On a recent Maintenance Tips You Can Use Today 3\23\06 from reliabilityweb.com by Jim Hall it stated you should lubricate prior to taking readings for trending. I would think that the trending information would be valuble in determining the frequancy of lubrications intervals with or with out the UE grease gun. Also would it not mask a bearing problem like in vibration, I personaly would not take vibration readings for a few days if I know it had been lubricated.

If readings will be compared over time, they should be taken in similar conditions, then rather you lubricate or not before the reading do it the same way each time.

"Greasing a bearing before trending", this was in reference to those using "Airborne Ultrasound" for trending. If you have ever picked up an ultrasound receiver and listened to a bearing you might automatically write the bearing up for removal & replacement. The sound of a non-lubricated bearing to those unfamiliar with the sound could in fact fool you into an early removal. However, lubricating the bearing first can quiet down a rather noisy bearing making it a little quieter. Less decibels in airborne ultrasound means less wear & tear.
For clarity, the lubricating the bearing before "ultrasonic trending", should be done several hours BEFORE the TRENDING is taken.
Yes, when using airborne ultrasound for bearing trending, keep the method "comparitive". Apples to apples, oranges to organes.......If you do not happen to grease before you take readings, maybe try changing your UT trend till a day or two after your next lubrication cycle.

P.S. I was not referring to the UE grease gun. I was referring to greasing in general. Wether or not your a fan of acoustic lubrication or not. I prefer acoustic lubrication for most because I prefer to squeeze off a half-a-stroke at a time and give sufficient time before the next half-stroke. I have seen far too many tech's start pumping full strokes...till they see grease!
Thank you for reading the ReliabilityWeb Maintenance Tips...
Jim H.

Jim,

Assuming I am not violating any copyrights, what parameter do you suggest to trend for bearing lubrication purposes: peak, average, peak/average ratio?

Do you follow vendor's recommendation to use 30KHZ center frequency or you vary +/- 5 KHZ around it to get the highest amplitude?

Did you have any luck in detecting lube problems in couplings using airborne sensor?
Do you also use this sensor for bearing lube problems detection?

Thanks,
David

This message has been edited. Last edited by: David_G,

David,
Its not trending for bearing lubrication. When taking readings using an airborne ultrasound instrument, you pay attention to the increase over a period of time.
For instance, most manufacturers of airborne ultrasound instruments support that an increase of 8-10 decibels over the previous baseline is "lubrication or Pre-Failure". Typcially you would now lubricate and retake your readings. However if you had already lubricated the bearing and not over-or-under-lubricated the bearing, you may only see a minimal if any increase in dB's.
If you have see a 10-12 dB increase over baseline it is the "Begining of Failure" mode, you may want to take note and watch.
If you have see a 10-16 dB increase over baseline it is the "Incipient Failure" mode, you may want to take schedule a removal/replacement next convenient time.
If you have a 35-50 db increase over baseline it is considerec "Catastrophic" remove/replace bearing now.
If this is all too confusing, I was taught that if an increase of 20 decibels over its normal baseline indicates a has entered a "significant failure" mode. Schedule a r/r.
This has been the normal for years.....However, today with the use of wave form analysis and integrating the airborne ultrasound instrument with a vibration analyzer, vibration techs are finding a means to determine lubrication, inner or outer race, ball defects, etc...
If you are not integrating the two technologies as yet, try it.
If not, I personally would much rather see you use a "Historical" trend for bearings when using an airborne ultrasound instrument for bearing analysis. There are very, very few cases were as you can "snapshot" or diagnose a bearing from just listening to it. In other words, so you hear repetative click, you can just about time it. Say its a spot on the race. Or, you can determine a ball defect, intermittent click. Okay. So its clicking this bearing may go for years under this load, this condition. Trending over a period of time is best.
Frequency......30 khz, "linear" if you are using a particular pistol unit (we do not name) and 38.4 khz "linear" if you are using a certain "blue" unit, (we do not name). Another "Red" instrument already has a setting built in for bearing readings.
Jim

Jim,

Thanks for the info.

I have successfully used in the past US instrumentation in combination with a vibration analyzer and few times was able to detect bearing faults using TWF and spectral analysis. I agree, data analysis these days is a preferable way over a "clicking-no-clicking-sound" type of analysis.

This time I want to hand the US gun over to a lube tech, so he can add proper amount of grease to a bearing without carrying another box and doing spectral analysis while in the field. Of course he can run into a bearing fault problem rather then lubrication starvation and not realizing it, but let's do not consider this case at this time.

Trending is always a good tool and assuming that lubrication is a potential problem and US readings are used to assess it, my question is:
'What parameter do you trend for this purpose: peak hold, average, ratio, or all of the above?'

Is it safe to conclude that you don't lubricate until increase over the baseline did not reach 8dB ?

Most of the instruments allow slight adjustment of the central frequency. Do you find it useful?

David

This message has been edited. Last edited by: David_G,

David,
Preferrably "average", I have not been a fan of Peak Hold with US because of the sensitivity of the contact probe. It can be unforgiving in terms of getting a high db.
You may have standard or regular lubrication cycles, be familiar with them. I am just saying if you do not have a particular schedule for lubrication, and you get an 8db increase you may want to lubricate and see if the reading returns to the normal or average baseline, if not you know that it the bearing is degrading. You may even use this to look back at your lubrication cycles and if they are too far apart.
I do not like to adjust frequencies. 30khz for one of the instrument is fine for its reading. It can really depend your unit and if it is digital or analog and that you can repeat the frequency setting again.
If it is analog "NO" stay with the same frequency. If it is Digital maybe, but I still like to keep everything the same as before. 2-3 khz differences can make quite a difference in db range. So why have the next guy taking the reading waundering if he is using the correct frequency?

(Cont'd)
David,
To follow-up further, anothe instrument you will find 38.4 the recommended frequency reading to trend at. Again, I like keeping the frequency setting on these instruments consistent as possible.
I might go up or down to maybe tune-in a sound, but to trend it for future reference, I will go back to the normal frequency.

If I may add one thing about frequency selection, or manufacture recommended settings (30khz, 38.4khz etc...) there are a couple things that should also be known.

1. Sensor and frequency setting. If a manufacturer recomends a particular frequency setting, its because the sensor provided with that instrument matches that frequencies optimal response. Since acoustic sensors have very steep frequency responses, or lets say very specific ones, it makes very little sense to use a 40 kHz sensor at a 30 kHz setting. True, there are other side band settings where the 40 kHz sensor may be responsive, but not as responsive as its true center frequency.

2. Bandwidth. Depending on the manufacturer, bandwidth of the filter is also an issue. If bandwidth is for example +/- 6kHz, that means the detector is sampling a frequency that is 12 kHz wide. While a wide bandwidth may be good for detecting some phenomena, it may prove problematic for others. Also, a wide bandwidth may introduce more than one harmonic per sample. Doesn't that make tunable frequency feature redundant?

Another thing you need to calibrate the sensors periodically if you are going to use it for trending, probably you have to send it to the manufacturer.