I just pulled a bearing with signs of BSF, and FTF across the baseline. The data showed signs of increasing activity. The HFD readings also were on the rise well above what the plant has set as fault alarms of 3-4g's. The overall energy was still low, but something was happening. If all goes well with the unit I wouldn't get another shot at this pump for a couple of years so I made the call to go for it instead of watching it. I pulled the pump early compared to where I would usually make the call, but I figured we'd use the bearing as a guage for case histories. However, the bearings were pretty dang clean. I went over the bearing with a fine toothed comb, but all I was able to find was very small signs of brinelling like that you'd find from transportation or sitting still near running equipment. This arrangment was two SKF 7312 thrust ball bearings placed back to back on a small pump. I'm just wondering where the increasing roller train frequencies were coming from if there is no visible signs of wear or damage? One other thing is that this pump was on oil mist and the bearings had darkened areas aroung the race and on the cage, but nothing to obviously very wrong.

Angular, similar to thrust bearings are very dependable on preload set up. Those bearings are always "higher" then others, defects are common in spectrum. Because of tight clearances, lubrication is more critical. So I watch for trends, and HFD is very limited for monitoring the preloaded bearings. The most failures happen during installation, this is why manufacturers made double row ball bearings, to avoid installation problems. If there is installation problem, bearings will fail in short time, in first couple hours.
I learned a lot about preloaded bearings from monitoring spindles, especially testing rebuilds, based on my first data I thought spindle would fail right away. I think in your case, discoloration is probably related to lubrication problem, not proper lubricant, or to much preload, if it was not enough preload it will fail quick.
Good luck.
Dragan

One of the techniques that can be useful for diagnosing installation problems and especially where excess preload is concerned is to monitoring and short term trending the overall high frequency levels and envelope spectra. Some variation is normal in this situation but increases in overall levels can be a good indicator of the inability of the bearing lubricant to maintain proper protection for the bearing surface. The first time I ever used and found a problem this way was on the support bearings an 1500 HP Eaton Dynamitic clutch with a duplexed bearing pair - over a few hundred hours, the bearings went from a 3 mil preload condition to 3 mils of end play and was later determined to be the result of Eaton having made the shaft about .0016 oversize. The clutch wa mounted horizontally, which was the less common configuration, and it became unstable at low frequencies after the preload was lost.

Waylon,

I have read (not experienced) that you can have problems if these bearings are installed front-to-front rather than back-to-back. Were they installed correctly?

Good Luck

Danny has a good point. I've seen thrust bearings in this situation where one bearing is loaded, and the other is unloaded. The unloaded bearing would be "loose", since it is unloaded, and would have vibration signature very similar to a bad bearing.

Did you have a long term trend on this pump?

B-to-B vs F-to-F is important for shaft stability but both configurations provide bearings preloads. The main question Waylon is asking I believe is why BPF and FTF showed up if nothing wrong was in the bearing. They should not have showed up even if preload was lost, which was not the case here. It is rather a reverse situation: preload was excessive or proper lubrication was lost. In both casess bearing may produce all bearing defect frequencies and FTF in particular as it is sensitive to lube conditions. How to assess severity? I believe by trending demodulated data.

Good post, David. Short term trending of the average, 1x, and 2x envelope components can help you decide what's going on. Sometimes, a bearing outer race can stick in it's holder, either temporarily or not so temporarily and this can be seen in the 2x component. If the outer race does stick, the 1x component may increase greatly and quickly because of compression load shaft bending but may also increase and decrease if the sticking is marginal; what the one 1x does is usually very depended on the shaft dynamics and whether the angle of the rolling element tracking is proper, that is if something is slightly cocked or not. To assess the possibility of shaft bending affecting the 1x envelope component, you also want to monitor the normal spectra. As you pointed out, if the loads are greater than the lubricant film can handle, you are likely to see anything and everything. This kind of problem is something that you really need to think through; it's something that most of us see very seldom but, when it happens and you understand the mechanics of how things work, you can get correct answers to what's really wrong and solve mysteries tha other analysts fail at.

This message has been edited. Last edited by: Duncan Carter,

I just received the monthly lube oil reports which indicated a sharp rise in copper and tin. Maybe a small amount of wear is present in the cage at looks like normal wear? I usually notice this as sharp edges on the roller fit, but not this time. This cage is a solid piece.

I would post some pic and data, but I haven't used the "ADD Attachment" feature yet.

Yhanks for the input.
Nice to see that some old faces migrated over.

We have found over the years that initial indications of bearig wear are very sensitive, and the defects may not even be visible to the naked eye. examination with a microscope will show defects that may otherwise not be seen.

How much is too much? That's the $100,000 question. It depends on bearing load and the application. Once a defect starts, it's an accelerating process. Each time a ball rolls over a defect, a little more damage is done. Based on my experience, if any damage is visible to the naked eye, changing a bearing that you may not test for another year of more was a good move!

By the way, a bearing can be severely damaged and still "feel good" when turned by hand at slow speed, so don't be too quick to discount your vibration data!

Jon
Spintelligent Labs