All,

It is known that real life is not back-and-white. Same could be applied to the vibration world and to the impact test in particular.

Refer to the applied plot as an example. Multiple small peaks and one large ( sorry, no phase data) are present. They could represent some internal components resonant frequencies as well as background vibration.

One will definitely pay attention to the largest peak, but how about the small ones, in particular at 3555 cpm. ( This is rinning speed of a motor and vibration is high at 1x when running. Background vibration should have been taken in this case although nothing was running in vicinity but unfortunately was not ).

In general, how seriously one should take small peaks? It would have been preferable to have a quantitative approach.

David

Bump_test.doc (44 KB, 57 downloads)

Can you describe what you were impacting?
Are those side-band looking peaks equally spaced or not?

I lean more toward what the waveform shows me in the ringdown direction. I see your waveform has something that looks repetitive. But then that is just my thoughts.

Doesn't answer your question directly, but if the ring down references a small peak rather than the large one, might say something differently than what the large one shows.

Looks like from the spacing of the waveform out to 1/2 sec. there is close to 14 peaks from the impact to 1/2 sec, with a time difference of about.0223 sec which might be close to 44.8 Hz or 2700 cpm. All these numbers are from visualizing the distances, might be wrong on the frequency but you could confirm if the peaks are equal and what the frequency of them is.

The significance of the largest peak is pretty obvious I think, but the rest.... who knows? Without proper frequency response functions (2-channel) I think it's impossible to tell about the other peaks.

High 1x when running? First thing I always look for is loose mounting. Had a fan once with high (and odd) 1x vibration and they assured me that all the pedestal bolts were "tight" and they were, but the nuts were bottomed out on the threads of several bolts, so the base was still loose. Adding some flat washers tightened the base up, and allowed the fan to be balanced without further problems.

I'd balance it. The phase reading of the reference run, and then the final position of the balance weight usually tell you where you are relative to resonance.

This is a little fan with the impellor on extended motor shaft. Motor IB is impacted in vertical.

Atached file shows peak frequencies. If those are resonant frequencies I doubt any relationship between peaks should be present.

Doc4.doc (35 KB, 22 downloads)

You're right -there is no logical reason for resonant peaks to have equal spacing resemblin sidebands. Just thought I'd check.

My wild guesses fwiw

Reasonable simple interpretation would be those other peaks are just noise in the skirt of the main 2800cpm peak and so the relevant number is separation from the 2800cpm peak (20%).

The devil's advocate position - there are several possible modes of interest. At a minimum one where the whole machine moves up and down together and one where it pivots. Bumping at that location may be efficient at exciting the 2800cpm main peak resonance, and not as efficient at exciting the a 3500cpm mode that shows up smaller in the bump test results. But that doesn't mean that that other 3500cpm won't be excited by the force profile generated by machine during operation. If you saw that same peak of concern show up in several different bump tests (for example impacting at the opposite bearing or impacting in middle of stator), that might increase the concern for it.

As you said - bump test is not exact.

An important thing to note - the main 2800cpm resonant peak could be lower by disk effect of the overhung rotor during bump test and stiffened by gyroscopic action of the overhung rotor when running (especially if the fan rotor has large diameter/length ratio). Both of these factors suggest your 2800cpm bump test resonant peak could be higher frequency - closer to 3600rpm with the machine running.

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

-----------------------------------------------------------
An important thing to note - the main 2800cpm resonant peak could be lower by disk effect of the overhung rotor during bump test and stiffened by gyroscopic action of the overhung rotor when running ...
_________________________________________________

To me this is a controversial point in general.

The question is: can we assume, when performing a bump test by impacting a motor body, that it will produce a separate peak in the spectrum defined only by shaft stiffness and mass which of course ( the stiffness ) is not affected at that time by gyro action?

If we had impacted the shaft -- then may be.

David,

Sorry for asking again, but what are the ringdown spacings in the waveform? Are there any other peaks showing if you expand the waveform out a little? I trust the waveform more than the spectrum in a bump test, but again, only my opinion and I could be totally wrong.

I'm not sure exactly what your question is:

quote:

electricpete:An important thing to note - the main 2800cpm resonant peak could be lower by disk effect of the overhung rotor during bump test and stiffened by gyroscopic action of the overhung rotor when running ...
_________________________________________________

David_G:To me this is a controversial point in general.

I'm not sure which aspect is controversial, but I will paint a picture of a specific example when it would act like I discussed. Let's say machine has a mode is rocking axially with a pivot point roughly halfway between motor bearings. The overhung fan has large diameter and somewhat flimsy or long shaft. We test the machine in three conditions:
1 - Bump test inboard bearing while stationary- excites the mode of interest since we are not impacting on the node. Measure resonant frequency and call it F1.
2 - Running condition. Similar mode excited by a couple unbalance or overhung unbalance. Measure critical speed (let's say by variable speed test looking for max response) and call it F2.
3 - Mathematical computation of resonant frequency when the fan disk is replaced by a concentrated mass (same mass) located at shaft centerline. Call it F3.

The results of these three experiments will compare as follows:
F1 < F3 < F2
Fbump < Fpointmass < Frunning

quote:

The question is: can we assume, when performing a bump test by impacting a motor body, that it will produce a separate peak in the spectrum

Impacting at the location we discussed (inboard beairng) can excite at least two different modes: the rocking mode and the simple vertical mode. And it is reasonable with antifriction beairngs that the shaft extension and fan join in that rocking motion also (probably the harder the impact the more likely the motion will be transferred to rotor, very small motion may be accomodated by bearing clearnaces without moving the rotor... and the bearing stiffness at low force may be less than with higher force... can certainly build a lot more complicated model...).

quote:

defined only by shaft stiffness and mass which of course ( the stiffness ) is not affected at that time by gyro action?

To my way of thinking, the 'disk' effects resonant frequency during bump test (makes it resonant frequency lower than if the fan mass were concentrated at the shaft centerline) and during rotation (makes the forward whirl critical speed higher than if the mass were concetrated at the shaft centerline). These two effects act in opposite directions and therefore both increase the difference between bump test resonance and running resonance.

Sorry if I have misunderstood your point.

Hope others will chime in also.

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

quote:

Originally posted by Ralph Stewart:
David,

... what are the ringdown spacings in the waveform? Are there any other peaks showing if you expand the waveform out a little?

Ralph,
Sorry, here is the TWF

Doc2.doc (44 KB, 14 downloads)