The discussion about vibration vs. misalignment on the 'Diagnosis Unknown' thread is interesting. I think it points out that a lot of what we are/were taught about vibration is in fact, "Old Wives Tales" with only a little basis in fact.

For example, you often hear, "You can't balance a machine running near resonance" or "You can't balance a machine with excessive looseness."

Last night I went to balance a fan at a flooring mill. Overhung, belt-driven, SKF tapered sleeve bearings, 1400 rpm, 6-blade paddle fan. Nothing special, just your basic fan. When started, it was obvious there was major looseness in the bearings, 0.7 ips, nearly all of it harmonics. We did a lift check, and the fan-end bearing had a 20-thousandths "bump" in it. Pulled the top half expecting to see a loose sleeve with the shaft 'turned down.' To my surprise, it looked good. Roller clearance was normal. What the heck? Closer inspection revealed a layer of crude on both housing halves at one corner... almost couldn't scrape it off, it was so hard. At some point (initial installation?) when they set the upper half in place they scraped crude off the fan housing and deposited it in the joint. Cleaning the joint, and removing a sizeable burr on the opposite joint, reduced the clearance and vibration by about half. The bearing had been loose in the housing, and the constant movement had wallowed-out the housing, so it's still loose.

Interestingly, the 1x displacement phase/amplitude readings changee very little from the original readings. We were able to successfully balance the fan in a normal manner and it was remarkably well behaved. The vibration at the problem bearing was a little higher than I like to leave it, but overall the vibration levels were quite good. Except for the noise, the quasi-random impacting, and tons of harmonics in the spectrum, it looked almost normal. It will run fine until they have a window to replace the bearings.

My point is, don't just blindly believe what you've been told/taught. When the situation allows, do your own verification. See for yourself. You will find that much of what you believe to be true, is in fact, not true at all... it's just an Old Wives Tale.

Good point Rusty. Just because someone has written something down that they saw happen once or twice and accepted it as the "rule" doesn't always make it the real "rule". Some things we have to take as a "rule" if we have no way of disproving it, but always beware of what you take for granted. In fact I never really want to take anything for granted or assume anything without proof, if possible.

I was told by a contractor who came into our plant many years ago, "you can only balance in the horizontal direction" I guess I am too much of an analyst and have to be shown beyond any doubt or at least very little doubt in my mind, that my line of thought is wrong, before I am convinced about anything and yet not ashamed to admit I was wrong if I was. My wife tells me, "you analyze everything too much Ralph."(not just pertaining to vibration alone, but life in general).

Can't help it, just my way.

Guess one good thing about becoming ancient is you forget all those 'you can't do that' stories. We had a bark hog (18000 lb rotor) that was resonant both vertical and horizontal at running speed. After much work, the resonant frequency was moved, but still affected the machine. Due to the vertical plane being the most responsive, we balanced the unit in that plane. The horizontal plan responded the same.

Have a good one,

Gary B

Rusty
Is lift check the same as clearance check?

My favorite is that if there is little or no 2X it can't be misaligned. That is a definate old wives tale that has been perpetuated through out the industry by hundreds of fault charts on office walls in facilities all over the world.

This will shatter the assumptions of many, but it is possible to be misaligned and have little or no 2X...(waiting for the fight to begin ).

Here are 2 more ‘gospel ‘truths’ to treat with caution, along with a common misconception:

One:

An instantaneous reduction in vibration when cutting the electrical power to a motor is an indication of an electrical problem .

This statement is not true for many situations.

When investigating motor related problems, if the opportunity presents itself, I have always added a tacho pulse and an appropriate number of magnetically mounted transducers and, using either an Adre or Zonic book, collected data in both delta time (with intervals of 0.1 seconds) and delta speed terms. This doesn’t give any worthwhile spectral data, but sensible overall, 1X and 2X amplitudes are achieved.

For a motor driven pump, vibration nearly always drops as soon as the power is cut. Looking at the speed v time plot shows that the speed also plummets as soon as the power is cut – the reason is that whatever is being driven by the motor (usually a pump but sometime a fan) acts as a brake, so it is impossible to determine whether the drop in vibration resulted from a disappearing electrical fault or a rapid drop in speed.

An uncoupled motor will draw little current, so only the most severe electrical problem will show up in this condition. However, an uncoupled run will give a long run down time and allow differentiation between really severe electrical and mechanical problems: Having said this, I have found uncoupled runs good for balance checks and isolating bearing problems, but I have never found an electrical problem with this.

Many of the motor problems that I have seen involve broken rotor bars: non-symmetric heating of the rotor causes it to bow giving a high 1X vibration – the bow does not disappear immediately upon cutting the power. The real test here is what the vibration is like immediately upon reaching speed from a cold start when compared to vibration after say 30 minutes on load.

I am certain that someone out there will come up with an example where ‘cutting the power’ did ‘prove’ an electrical problem but I would suggest that this diagnostic tool needs to be used carefully.

Two:
'A truncated waveform signifies a rub’

Well, I have seen truncated waveforms that have resulted from a rub, but only from proximity probes and I can understand how this happens as some obstruction stops the shaft moving within its bearing clearance and 'knocks the top off' of the sine wave.

I just cannot see how this same mechanism can be extended to bearing casing measurements. The casing moves in response to forces transmitted from the shaft. Just because the shaft motion within its clearance is restricted doesn’t mean that that casing movement is also restricted in the same manner.


Three:
Maybe not qualifying as an old wives tale, but a common misconception.

You need ‘a demodulated measurement to diagnose a bearing failure’


Sorry, but I have never seen any evidence of this.

I know there many fans of ‘demod’ out there, but I have never had a satisfactory answer to the question ‘What does a ‘demod’ reading bring to a bearing diagnosis that acceleration measurements do not. Trends of overall acceleration, acceleration spectra to 5kHZ and a waveforms are more than enough to adequately identify a bearing failure in a timely manner (and avoid having to resort to the ‘it’s a subsurface defect’ explanation when pulling bearing and finding that it has no visible problem)

I am not suggesting that demodulation doesn’t work, just that its not as good (or as essential or as valuable) as the vendors claim. Take a look back through many postings in this forum to see how demodulation readings are misused/misunderstood – often the reply to a question regarding a demodulated reading is ‘how about more readings in terms of acceleration’ or what ‘does the acceleration waveform look like?’

Buzz - I like all three of those.

I honestly think most of the wives tales start with the infamous diagnostic chart. Don't get me wrong, it's a great tool for learning and referring to. But our business is not always that simple and a few of those aren't completely explained.

The truncated waveform for rub is shown on the chart. I have heard several people claim it applies to bearing housing measurements. Just as you said, appears to be taking a plausible symptom for prox measurements and misapplying it.

Another one is rotor bar pass pattern vibration. RBPF +/- 2*LF is a symptom of broken rotor bar on the charts. I wonder at how that even got started. I guess somewhere someone must have seen a large increase in RBPF pattern during a broken rotor bar. For the most part, I ignore RBPF since it is often present on 2-pole motors and known to vary with load.

On point 2; the signal can be truncated via clipping since there are a number of ways to collect data and process data at least three come to mind.

Point 3; demod.......... I can find no use for demod as acelleration does the job head-and-shoulders beyond demod. Demodulation is ???? AT&T's 1937 reprocessing back to the original or reprocessing to what? How consistant? I don't think a new acronym needs to be added every week or so to re-invent the wheel. Work within the transducers parameters - integration is only relative, not absolute!

Gee,

I wish this discussion and the comments about misalignment and 2x could have come out 2 days earlier!!



Aubrey

quote:

Trends of overall acceleration, acceleration spectra to 5kHZ and a waveforms are more than enough to adequately identify a bearing failure in a timely manner...

For rolling element bearing failure detection, I couldn't agree more. But more interesting, you raise a great question:

quote:

What does a ‘demod’ reading bring to a bearing diagnosis that acceleration measurements do not?

Obviously, demod can give you "advanced warning of bearing failure" by detecting high freq impacting, but too often, it seems, that bearings tend to get replaced that are "less than perfect" rather than at the true end of their lifespan.

What say you, defenders of demod???

I almost always use Demod but "never" exclusive , always in companion with other indications but as a confirmation. It is if for no other purpose convinient as you can set alarms w/o considering most other mechanical problems as it´s filtered. Basically I agree, a HP filtered accel reading would be as convinient. Starting to use demod was with cavitating pumps as the noise from cavitation is suppressed in the demod process, some bearing defects could be found, the other case was for slow rotating things like dryers where in my view data was easier to analyze. So the cavitating pumps are almost extinct due to VFD so that argument is mostly gone, the other couple of details may still apply. I see no real reason for arguing this to hard, most systems can have you take both readings and I most often do that and in some cases one will be better than the other.
Calling bearings is also an art of human interface. I visit a plant where they may wait up to 12 months for changing bearings I called, then you need to be observant. I now have a double list, one with bearings for immediate attention and one with bearings to be swapped within a year. Some other plants try to only get 1 stop per year so they prefere to fix any problem that may get serious within a year and that make them do work earlier then really needed. So the world differs, normally I try to not wait to long as bearings are unreliable they may sometimes fail whenever they like and normally last longer than expected. I try to make sure even by tvisting arms that bearings are swapped before they fail as my work otherwise is a complete waste. I also try to make sure they are bad enough that damage is visible and that is normally possible with or w/o demod. Olov

Oli,

I agree with you on the demod for slow speed stuff. I think it is more useful (read that as easier to interpret) on slow speed stuff than higher speed. Also good for sleeve bearings (lower filter) and any "type C" looseness.


Dave

quote:

Some other plants try to only get 1 stop per year so they prefere to fix any problem that may get serious within a year and that make them do work earlier then really needed. So the world differs, normally I try to not wait to long as bearings are unreliable they may sometimes fail whenever they like and normally last longer than expected.

An excellent point. OLI, I normally don't need to make a call a year in advance, but for those who do, I can see where demod would be essential.

By the way, I don't believe Buzz was trying to say that demod is "totally worthless", his point was that for BEARING FAILURE DIAGNOSIS (not cavitation, looseness, etc.) all the information that one needs can be found in acceleration. We could "split hairs" on details like filtering, resolution and Fmax, but from my experience, he is correct. A one-year lead time??? Well, maybe that's another story...

Can't speak for "Demod" since I've never used it. PeakVue however, if correctly applied, can give very good insight into the state of lubrication in a bearing.

quote:

I honestly think most of the wives tales start with the infamous diagnostic chart. Don't get me wrong, it's a great tool for learning and referring to. But our business is not always that simple and a few of those aren't completely explained.

Pete, I couldn't agree more. When we first started in this function, the wall chart was useful. Now, I tend to caution the blokes against using it much.

Ian,

Not to start any more "heavy discussion", but I think when "Demod" was mentioned above, it included Peak Vue.

Dave

Dave,

Yes, I almost put a disclaimer in about PeakVue being the same as demod. I'm not sure that Mr Robinson would agree. As I understand it there are some differences, but I don't know enough about the signal processing that goes on to really be able to articulate them.

All that being said, we have for some 4 or 5 years now used Peakvue very effectively in identifying lube isues, and this has saved us a motza. Slightly off topic perhaps, but the challenge was thrown out to defend demod. We take a Peakvue reading on nearly everything, and it occasionally (read quite rarely) helps enormously in defining the the source of a problem or fault. Certainly for most "run of the mill" type bearing faults, it doesn't add much value. But then it only has to help find one or two problems, and it justifies itself for a long time to come.

Ian

quote:

that if there is little or no 2X it can't be misaligned. That is a definate old wives tale that has been perpetuated through out the industry

This would probably be my number 1 as well. We use few gear couplings in our industry. When they were used, I think this had some value.

Certainly, on power gen equipment with solid couplings one does not expect 2X either. In fact, on a 2-pole generator it would be hard to figure out from where the 2X came. Mis-alignment on a 2-pole generator should (does) change the 2X but try teasing that out from the data.

Bill, there are exceptions to everything, we used to have a couple of 4-500MW power gen´s here where specific in one LP coupling there was "severe" 2xRPM, in comparison that is, reason was that the coupling when turned in the lathe or whatever when produced was for unknown reason not 90 DEG to the shaft and still being one piece of steel it wobbled all it´s 25+ year life, it´s now been replaced by circle W rotors. It must have taken some skill to do that. On the 2-pole gen you may look at the data when not on the grid? It may not be the same w/o load but on the rotors above, 2xRPM was always there. Olov

Oli,

Your example doesn't seem that the real problem was mis-alignment.

As to looking at the generator off the grid, the design of the rotor creats 2X vibration. The asymmetry should react to mis-alignment similar to the gravity loading with 2X vibration.

I recently saw some data from a 50 Hz machine that had 400 micro-meters at 2X (1/2 of a resonance). This is considered 'normal' by some. This vibration is due to many things including the alignment (alignment again) of hte planets, but it is mechanical in nature, rotor design, bearings, supports, speed probalby play parts (This would be intersesting to deal with.) - the unit is not online when this occurs. This doesn't appear to be isolated to a single machine either.

I have even heard many times over the years "forget doing an alignment check we can just have vibration tell us if it is aligned"

THE ONLY WAY TO CHECK THE ALIGNMENT IS TO MEASURE IT

Should I repeat it?

In other words you can have misalignment without vibration.