I use both ultrasound and temperature measurements on grease filled pillow block bearings. The bearings are 3 15/16" SKF 2220 grease filled variable speed (1500-1800 rpm) split housing used on fans. I keep coming up with the same results for these fan bearings. That is, when the temperatures get very high; as much as 50ºF over ambient; my ultrasound readings become much lower. The average dBµV data for all these bearings is 45 dBµV, but for high temperature measurements that ultrasound number drops to 27 dBµV. That is 8 times quieter! This appears strange to me as I would have believed that higher temperature would mean higher ultrasound. I’ve concluded that this must be due to the oil separation from the grease. In some recent cases I’ve opened up two bearings that Ultrasound dBµV readings were both relatively low (27 dBµV) but at the same time had temperature issues (about 25-30 degrees above other similar bearings). Both bearings were at the threshold of being lubricant deficient and the grease was hard and in some areas of the bearing the grease was crusty. There was however no bearing damage and in both cases the bearings were cleaned and repacked. After a few days of running I’ve began taking temperature readings and Ultrasound readings and the results are the temperatures dropped but the dBµV readings have nearly doubled. My question is, How does one trust ultrasound readings that trend low dBµV readings but show high temperature readings? Is this something that is unique to my situation or is this something that you also come up against?  

What's interesting about this question is that the lubrication tech is comparing two sets of data; one from ultrasound and one from temperature. On the surface it appears that the data is conflicting (higher temperature/lower ultrasound) but its really not. The inspector has faith in temperature (and why wouldn't he, its a tried and true technique for identifying faults). Higher temperature should in fact mean more friction but the higher friction is not showing up in the ultrasound data. The answer to his question is straightforward and should restore his faith in the ultrasound measurements which are a safer indication of trouble. But before giving the answer its useful to know a few facts:

• Were the two bearings in comparison on the same fan or different fans?
  • YES - SAME FAN/SAME SIDE
• Is the fan overhung?
  • YES
• Is the fan belt driven?
  • NO DIRECT DRIVE
• What is the operating temperature that the fan and fan shaft are likely to be?
  • 130ºF IN SUMMER, 15-30 DEGREES OVER AMBIENT DURING OPERATION. THE BEARING UNDER INSPECTION WAS 51ºF OVER AMBIENT
• What is different about the operation and ultrasonic behaviour of #3 bearings in question
  • THERE ARE SOME SPEED VARIANCES THAT ARE PROCESS DEPENDENT BUT THAT IS MARGINAL AND NOT AN IMPACT TO THE ULTRASOUND. ALL FAN BEARINGS BEING MONITORED ARE AROUND 45 DBµV BUT #3 FANS ARE MUCH LOWER (27DBµV). SO PROBLEMATIC BEARINGS ARE ULTRASONICALLY QUIETER BY A FACTOR OF 8 (45DB VS 27DB).
• How are the bearings greased?
  • BEARINGS ARE GREASED ON CONDITION BASIS, WHEN TEMPERATURE AND ULTRASOUND INCREASES
• Explain how the bearings were cleaned and repacked.
  • BEARINGS ARE SPLIT HOUSING. LARGE AREAS OF DRIED GREASE ARE SCRAPED AND THEN A SOLVENT WAS USED TO RINSE REMAINING GREASE. REPACK: GREASE IS FORCED INTO THE BEARING AND THEN RESERVOIR IS FILLED TO 40% CAPACITY.

Some follow up comments:

If the ambient temperature is 130ºF and the operating temperature was as much as 50ºF above ambient, that is a seriously hot bearing. The fact that there was dried lubricant means that the grease exceeded its drop point. For the grease used, it should be possible to find out what that drop point is and that will give more accurate indication of the actual temperature these bearings have seen. The temperature was taken with a portable IR gun and in these circumstances the accuracy can be hit or miss. Without set target points of known emissivity there is possibility for errors and those errors rarely average out.

Having completed the inspection, cleaning, and repacking of the bearings, and a new ultrasound benchmark of around 50 dBµV compared to others around 45dBµV it is safe to say that the efforts taken have restored a common benchmark to all the fan bearings in the survey.

If the bearing was sufficiently packed with soap, it is quite common for the rolling elements to stop rolling and instead for the whole bearing to rotate as one solid lump. At which point the ultrasound signal will reduce quite significantly. Until that is, the day when the soap starts to break up and the sliding elements are now wearing through the lubricant film on the bearing housing – at which point you have a large and sudden change in the ultrasound signal. It is also normal for a bearing which is over-lubricated to generate heat because the excess lubricant increases, rather than decreases, friction.

In the failure life of a bearing it is possible for the ultrasound signal to go down as well as up. SDT170 measures RMS energy. Consider the transition from friction to impacting and you have a good example where its possible for the ultrasound signal to decrease. Why? Because there is less RMS energy in an impact spike than there is from continuous friction.

There are two many variables to draw a safe correlation between temperature and ultrasound measurements. That's not to say both sets of data are not relevant, but its unsafe to mistrust one data set when the other data set is not fully explained or understood. Ultrasound is very good at detecting two aspects of a bearing's state: friction and impacting. Friction will generate more ultrasound in the early stages than it will increases in temperature, and a large amount of friction is necessary to generate 50ºF over ambient. Impacting will not generate heat, so it is possible for an ultrasound signal to be quite high, but there will be little or not heat. Remember that the friction sounds we are listening for with the SDT170 are the steady frying/crackling sounds.

Conclusions and Suggestions

It was interesting that while answering this question I was also helping one of our excellent customers at Holcim Cement to finalize a lubrication procedure using ultrasound and temperature. What I found significant was that Holcim had already concluded and implemented the following statement about temperature measurements:

(In addition to ultrasound) ambient and bearing temperatures are also collected and monitored.  The temperature readings are used in conjunction with the ultrasonic data to determine over-lubrication or under-lubrication.  Stable ultrasonic amplitudes with abnormally increasing temperatures are an indication of over-lubrication.  An increase in ultrasonic amplitudes, especially combined with temperature increases, is an indication of under-lubrication.  Temperature readings can be deceptive if consideration is not given to ambient temperature.  For that reason, ambient temperature is also monitored.

Cyclic fluctuations in temperatures of bearings lead to premature metal fatigue resulting in spalls, looseness, and internal clearance changes that shorten bearing reliability.   

The bearing housing shown here was completely full of grease and unable to dissipate heat properly.  Temperature swings were believed to be caused by random movement of the compacted grease. The bearing cavity was cleaned of the old grease, flushed and re-inspected for other flaws.  The bearing was not expected to have physical defects in the raceways or on the rollers as indicated by vibration analysis and ultrasound. The bearing was repacked with grease and housing cavity filled to the proper level. A ultrasound baseline was collected and routine collection and analysis continued for both ultrasound and temperature. Photo courtesy of Holcim Cement, Holly Hill, South Carolina

Why this is an interesting statement from Holcim Cement, is because it helps to explain the increase in temperature, but we still did not explain the decrease in ultrasound levels.

Why do we get higher temperature in bearings? Without friction there is no increase in temperature so we know where the heat comes from, but is it excessive due to too much friction or too much lubrication, or worse? Well, as described by Temple Inland's lubrication technician, after removing the two bearings a build up of dried grease was discovered, thickener was blocking the bearing, and in some cases it was crusted. Having all this grease clogs the bearing and doesn’t allow the heat to dissipate. Finally, because of all the grease encrusting the bearing we get a muffling or dampening effect which absorbs the ultrasound signal and make the bearing appear quiet when in fact it isn't. By cleaning, inspecting, and repacking the bearing the heat can dissipate and ultrasound from the bearing can transmit from the rolling elements to the contact probe of the SDT170.

Was confidence in the ultrasound reading restored? We will follow up with Temple Inland to confirm, but it most certainly should be. This is not a unique situation, and in fact a similar event at a General Mills cereal plant was detected using SDT170 and a shutdown was averted on cooker lines where excessive heat had dried out and cause separation of the oil from the grease thickener. Simple maintenance was used to keep the lines running. Was this article interesting or helpful?