Visite nuestro sitio en EspaƱol:    RSS | Contact

Detecting faults in rolling element bearings

April 15, 2009
(Vibration Analysis)

Detecting faults in rolling element bearings

One of the key goals of the vibration analyst is to detect faults in rolling element bearings. Spectrum analysis can be successfully used to detect bearing defects, although it must be said that by the time you can see signs of failure in the velocity spectrum that faults has probably already progressed to Stage Three. There are a number of ways to detect the defect at an earlier stage including acoustic emission (ultrasound), Shock Pulse, Spike Energy, Peak Vue, Demodulation, Enveloping, acceleration spectrum, time waveform analysis, and viewing the velocity spectrum in logarithmic format.

Almost all of these techniques require you to know the bearing defect frequencies: the fundamental train (or cage) frequency, the ball spin frequency, and the inner and outer race frequencies. These frequencies can be calculated if you know the physical properties of the bearing, or you can use a bearing database to determine these frequencies. But what if you don’t know the model number of the bearing? And what if the model number you have is incorrect? And what if the design of the bearing has changed (did you know that the internal dimensions of the bearing, and the number of rolling elements, can change yet the model number will remain the same)?

If you don’t know the bearing number, don’t worry!

There are a few facts about the vibration generated by rolling element bearings that enable you to confidently and accurately detect defects without knowing the bearing number:

1. All bearing defect frequencies are “non-integer” multiples of the turning speed of the shaft. For example the frequencies are 3.09, 6.91, 4.45 times the turning speed of the shaft - they will not be multiples such as 3.0X, 5.0X, 7.0X.

2. Because of the nature of the vibration generated when bearings begin to fail, there will be harmonics in the spectrum. Therefore there will not be a single peak at 3.09X, instead there will be peaks at 3.09X, 6.18X, 9.27X, 12.36X and so on.
3. If there is a defect on the inner race of the bearing, there will also be sidebands in the spectrum, separated by 1X (i.e. the turning speed of the shaft). If the defect is on the outer race of the bearing, there won’t be any sidebands (in most cases).
4. If there is a defect on the rolling elements (i.e. the balls or the rollers), there will also be sidebands in the spectrum separated by the fundamental train (cage) frequency, which is typically between 0.35X and 0.49X turning speed.

There are a few other tips regarding the bearing defect frequencies, but we can cover them next time! But notice that if you accurately determine the turning speed of the shaft, and find that there are peaks in the spectrum that are at a non-integer multiple of running speed, and it has harmonics, then it is a pretty good indicator of a bearing defect. If you also see 1X sidebands or sidebands spaced at less than half running speed, then you have an even better indicator of the bearing defect.

The clues are all there. If you follow this logic you will accurately detect bearing defects with confidence without ever looking up a bearing database.

We love to teach at Mobius Institute, so if you love to learn, visit our Web site  or give us a call: 206 842 8721


Previous tip: Root Cause Analysis (RCA) - Causal Analysis
Next tip: Milk of Magnesia - Antacid for Hot Heads

« Back to all maintenance tips

If you enjoyed this tip, fill out this form to receive our weekly newsletter. You can unsubscribe at any time by clicking on the unsubscribe link at the bottom of our emails.


There are no comments for this article yet. Why not add one?

Have your say

Comments are moderated prior to publication.
Please fill out the fields below.
Email addresses will never be published.

Comment guidelines

You can use basic HTML (a, strong, em, blockquote).
Links automatically use the nofollow attribute.
Off-topic or inappropriate comments will be edited or deleted.

ReliabilityWeb on Flickr