Working on one of the very few v-belt driven Ingersoll Rand SSR compressors I cover, My data collector first showed me about 8 ips as an overall on the input shaft bearing of the compressor. I figured that I had broken a cable or had an accel go bad and took it again-9 ips. Once more and over 13 ips. Last month was .425 ips. Looking at the input shaft I can see something moving axially when the compressor loads and unloads. I think it may have been the inner bearing race that I was looking at.

The maintenance manager doesn't seem too alarmed about the situation and didn't think much of my prediction that it might not make it through the day. This compressor has a history of high vibration at running speed which is probably due to resonance in the flimsy frame which is not bolted to the floor. It has been like this for years and they expect it to continue for years more. I gave him the number of the closest compressor rental place just in case.

sicplantssr.doc (84 Kb, 140 downloads)

Well Danny, I would say that your "data" is bad. Meaning the velocity data is pretty much useless because the integration is overwhelmed by the acceleration energy hence the "ski slope". However the acceleration data looks "good". I assume you're using an accelerometer. There is definitely a difference from now to the last reading. I would be concerned and may take some additional readings using HFD and some time waveform data. Maybe even some other directions to help prove the axial movement. Then at least you can say you did what you could even if they don't listen to you.

I would say think the ips values are probably incorrect and caused by accelerometer overload.

But the waterfall seems to provide a dramatic indication that something has changed. First the appearance of low magnitude stuff around 500kcpm, then it got dramatically higher.

If the base isn't bolted down and it's swaying back and forth, I say that's where the ski slope is coming from. Like Ron said, I'd look at the waveform data. Have any to post?

Sorry, no waveforms here.

It has never been bolted down, so that wouldn't be the cause of the sudden increase.

I use a 2" magnet here and there is probably enough high frequency energy present to excite the low natural frequency and cause the overload on the accel.

Danny, I thought you would be out golfing on a beautiful day like today!!!! I saw your data and as you know I know the Maint. Manager. The accl is (as I am sure you know) is being swamped. The g's looks kinda scary to me. I am sure the customer has a calibrated penknife that tells him everything is OK. I always recommend oil samples in cases like this, not that the customer ever does it!!!!

Doc,

No golf today. Too much work to get done before vacation next week.

He's a pretty confidant guy, alright.

I showed him the moving parts with my strobe and he didn't seem impressed. I'm no genius, but when I get a satuated signal and see things moving axially along the shaft- it's about time to get out a wrench at least. He said that they had increased the load on the compressor since last month. Probably pushed it over the edge that it has been running on for so long. The production people were elated to hear that the compressor was about to go. They said that they never had enough air and it screwed things up enormously for them.

Tell Val I'm working on her report.

Danny

Danny,

First, it is hard to rely on data consistency with Fmax=10 KHZ and magnetic attachment. Therefore, I'd disregard increased white noise in the 500,000 CPM range.

More troublesome is the ski slope. To examine this closer I'd take low Fmax, high #lines to get a longer block of data in the TWF. I'd examine TWF in g's and ips for transients and for clipping. (Use transients program if possible).

Well, it's settled. The maintenance manager was right and I was wrong.

The compressor made it through the day and two hours into the next morning before self destructing resulting in a total loss.

He was so right, he cost himself $24,000 for a new compressor, 12 or so hours of down time and 2 months rental on a temporary compressor, not to mention a nights sleep and a great deal of credibility with upper management.

And I get another dinner from the compressor rental guy.

Does the maintenance manager claim that somehow, it's your fault? BTDT.

Yeah! He probably said that Danny should have bend his arm up of his back to force him to act.

After publicly questioning my accuracy, then not acknowleding the correct call, I would expect that this is my fault-when I'm not around which is 30 days a month.

The consultant always makes a good scapegoat

Danny,
Nice call even if you were off by < 5%. Sometimes it just doesn't pay to go to work. You;d have been better off golfing.

I still think acceleration waveforms should be saved on every rolling-element-bearing, and that trending maximum peak acceleration and crest factor is one of the best indicators of bearing condition.

Nice call Danny..Atleast you may have the self satisfaction even the plant couldn't save their equipment.

Long before, I got a similar case like what Danny got.. Horzontal velocity was at 2329 with a ski slope at the spectrum. And a weird waveform in acceleration reading. You can see it in the attachment.. But we could be able to save it. I will give full credit to peakvue for that. Eventhough the veocity and acceleration readings are showing weird signature, the peakvue showed clear peaks at bearing fault frequency.

Have a nice day!!

Weird.doc (40 Kb, 26 downloads)

Why is it that guys like that always seem to make it to maintenance manager?!?!?

Denny,

Even when I shoot 100, I am better off golfing.

Rusty,

10-4, good buddy. An inherited set-up that I am planning to change.

Regardless, this was a sudden increase brought on by an attempt to increase the load on an already overloaded compressor. OK one month, dead the next. I didn't need any twf's or tsa or high resolution spectra to find this-just a walk past the room and a quick glance with the strobe was sufficient for me and probably most anybody on this board. Better data might have helped me convince them that this little compressor was already walking a thin line, but nobody asked my permission before increasing the load.

Michael S.,

It seems that the cream is not the only thing that floats to the top. I figure him to be a sharp individual, who occasionally lets his arrogance dull his judgement.

Thanks again,

Danny

quote:

Originally posted by Danny Harvey:
The compressor made it through the day and two hours into the next morning before self destructing resulting in a total loss.

So, after dust has settled what was found wrong with this compressor: loose hold down bolts, bad bearing, unbalance, or .... ?

You mentioned before that the compressor frame was not anchored to the slab and vibration at low frequency could be observed almost with a naked eye. Could this looseness cause self destruction?

David

David,

I think that a combination of looseness, and a resonant response to misalignment in the v-belt drive allowed to continue over a period of many years drove this thing to a slow, agonizing death.

The lack of anchoring probably allowed the structural crossmember supporting the compressor and motor to drop in natural frequency (less stiffness and damping) and become resonant with the 1 x running speed due to misalignment and sheave wear. Added load probably required overtightening of the belts (I havn't been able to get confirmation or denial on this) since the sheaves were worn, resulting in the failure of the bearing and the shaft breaking off. Internal looseness did the rotors in before the shaft broke.

Of course, all of this is conjecture, since the evidence was promptly disposed of.

Moral of the story (one of many):

"It has always run like that" is not a good answer. Things that run at high vibration levels for years eventually fail, prematurely and should be corrected.

Yes I agree with your statement ""It has always run like that" is not a good answer. Things that run at high vibration levels for years eventually fail, prematurely and should be corrected." This week a 1250 HP motor connected to a gearbox failed due to years if high vibration not being corrected.

Tommy