Wear and vibration in a bearing

[Manini]

Hi!
Could someone suggest me how to correlate a bearing vibration signal with its degree of wear?
Thanks
By
Manini

 

[electricpete]

As you are probably aware, defects on the races and balls of rolling element bearings can easily be detected based on impacting at unique frequencies for inner race, outer race, and ball pass.

For example for outer race defect:

In acceleration time waveform look for impacts spaced at the interval 1/BPFO for outer race defect.

In velocity or acceleration spectrum, look for a series of harmonics of BPFO peaking in the neighborhood of the bearing natural frequency 30kcpm - 100kcpm.

Demond spectrum you will see peak at BPFO.

If the defect becomes worse and starts spreading over all balls and races, then the noise floor will start to raise.

This stage might be considered wear? Not sure.

Another aspect of wear is increased clearance. Increased radial clearance may result in slight increase in 1x running speed due to eccentricity and also may result in 1x sidebands around bfpo harmonics due to rotating load zone. (assuming stationary outer race rotating inner race)

If the machine can be shut down in some cases it may be able to check clearance in place by seeing how much the shaft will move.

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

 

[EnglishMuffin]

As electricpete says, you can sometimes detect one type of "wear", such as spalling/pitting caused by subsurface fatigue, by looking at the ball pass frequencies. This type of failure usually occurs suddenly, initially at one point on the raceway, and its onset is what one is calculating when one determines bearing "life" from manufacturers catalogs. There are complications of course. If you have two pits, for example, it might be difficult to interpret the readings, depending on their location. Also, with angular contact ball bearings, the contact angle, which is one factor which determines the ball pass frequency, is to some extent an unknown quantity, since it partly depends on assembly preload and speed.
There is another technique for detecting gradual wear, known as "acoustic emission monitoring", which works at ultrasonic frequencies and there are a number of companies which sell systems which are claimed to be effective. The main problem with this approach is determining what the signals mean for any given bearing from a quantitative perspective, without prior benchmark failure data.

See this link for an overview - a web search should help you find the various commercial companies involved.

http://www.nasatech.com/Briefs/Aug00/MFS31468.html