This one stumped me for a little while until I looked a little closer. I'm holdiong back a few details to make it fun.

This is taken at the inboard bearing of a 100hp motor driving overhung centrifugal pump through Thomas Shimpack coupling.

Spectrum is shown attached. The dominant feature at first look is peaks with 0.8x spacing. Inspection of spectrum ruled out rbpf pattern (no 120hz spacing).

See attached. What do you think?

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

quiz_point8spacing.ppt (162 Kb, 176 downloads)

I would guess either inner race defect or Ballspin with 2X cage sidebands.

My guess is that the majoriy of the peaks are sidebands of running speed. They are spreading into each other on the higher harmonics of the race defect frequencies.
The harmonics of the race defects are very low into the base of the spectrum and show up better in the log plot.
The previous post had said BPFO and BPFI.
I agree with the BPFI at 4.54 orders.
Unusual that the bearing would therefore have 8.31 rolling elements. unless the bearing is seriously out of dimension.
So, not sure about the BPFO, it's probably around 3.46 orders, if the bearing is still reasonably shaped.

I would say we are looking at a series of upper and lower sidebands around a center frequency. The center frequency is 4.16x, likely a BPFO. Then there are multiples of the 4.16 and again the sidebands. The sidebands are separated by ~0.4x, a cage frequency?

One or more of you guys has pretty much got it already.

To go with the kind-of strange spectrum there is a kind-of strange TWF (attached). Looks like impacting at 1384 cpm pretty close to FTF=0.386 orders for this 6316 bearing. What is your diagnosis now?

Other fault frequencies:
FTF = 0.386
BSF = 2.07
BPOR=3.084
2*BSF=4.14
BPIR=4.914

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

Do_you_think_its_cage_impact.ppt (98 Kb, 78 downloads)

I'm sorry I didn't notice before but Jan K from Alberta Canada can only be one person. Welcome back Jan!

I posted another powerpoint attached which shows everything I know about the problem.

Slide 1 - is the same as the log spectrum posted previously with frequencies labeled. It exactly fits the pattern of 2*BSF harmonics with FTF sidebands. The strange thing is that the 2*BSF harmonics are so much smaller than the sidebands. Presumably the original signal is completely modulated (multiplied).

Slide 2 - same TWF as before, repeated for information only.

Slide 3 - zoom-in on the TWF. I presume that what is going on is that there is impacting occuring at 2*BSF and it has an envelope which peaks at a frequency of FTF. I forgot to use my sideband cursor when I was at work where I have access to the database.... tried to do the same thing with powerpoint and results were a little inconclusive from this slide alone.

All of the previous slides were taken at our normal monitoring point for motor inboard, which is on the stator frame (they were all taken at 11:09).

We also decided to take a vibration reading at the bearing housing which sticks out from the endbell around the shaft, approx 2" long and 8" diameter. The remaining slides (4-7) come from that point. (they were all taken at 11:12)

Slide 4 shows the bearing housing spectrum. It looks the same as the previous spectrum to me. I have highlighted the 2*BSF harmonics and you can see the sdiebands on each side.

Slide 5 is the bearing housing TWF. This one has an envelope which peaks at 2*FTF! (previous one peaked at 1*FTF).

Slide 6 is a slight zoom-in on TWF. Again you see the 2*FTF envelope peaking frequency. Also you see the envelope is more smoothly rising and falling than at the other point.

Slide 7 zooms in further on the TWF. The 2*BSF impact rate (4.19x=14,927) is highlighted and IS conclusive. It is clear this time waveform comes from impacting at 2*BSF which is modulated by FTF in some form.

At one point after looking at TWF that I thought was impacting at cage frequency, I was concerned there was a cage problem. But now I'm very sure based on all the above discussion that it is a ball defect. (pretty much what Jan and Ralph said). On log scale I could detect harmonics of BPOR and BPIR but smaller magnitude and no sidebands on those.

I thought it was interesting case from the standpoint that both the spectrum and TWF were a little bit misleading.

So some followup questions:
1 - what do you think is the severity?
2 - Why do you think we saw the difference in TWF between stator frame and bearing housing, both measured at motor inboard? I'm not sure if that was a change due to change in vibration over time or change in location.
3 - Any other comments?

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

TwiceBSF.ppt (486 Kb, 57 downloads)

Pete,

First, thanks for sharing this lesson.

Secondly, 2nd harmonic (and possibly 3rd) of FTF modulation likely could be seen in the spectrum as well. Also, in the latest TWF, every other batch of 2xBSF is higher. 2xFTF modulation is slightly visible in the old TWF as well when it was taken on the frame, but when taken on the bearing housing, it became even more clear.

In my opinion there is a clear ball defect. Hard to assign severity though. May be you want to watch it closely for a while to see its dynamics.

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

Well! Who passed the "quiz"?

Ralph - IWOYOAYWTR
(It was only your opinion and you were totally right) ;-)

At least in my opinion. Although I'm embarassed to say it took me quite a while before I ever thought of ball defect.

Everyone passed.

The bearing should be changed out within a few months and I think I will be tasked with "root cause" determination since this motor's bearings were just changed out last year (due to a different defect). Maybe we missed something on our last cause evaluation. I'll post the pictures when the bearing comes out.

Great!!! Sounds good.

I did get to look at this bearing when it was removed in January 2006. Photo's from this inspection are attached.

The distinguishing features:
1 - discolored (stained) surfaces on races, bearing faces, shaft, cage, rollers indicating overheating.
2 - discolored grease indicating overheating
3 - Some of grease was hardened to the point it formed a clump difficult to remove from ball with fingernail - shown in photo.
3 - Ball track slightly offset from center indicating possibly some axial load.
4 - Not a lot of grease left in bearing. What was there was overheated. There was a lot of fresh red grease in the housing (but not overfilled). I'm not sure whether the lack of grease in the bearing itself indicates the bearing itself was not packed upon installation, or that the heating caused the grease to bleed out. Any thoughts on this question?

As it turns out, we have the IDENTICAL symptom right now on a completely different machine except much higher magnitudes of these non-syncronous peaks spaced at twice FTF centered on harmonics of BSF. I'll post that info if I get a chance. As it turns out, both machines also have in common the fact that these symtpoms developed between 1 and 3 years after a previous bearing replacement. We are intensely interested in this other machine so I want to learn whatever lessons I can from this previous machine.

MSDT12BBearingJan06Inspection.ppt (2,678 Kb, 55 downloads)

Now that I think about it. We may have used a torch to remove the bearing from the shaft (don't ask why, sometimes we do that if we have trouble with a puller).

So some of the heating shown in the disassembled photos may have come from that (I'm not sure). But the discoloartion shown in the first slide was definitely before that point in time (we wouldn't put heat on it without cleaning up the grease first). Also I think the discoloration on the race was not a result of using a torch based on the pattern. I think the discoloration shown on the balls was a result of using a torch.

Pete,

I do not see any damage on the balls (am I right?) to cause relatively high amplitude of 3 g's in the TWF. Have you tried to look at them under microscope?

From the above statements it appears you do not see any bearing (balls) damage other then discoloration.

Looks to me like the balls were sitting in some sort of a stationary position and spinning in a tight circle and not rolling over the full circumference as they should have been, probably due to being under too heavy of an axial load, from the appearance of the inner race path of the balls.

Only my opinion and I could be totally wrong.

Thanks, those are good comments.

Ralph- From looking closely at those marks on the ball, I am thinking they are heat discoloration marks from torch-heating of the bearing. They show up pretty well on both sides of the ball and are missing in a band where the cage would be. They happen to form concentric circles which gives the illusion they were formed by rotating motion of some kind, but I don't think that's the case. Just my opinion (I have the luxury of looking at these partsfirst-hand here in my office). But you never know.

David - I did not see any defects but I didnt look with a microscope. I guess there could be tiny defects I didn't see. Also I'm guessing maybe there could be impacting associated with caked/hardened grease particles that act like a solid particle. One example I tried to photograph was the ball shown in upper right hand corner of the last slide which had a hard patch that I couldn't remove with my fingernail. I realize it is possible that this may also be result of torch heating, but clearly caked hardened grease was also present before torch heating as shown in first slide.

Pete, I am trying to put myself into your shoes thinking of what I'd have told the manager. Was it really a bearing failure? Or just a lubrication problem which could be resolved simply by lubing the bearing? For vibration people though (at least for me) the symptoms were sufficient to justify motor removal from service.

So, what is the failure mode here if any (except of axial loading which was not detected in the data)?

Finding these items including "staining" on the metal indicating cooked grease is every bit as important as finding a bearing with some visible defects on the races. Either way, the time to failure is indeterminate. I would argue this lubrication condition is more severe than most race defects we have found. Both of these machines happen to be some of my problem child "high" D*N machines which I've talked about before. The first one was a 6313 or 6316 operating at 3600 rpm. The second one (that we haven't inspected yet) is a 6323 operating at 1800 rpm. These high D*N velocities are at the upper limits of grease lubrication (better suited to oil lubrication). The dried caked grease is a contaminant which can't be eliminated without replacing the bearing and makes the already poor lubrication (high D*N) even worse and will never get any better.

The failure mode (cause) - I'm still pondernig. Open to suggestions.

Here is a pump with BSF and FTF sidebands that I had pulled. The brgs had no visible damage. I *think* the brg is under excessive load, and thus minute irregularities of the balls appear as significant defects.

If I may borrow the line, TIMOAIMBTW

BSF.ppt (54 Kb, 29 downloads) Ball spin w/ cage sbs

Some excess load problems are caused by fit and clearance issues. One way to test for these problems is to trend envelope spectra through a range of temperature such as treending from a cold start. Usually, the inteval between measurements should be as quickly as you can repeat the measurement and should continue until the levels and temperatures stabilize. Frequencies of particular interest are:

0x - the average level

1x -

2x -

The high frequencies that are commonly used for bearing condition assessment are load generated, mostly by friction and secondarily by impacts.

If load is not a factor, the envelope spectrum signals should be stable.

To me, Michael's suggestion regarding excessive loading rather then hardened grease as an explanation for BSF appearance sounds plausable. Duncan also talks of it but these conditions in his view manifest itself in the enveloped spectrum as 1x, 2x rather then other bearing fault frequencies due to friction.

It would my speculation only to assume that BSF appearance occurred in this sequence:
loss of lubricant -> heat generation -> reduction of clearance/oil film removal -> excessive loading -> bearing imperfection defects appearance.

I wonder if SpikeEnergy has indicated a problem ?