I have a very puzzling issue with a 6-pole motor driving a center hung I.D. fan. First, an overview of some of the relevant frequencies that could (or maybe should) be generated by this motor while in operation.

Motor Info:
Mfg.- Teco Westinghouse
Horsepower- 1000
Synchronous Speed- 1200 RPM
Actual Operating Speed- 1193 RPM
Line Frequency- 60 Hertz
Slip Frequency- 7 CPM
Pole Pass Frequency- 42 CPM
# Rotor Bars- 56 (56 Rotor Bars X 1193 RPM = 66,808 CPM)
# Stator Slots- 72 (72 Stator Slots X 1193 RPM = 85,896 CPM)
ODE Bearing- 6320 (Ball Bearing)
DE Bearing- NU226 (Roller Bearing)

Fan Info:
Direct Coupled Center Hung Fan w/ 16 Vanes
Inboard Fan Bearing- Sleeve/Babbitt
Outboard Fan Bearing- Spherical Roller

Over the past several years we’ve seen a frequency at ‘67X RPM’ dominate the velocity, acceleration, and spike energy spectra. The amplitude of this frequency has always been pretty significant ranging from 5-25 g’s of acceleration and varying widely dependant on load. I’ve always assumed that this frequency was ‘Rotor Bar Passing Frequency’ until yesterday morning when a true count of the rotor bars was made and we found there were only 56 bars.

I've played with all the electrical and mechanical frequencies in hopes of determining why the frequency at ‘67X RPM’ dominates the spectra with such great amplitude and I have several theories. I believe it to be a resonance, but the question that I can’t answer is... What is exciting the resonance and why does the direction and amplitude vary so greatly dependant on load?

Some of the things I've been 'kicking around' are as follows, but again, these are only thoughts;
· The dominant peak at '67X RPM' is a resonant frequency excited by a '1X RPM Sideband' of the 'Stator Slot Frequency'.
· The dominant peak at '67X RPM' is a resonant frequency excited by '18X the Ball Spin Frequency' of the Drive End Motor Bearing. (Perhaps the 'Ball Spin Frequency' is excited due to insufficient loading of the roller bearing!)

Any ideas as to what this frequency could be or what could be causing it?

6_Fiber_Drier_ID_Fan_Motor.doc (260 Kb, 41 downloads)

Michael,

My guess is that vibration at (or close) 67xSS is a 120 Hz (7200 cpm) sideband of the stator slot frequency that is the exciting a resonant structure (stator or motor case). The 120 Hz is caused by magnetostriction force and usually modulates both the rotor bar and stator slot frequency amplitudes in all induction motors. Motor load can affect both magnetic force and rotor positin in air gap. This can be confirmed by measurement with a 1600 or 3200 line acceleration spectrum. Use log amplitude scale and sideband cursors to verify which family of sidebands fit the center frequency (rotor bar or stator slot). I would also follow vibration amplitude changes with load at 1xSS and at 120 Hz. I would feel around motor housing and measure vibrations with a light weight accelerometer or microphone to possibly find resonant structure with motor running. I would use impact test method with motor off to identify possible resonant motor case. If the resonant structure is not external on case, then it is most likely internal. If motor is open frame, then there may be fouling in the air gap that could be cleaned.

The attached file contains bump test data taken on a brace on the motor stator accompanied with a photo of where the bump test was physically taken.

So... I think I've confirmed this as a resonance excited by something, but I'm not sure what1(???)

I'll agree that my resolution is not great, but... There is a small peak at '66X RPM' (or sideband of 67X) that lines up pretty close to a 120 Hertz sideband (to the left) of 'Slot Pass'.

I guess my preliminary recommendation to the folks in the motor shop will be to add additional bracing to the stator braces illustrated in the picture.
Do you think it's really that simple!?!?!

I guess the lesson I've learned is that you can never definitively identify a 'RBPF' until you've physically counted the bars...

Thank you Walt for inducing thought!!!

6_Fiber_Drier_ID_Fan_Motor_Bump_Test_Data.doc (1,759 Kb, 35 downloads)

Michel,

I don't know how you did the "Bump" test, but the data is suspect. Natural frequencies don't have harmonic frequencies. The frequency of interest is 1667 Hz, so I would use an analyzer F-max of about 4000 Hz with acceleration units (not velocity). Be sure to average about 4 spectra. The hammer tip should be hard plastic or metal.
The 1st vibration mode of those stiffening tubes is probably below 1667 Hz, so be carefull where you hit and measure response to avoid nodes. I would test other components including the stator core for a "ring" vibration mode.
You can contact me at w_f_strong at msn.com for additional support.

Walt,

It seems as though I've made foolish mistake!

I failed to accurately follow all the steps in the process of my 'New and Improved' Bump Test!

I'm going to experiment more with it in the office before I apply it to this machine.

Thank you for bringing this to my attention.

Some are talking about sidebands exciting a resonance. It is my understanding that sidebands are not true frequencies. They are created by the FFT algorithm. For example, if 7200 cpm sidebands are present, a true 7200 cpm frequency (2X line frequency) is modulating the signal and the FFT process creates the sidebands in the FFT, which are not true frequencies. A natural frequency cannot be excited unless energy at or near that frequency is present. The problem could actually be an excited natural frequency (resonance) and the energy for that excitation is one of the true frocing frequencies of the machine which is close enough to the natural frequency to excite it. This could be a housing or endbell resonance, but I would be concerned that a winding end turn is vibrating and will eventually wear through the insulation.

Good Luck,
John J

Sidebands such as a 2*LF sideband around RBPF are true frequencies which can excite a resonance, even though in some cases you would have a tough time picking them out of the TWF.

For example if my signal is sin(w1*t)*sin(w2*t), it will have sdebands at w1+w2 and w1-w2 can excite resonance. You can see the w1 and w2 frequencies easily on TWF (assuming w1>>w2 or vice versa) but in the spectrum you would see those sidebands, whether you generated your spectrum with an FFT or an analog spectrum analyser.

The FFT when properly applied (windowing, antialiasing filters etc) does not create frequency peaks of any significant magnitude that weren't in the original signal.

There was a long discussion about this on the old board. It is still available courtesy of our own Oli (thanks again Oli):
http://www.vtab.se/PHP-NBoard/html/images/materiali/Forum2/HTML/002780.html

As far as the 67x, that's another strange one. RBPF +/-k*2LF pattern does increasee with load. But hard to imagine why electrical/magnetic forces would create this frequency 67x.

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

Michel,

I am known to ask stupid questions, so here is one:
Did you also verify the # of stator slots?
I ask because if you had bad info about rotor bars ...

If the stator count is correct, I would suspect that
BSF is being multipled by the 18 rollers in your NU226.
Remember, actual bearing installation can change the
internal geometry of a bearing (heated fit vs. slip fit).
We're talking 3.68 versus 3.72 for BSF. (seems fairly close!)

Pete - whoa! bad flash back! that was some discussion!

OLI - thanks for keeping it all alive!

Take Care,

Some instabilities are more like a critical speed of forced vibration than like a true instability. Indeed, some can even be driver throught, a higher speed can be found where the rotors regains its stability, however the speed range of instability is wide.
Parametricaly excited are :
Assymetric shaft stiffness, assymetric rotor inertia, pulsating torque, rotor/stator rubs, excesive bearing clearance.
Effective solutions are :
Squeeze film bearing dampers, remove assymetries, isolate pulsating torque with torsionally soft coupling and align bearings to preven rubs.

Regards
Mike