This question is probably as old as vibration analysis itself, but I am not sure if there is always a clear answer.

The problem I have is with a power train consisting of an overhung fan directly driven through a grid coupling by a 250HP 4 pole motor.

Lately I've noticed an increase in 3x only, which has actually become the dominant peak ( about 0.2-0.3 IPS). It appears there are a few very, very, very small harmonics of 1x. Vibration is present in radial and axial on both motor and the pillow block ( at a smaller magnitude) bearings. It is not a BPFO, which sometimes is close to 3x.

Yes, there is known misalignment present. The coupling has been checked and lubed.

Any experience with this? Thanks.

David.

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

DAve,

I have seen this caused by excessive clearanc in the fan bearings. Theremay be a resonant structure that amplifies the level.

Walt

Dave
A few years ago I too had characteristics of high 3X and just as Walt mentioned, it turned out to be loose fit between outer race and housing.
At first I thought it was misalignment too...I checked (not too bad) and got within spec - guess what - no change.
Since it didnt cause much problems, unit ran OK for quite a while. When it was overhauled, I became curious and thats when I found loose fit.

I remember Charlie Jackson mentioning that the physics of 3X for loose bearings was not clear to him and asked anyone in audience to explain if they could - no one could offer explanation.
But, many agreed this was not uncommon. So, for whatever its worth, maybe take a look for this possible cause.

Anyone out there hear of a good explaination for this 3X??

Regards
Jim P

I have always thought that if you see harmonics of a frequency like 3x, it makes it much more likely that there is some kind of three times per revolution event (as opposed to looseness which happens to peak at 3x).


What could be the 3x in this case.. 3 bladed axial fan? Shaft with three 120 degrees apart keyway?

Forgetting about the harmonics... on the general subject of why would bearing looseness cause 3x in particular (as opposed to 2x, 4x, 5x etc)... I also can think of no physical reason. If there happens to be half-wave symmetry [x(t) = - x(t+T/2)], then there can only be odd harmonics present. Would not seem to be a factor if your spectrum includes 6x standing above the neighboring harmonics as I understood you to say.

I think the highest peak resulting from looseness is a somewhat random result of resonant frequency of the impacted structure, overlaid with sharpness of the impact at it's orgination (like if you impact a structure with a hammer, the response is affacted both by the resonant frequency of the structure and the bandiwidth of the frequency excitation supplied by your particular hammer tip/surface combination). I have a hard time buying into the theory that there is something magic or special about 3x (vs 2x, 4x, 5x) that points to bearing looseness. It may be just my inexperience, but that's my opinion until someone convinces me otherwise.

In my experience 3xRPM and harmonics is a very strong indicator for looseness, specially the presence of 3xRPM. On the other hand, it may be anything that is loose, from bearing fit to machine mount. I have no theory why that is the case, just that it is in IRL so keep on trying to explain why. Olov

Years ago we had a 3x vibration on some fan bearings with a tapered bore. Further investigation found that they were not using the adapter sleeve that came with the bearing. There were using cheaper ones that were manufactured in a lathe using a 3 jawed chuck causing 3 high spots. They started using the real adapter sleeve and the 3x went away.

quote:

What could be the 3x in this case.. 3 bladed axial fan? Shaft with three 120 degrees apart keyway?

If the shaft acts like a beam, there will be principal axis for the area moments - it is possible that they are equal like a round shaft. 3 identical equally spaced keyways would seem to have no asymmetry when considered as a beam.

Yes, you are definitely right. The diametral area moment of inertia of that configuration doesn't change with rotation. It was probably a bad example. Still if you are seeing 3x/6x/9x, the most likely candidate is 3 of something imo.

I agree with Pete ( with all due respect to practical experience ) that expecting looseness to give give rise to 3x only, unless there is a resonance close to 3x, is hard to explain. I'll try to check for resonance on a down day.

Also, there are no 3 blades, 3 keyways or anything else I know of that may cause 3 events.

It also appears that 3x is NOT fundamental.

Important, this vibration is mainly present in the motor vertical and axial. Much less on the fan shaft pillow blocks.

I'll post plots later.

Thanks,
David

Another example of dominate 3 x turning out to be bearing slipping on the shaft:
http://www.reliabilitydirect.com/brg_slipandloose.htm

If I understand what Oli and Dave G are saying, it is not just 3x, but harmonics of 3x (3x/6x/9x) that signals a bearing looseness problem. Does anyone else view it that way?

For that pattern (with the 6x/9x etc) to be associated with loosenss and no obvious 3x exciting force is even more surprising to me. I'm not saying it's wrong, just hard to comprehend... and would like to see more proof if there is any.

A question for Dave G - do you know what type of coupling was on that machine? Is it a small machine with a Lovejoy jaw coupling perhaps? Was there an 8-element bearing... and was the resolution sufficient to distinguish 3x from bpfo?

I just thought that a very energetic and fruitful discussion that recently took place on 2x may give some new ideas on 3x. Attached is some vibration data I was referring to previously only in description. Details are given inside.

Please note that 3x is standing out ONLY at MIV. 3x is also present on other points but at very small magnitude. Also note that 3x is not a fundamental frequency, and there is no 3x/6x/9x pattern.

The motor and fan shaft are sitting on a common pedestal. Of course, there is about
10 mil of offset and 4mil/10" of angular misalignment in Vertical; and
7 mil of offset and 13 mil/10" of angular misalignment in Horizontal.

Misalignment is on the high side but can not be fixed at this time. IMO with a grid coupling it is not that causes 3x.

Thanks,
David

Sys_fan3_3x.doc (148 Kb, 28 downloads)

It may be circumstantial evidence, but if you have known misalignment and unexplained 3x, my s.w.a.g would be that they go together.

Even though we know that 3 events per revolution are not required to produce 3x, it would still be interesting to inspect the specific grid coupling in question. Maybe it will show an obvious 3x pattern unrelated to the grid spring, like the number of bolts or other obvious construction feature.

David,

What did you use for the impact test? Did you use a fairly soft rubber tip and a good size hammer to get enough energy at the 89 Hz frequency. Besides the motor case, did you also impact the support base and shaft/coupling? I still think that a natural frequency may be present. I would also inspect the grid coupling housing to be sure that the anti-rotation nub is still intact and wear marks on the hub interior, since the hub could be acting as rotating looseness. There could also be looseness in the bearing that is affected by misalignment force in vertical direction.

Walt

Pete,
The reason I think it is not coupling abnormality in terms of 3x, is because there is no 3x/6x/9x pattern. There is nothing 'of 3' in the design either.

Walt,
I did a bump test on the pedestal, but unfortunately not on the shaft or coupling. I'll try it. I also believe in resonance in this case, just can't catch it yet. Will open the coupling as well.

Regarding the bump test.. Yes, I used a 5 lb rubber mallet. But how does one know if enough energy is imparted into the structure. I thought that as long as I do have a response in the frequency range of interest ( with this mallet a response has been observed in the range of
0 cpm - 30000cpm ++ ), the mallet and amount of energy is right.

David

Regarding an impact test with a mallet instead of a force hammer: You can get a response, but you don't know if it came from the hammer. A dual-channel measurement with Coherence and Transfer Function is the only way I know of qualifing the data. Not to say that a plain mallet can't be used, it is just limited in the data provided.

Walt

Walt,
I totally agree in regards to the instrumented hammer. I just don't have it...

David

David,

I have three sizes of load-cell impact hammers (1-lb, 3-lb, and 12-lb). Have hammer will travel!

Walt