The following spectrum is from a belt-driven, horizontal, rotary seperator, with rolling-element-bearings. The measurement was taken at the non-drive end, horizontal direction, on the bearing housing with an accelerometer, 400 Hz, 800 lines. This is a heavy unit; I'd guess the housing weighs 1500 lbs. with a 4000 lb. rotor running 1217 rpm. The base has 4 bolts across either end, with the base plates sitting on rubber/cork pads. They are squeezed down so tight, that there is very little "flex" left in them. This unit does have a lateral resonance around 700 rpm.

I was just curious as to what your opinion of this problem is.

Here is the waveform.

I love that waveform. Noticed that it is in velocity. I use that when I suspect an alignment issue, worn insert etc. Is that what you're thinking?

It is obviously looseness, likely in the housing mounting. The question is 'Where?'. Firstly, I guess vertical reading could've given a clue. Secondly, horizonatl reading TWF points to a restriction causing the housing to bounce off in one half of the cycle.

The problem is not in relatively high velocity reading but in the pattern. If the cork isolators were not damaged a SIN wave form should've been generated in a perfect case.

My two cents.. (without knowing the equipment in details)

David

I would guess that the worn pads are allowing the base to be sprung and that the natural frequency will change when they are replaced.

Then you may find some loose anchors or something like that along with imbalance or misalignment.

The TWF is one of Misalignment. The "W" pattern.

Dave

Rusty,

I'd be looking at V-belt alignment and tension.

Walt

Dear Rusty,
the spectrum shows an orginal case of looseness..... even though waveform doesnt....
i suguest u check the face difference between bearing houing and the base"verticaly"...difference in magnitude of vibration and the face would show looseness in bolts or in structure ....even though to me this case looks like internal looseness cass inside the bearing due to partial harmonics.....so u might want to check bearing clearance ,shaft wear and bearing housing
peakveu shoot would be helpful too
regards

quote:

Originally posted by eng mohamed teleb:
Dear Rusty,
the spectrum shows an orginal case of looseness..... even though waveform doesnt....
i suguest u check the phase difference between bearing houing and the base"verticaly"...difference in magnitude of vibration and the phase would show looseness in bolts or in structure ....even though to me this case looks like internal looseness cass inside the bearing due to partial harmonics.....so u might want to check bearing clearance ,shaft wear and bearing housing
peakveu shoot would be helpful too
regards

lol...sorry i ment "phase"......honest typo mistake

I guess I would look closely to the belt alignment and would be interested in what the phase readings looked like on the bearings and base.

Wow, you guys are really good! One point, one spectrum, one waveform and you have the diagnosis.
Just having fun. I'm sure Rusty is setting us up for the old "baseball stuck in the blade" or classic "technician left his right shoe in the seperator" surprise, or something similar. It is fun to guess with no consequences tho.

It wouldn't be too interesting or fun if it was just plain ole' misalignment/looseness now would it?
How long will Rusty continue the mystery? Stay tuned...

quote:

It is fun to guess with no consequences tho.

For a change

It's typical of many questions posted where little or no data or no detailed machine descrition is available, and the questioner wants an exact answer. I doubt that a single spectrum and waveform was all that Rusty had available to make a diagnosis! I also doubt that he would do that without observtions and more data. Games!!

Walt

quote:

Wow, you guys are really good! One point, one spectrum, one waveform and you have the diagnosis.

Really difficult sometimes to analyze not knowing what it looks like and with only one data set of info and that being opposite the drive end.

quote:

This is a heavy unit; I'd guess the housing weighs 1500 lbs. with a 4000 lb. rotor running 1217 rpm. The base has 4 bolts across either end, with the base plates sitting on rubber/cork pads. They are squeezed down so tight, that there is very little "flex" left in them. This unit does have a lateral resonance around 700 rpm.

So with what data there is:

weight, (heavy)

a lateral resonance, might mean it is "twisting" at 700 cpm and has no effect at its running speed

very little, to no flex in the pads {tight on its base}

hold down pattern, 4 bolts across each end, making it seem to be extra wide and not very likely to move horizontal very much on a base that has very little "flex" in the pads which might mean it is not loose on its base and might mean it is not going to "rock" much in the horizontal direction being so wide and tight and heavy,which might mean something else is too weak to hold still

an ampltude of 8 mils in the horizontal direction with harmonics, but the end is mounted quite tight with 4 bolts, so is it loose at its main pedestal base, if it has one, or is it sitting flat on a small base on the ground level

The waveform is in velocity, so there must not be much impacting in Gs in an acceleration Waveform to give an indication of internal looseness, unless the discharge pressure is so high that the shaft is being restricted in its movement to give an impact

The spectrum is autoscaled with a vertical scale of .7. Making any raised floor appear low, but does show what appears to be a 1/2x peak as well as possible low amplitude belt frequency sidebands on some of the harmonics

a max amplitude of .5 in/sec (8 mils) at 1x and 5 harmonics, 2 of which are half as high as the 1x and 3 lower. sort of points toward some sort of problem with balance and looseness since the data is from the non drive end and must be the highest area of vibration, but then again, if any misalignment happened to be present, it may be shaking the weakest point, which might be the non drive end.

drive end data not available, which lowers the misalignment probability at the sheave since the possible highest amplitude is on the non drive end, but raises the question, is it running at a resonant frequency in the horizontal at the non drive end or is the lateral resonance creeping back into the picture

UNRELATED QUESTION, I guess,
Where did the theory of the "W" pattern in the waveform being misalignment come from originally? I have heard it forever but does it really say this? All the "W" represents is the fact that there are going to be several running speed harmonics in the spectrum, doesn't it? Any one have any hard evidence that a "W" means alignment and nothing else?

That's a lot of responses.

My take on the W is a combination of alignment/looseness. That's just my misguided opinion.

Rusty,

What is the sample size for the waveform (1024, 2048, etc.)?

John J

Hi, The waveform looks like pure sinewaves added to each other so I played around in excel and added the freq and amplitudes in the freq spectrum and was able to generate a waveform very similar to that shown by adjusting phase angles see attachment. So looseness is unlikely unless reading is taken away from the source of looseness and HF impacting is damped out. Like others say need more data. Good test though!

Copy_of_waveform_0.xls (278 Kb, 29 downloads)

No tricks. This is not some sort of test. If I knew the answer, I wouldn't have asked. I think it is clearance in the bearing housings, but I could be wrong. Just wondered what you see in the data, as I may be missing something.

Thanks for all the good input. Ralph you win the prize for doing a thorough, thoughtful analysis based on the information given; I appreciate your effort.

And I balanced this beast once and removed all the 1x. The waveform looked the same, just without the 1x. Balancing with Looseness Present

quote:

Originally posted by rustythevibeguy:
I think it is clearance in the bearing housings, but I could be wrong.

IMO, it was a text book pattern consistant with looseness, possibly amplified by imbalance.

What is interesting though in general is what particular information in the data can lead toward location of the a looseness and point to a bearing fit in the housing, on the shaft, or structural. In this particular case structural looseness has had a good chance since a defective isolator may loose stiffness completely thus creating looseness.

Regarding balancing in presence of looseness... Why not as long as phase is stable?

David