, as William has described in the extract from the API paper, as well as proximity probes.

When monitoring both, I figure that some sort of correlation could be established with the temperature rise and the shaft movement away from its hypotetical rotating center.

What are your feelings about this approach ?

Not very scientific but perhaps more significant in preventing some disasters.

Marko Leo

We use all the techology's mentioned above except the probes which we use on our Turbo machinery.

The Oil Analysis is a great way to check Journal bearings.
Does this machine run 24-7 ?
If not why would you not be able to do a lift check with a dial indicator and Hydraulic jack like I use on big fans and motors.
If you lift check it every so often you get an idea of bearing wear?

Later, Jason

yes, we do run 24-7. but thanks for that idea.