When to change a bearing?

Messrs. Forum:

Good afternoon, I would appreciate sharing information about when to change a bearing? I commented that we have a company that monitors the condition of the bearings and in their vibrational analysis reports they only analyze the acceleration envelope spectra and have placed as alarm limits of SKF Acceleration Envelope.

We have disassembled the bearings that supposedly had wear detected by the Acceleration Envelope and no anomaly was found.

I will appreciate your technical comments.

Original Post

A good discussion of the four stages of bearing failure can be found at https://reliabilityweb.com/art...e_stages_responses/.  It sounds like the approach you describe is catching the bearing at the 2nd stage where clear visual indicators of distress are often not present.  If the operation of the machine permits it generally is advisable to operate to stage 3.  That gives a better indicator of the actual cause of failure, especially if the bearing is analyzed by the bearing OEM.

The defects may not be visible with the naked eye. John gave a good resource on bearing failure stage. We typically wait until a bearing defect shows in velocity before changing. Our plant is not typical though, because we have 3 down days a week. It is also worth mentioning that a bearing in stage 2 failure can have less than 10% of the bearing life left. Depending on the service of the bearing it is not always advisable to allow it to degrade to a point of seeing defects with the naked eye.

Also depend on the machine.  Worked at an aluminum rolling mill where bearing defects would also be seen in the aluminum product.  Very costly so we changed those as soon as the spike energy data indicated something was wrong.  (no visible indication)  Now I am in the power industry and we usually wait until the higher harmonics start to show up in the normal spectrum.  It also depends on what the trend is doing.  I have bearings with defects that have been steady for years.

MARCO20002000 posted:

Messrs. Forum:

Good afternoon, I would appreciate sharing information about when to change a bearing? I commented that we have a company that monitors the condition of the bearings and in their vibrational analysis reports they only analyze the acceleration envelope spectra and have placed as alarm limits of SKF Acceleration Envelope.

We have disassembled the bearings that supposedly had wear detected by the Acceleration Envelope and no anomaly was found.

I will appreciate your technical comments.

Why do they use only or report only the use acceleration enveloping for analysis and making decisions on faults?

Would it not be better to use velocity as the main source of analysis and the making of a decision to replace the bearing?

Sounds like someone might be making the wrong choice in their choice of units, Fmax, Resolution, type of final analysis, etc.

Do they not collect conventional data using velocity units rather than using  a possible filtering system of analysis with a high frequency filtered nature, if that is what you are calling the "acceleration enveloping" data.

In my opinion and based on what you have said, whoever is doing the collecting of the data and the analysis of the data might have been mislead by someone during a training class or instruction class or introduction class, in the art of condition monitoring and analysis of data.

You mention assigning "alarms limits". Are they going by these alarm limits to make the calls from?

There are numerous theories floating around in the world on how to do analysis, some work and some seem to fall short of working. I think I would seek a second opinion on the reports being issued by the provider.

Just my opinion and I could be totally wrong.

Thanks and Have a Great Day,

Ralph

Acceleration Envelope is a good tool to identify bearing issues..but sticking only to that is not wise. there are so many other things to look at.

I would say replacing the bearing is the sole discretion of the analyst who has been monitoring them continuously as many things are to be kept in mind while taking the call. depends from machine to machine, bearing to bearing, next shutdown....and so on..

 

many of us want that the removed bearing should have some visible faults, but is not possible every time.

Typically acceleration enveloping, spike energy, peakvue etc... is used in order to see the initial indication of bearing or gearmesh defects which are then tracked for changes. We usually use straight acceleration peak to peak time waveform readings and not the algorithm (enveloping) to determine severity and generally look for physical movement (defects showing up in velocity) before calling for a change out. Of course this depends upon the ramp up (rate of change) of acceleration as well. Also keep in mind secondary indicators such as oil analysis, structure borne ultrasonics and infrared may also be used to reconcile what you are seeing and give a better view of the severity. Nothing beats historical data especially when you take pictures of damage which can be associated with specific time waveform and spectra data for the future.

Had a customer forced into numerous bearing changes by a third party vibration outfit using only a single peak in spike energy plot.  The owner threw a 100 page article by CSI on PeakVue and said, 'Read it, maybe you will learn something.'  

I read it alright.  Half way through the article, they warned users about making a bad bearing call based on demodulated spectra alone!

Ron, I agree.  If there is any fault, no matter how small that affects the final product, the replacement call should be made.

 

If i see something in the peakvue or impact demod or esp and nothing in the velocity spectrum I only ask to have lubrication checked. If i see a bearing defect peak with several harmonics but are low. I tell the customer to keep that in mind. If i see a defect and impacting in the waveform I ask that it be changed. Bearings can go from slight defects to very bad in a short while. When the bearing is removed you can bet there going to look for the defect you called. I had that happen in a paper mill for a while after starting there. The mechanics found the defect each time they cut open the bearing and it eventually got to where they didn't even question it. Just changed the bearing and moved on.

It depends on customer decision. The analyser will give a recommendation based on data. Also, follows up the trend of acceleration envelop. If found bearing frequency in the spectrum, first he will ask to relubricate the bearing. If the trend follows up, He may recommend changing bearings. Bearing frequency will go up very fast damage the bearing (maximum 2 months, based on my experience )

 

Add Reply

×
×
×
×