Synchronous Motor Vibration - Approx 120 Hz w/ 1X sidebands

[MER21]

Hello,

I've got what appears to be frame-resonance related vibration on several synchronous motors. I would like to understand the source of the vibration, what further testing should be conducted, and whether it is detrimental to the motors.

We have 5,000-5,500 HP E-M synchronous motors, 327 rpm, driving D-R HHE recip compressors. The motors have 144 stator coils and 22 motor coils. The rotors have 22 fan blades on each end. Motors installed pre-2005 are rated at 6,600V. These have robust frames with 280-350 Hz natural freq. Motors installed post-2005 are rated at 4,160V. These have a different frame design which is less robust (lacks several stiffening plates found in the 6,600V motors), with nat freq at/near 120 Hz in several locations. Our 4,160V (post-2005) motors have high vibration at approx 120 Hz (0.62 in/s pk overall, 0.56 in/s pk at 120 Hz). The problem appears to be resonance-related. Compressor vibrations are normal, and similar between all motors. The attachment shows data for a "normal" (6,600V motor), and a "high vibration" (4,160V) motor. We confirmed the absence of other electrical (power supply quality) and mechanical issues (i.e. - air gap alignment, softfoot, etc.).

1. Regardless of resolution, I am unable to obtain distinct peaks separating the 119.9 Hz (RBPF / BPF) from 120 Hz (2 LF). Is the resonance masking the distinct peaks?
2. What is exciting the resonance? In all motors, regardless of whether the vibration is normal or high, there is dominant frequency at approx 120Hz, with 1X sidebands. All time waveforms show 22 distinct peaks over the 1X period. It seems to me the vibration must be related to the 22 motor coils (RBPF) or the 22 fan blades (BPF). Is it possible for an electromotive force or blade pass frequency to be amplified this much? What sort of electromotive force is being generated by the motor coils? We noticed no peaks at the stator coil frequency (144 X 327/60 = 784.8 Hz).
3. What kind of damage could result from this high vibration? At 120 Hz, we should have had fatigue damage by now. Our electrical PMs uncovered no electrical deficiencies (shorts, thermal damage, etc.). Are there other tests we should consider?

Thank you for any insight you can provide.

 

[electricpete]

docx does not compute for me. Got any doc?

=====================================
(2B)+(2B)' ?

 

[electricpete]

1. Regardless of resolution, I am unable to obtain distinct peaks separating the 119.9 Hz (RBPF / BPF) from 120 Hz (2 LF). Is the resonance masking the distinct peaks?

That is expected. On a 22-pole sync motor, 22 times running speed is same frequency as twice line frequency.

=====================================
(2B)+(2B)' ?

 

[electricpete]

Does the vibration pattern remain when machine is uncoupled?

=====================================
(2B)+(2B)' ?

 

[waross]

Hi Pete. .docx is an open office format.

http://www.openoffice.org/

to download.
Open source, parallels MS office.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[ScottyUK]

.docx is the latest MS Office format.

Here's an Office 97 - 2003 compatible version.

----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[MER21]

electricpete, I see that ScottyUK has posted an MSWord 2003 version of the graphs. The first page shows spectrum and time waveforms for a 6600V motor (unit C25) with acceptable vibration levels. The second page shows identical graphs (0-1000 Hz, 0-0.341 s) for a 4160V motor (unit C30). The third page shows a higher resolution spectrum for the C30 unit.

I'm just curious as to the source of the vibration. I see a lot of information on induction motor vibration, with various spectrum signatures - then the literature jumps over to synchronous motor vibration and discusses only stator coil pass frequency. Does the discussion on induction motor vibration apply to synchronous motors?

Is it possible to differentiate between 22X (119.9 Hz) and 2LF (120 Hz) spectrum peaks. With induction motors, the literature always shows different peaks at 2X run speed and 2LF. Are 119.9 and 120 Hz too close to differentiate, or does the resonant energy blur the distinction? I've read that line frequency doesn't generate sidebands. Since I'm seeing 1X sidebands in my spectrum and 22 peaks in my time waveform, it seems like the resonance is being excited by an "electromotive" of the 22 motor coils, or by the BPF of the 22 fan blades. I'm thinking either of these forces is normal, but the vibration is being amplified at resonance. Are the 22 motor coils generating a force in the normal operation of the synchronous motor?

 

[electricpete]

Is it possible to differentiate between 22X (119.9 Hz) and 2LF (120 Hz) spectrum peaks.

For a 22-pole induction motor - yes (when under load). For a 22-pole sync motor - no... there is no slip... 22 x running speed is the same thing as twice line frequency.
SyncSpeed = 2*LF / poles.

For 22-pole motor motor
SyncSpeed = 2*LF / 22

For 22-pole sync motor
22*RunningSpeed = 22*SyncSpeed = 22*2*LF/22 = 2*LF

=====================================
(2B)+(2B)' ?

 

[electricpete]

By the way, thanks Bill and Scotty for your help.

=====================================
(2B)+(2B)' ?

 

[electricpete]

I don't know anything that fits that pattern exactly. Might be simple 2*LF pattern plus a lot of harmonics which are generally expected when driving recip load. If it is really 1x sidebands around 2*LF, then we might think something like rotor moving eccentrically within the airgap at 1x. Actually I have seen something like that, but not exactly... let me try to dig it up Monday.

Do you know if vib was present when machine is uncoupled?
It would be interesting to know: does it change with change in excitation or load?
Is the vibration directional (for example horizontal much greater than horizontal)
For the velocity time waveform that you showed: can you post a zoom-in closer? I would be interested to know if the waveform is really going up the way the plot implies, or maybe it is just a sampling anomaly. We could judge that by a zoom-in that shows how fine the samples are.... whether the waveform looks rocky or not. If it goes with the spectrum, I guess sample rate was more than 2000hz?
Have you looked at the current waveform? It would be interesting to see if there is 1x sideband around 1*LF.
Also if you have strobe handy, it would be interesting to take a look to see if anything unusual is obvious.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Clarification in bold:

Might be simple 2*LF pattern plus a lot of harmonics which are generally expected when driving recip load.

should've been

Might be simple 2*LF pattern plus a lot of running speed harmonics which are generally expected when driving recip load.

=====================================
(2B)+(2B)' ?

 

[electricpete]

with nat freq at/near 120 Hz in several locations.

Can you explain more about that?.... where was 120hz natural frequency seen?

These are horizontal sleeve bearing motors?

Have you already tried loosening 1 foot at a time to see effect on vibration?
(I tend to think it is more relevant for small machines than large, but for 2-pole with 2*LF, we have to ask).




=====================================
(2B)+(2B)' ?

 

[MER21]

electricpete, Bill, Scotty, thanks for your help and responses so far.

Each motor has a pedestal-mounted sleeve bearing on the outboard end. The inboard end has no bearing - the rotor is directly coupled to the compressor. Therefore, we cannot run the motors uncoupled.

The vibration does not change significantly with load.

The vibration is not directional, but varies with location around the motor frame.

We checked for softfoot and made air gap measurements when the unit was down. We have not tried loosening one hold-down bolt at a time, but I suppose we could.

We bump tested several locations throughout the frames (external, walls, internal plates, stiffeners, etc.) of a 6,600V motor and a 4,160V motor. Several areas of the flat plates (inboard and outboard covers) in the 4,160V motors had natural frequencies within +/- 5% of 120 Hz. You can see the flat plates in the attachment. the 6,600V motors have 3 plates (including center plate) braced with stiffeners. The 4,160V motors have two plates (no center plate).

I've attached close-ups of the time waveform, as well as pictures of the 6,600V and 4,160V motor internals.

I think this 120 Hz vibration is normal, and is being amplified in the 4,160V motors. I'm just curious as to what's forcing it. Thanks for explaining the 22X = 2LF, that makes sense.

 

[electricpete]

Thanks for the info.

The characteristic of your time waveform I was wondering about is that it wandered up and down (peak does not come to the same height every cycle). Now that you have zoomed in to show how coarse the sampling is (maybe 3.5 samples per cycle of 7200), I think that feature is just a result of the sampling process... I’ll bet that if a higher sample rate was used, it would be a relatively clean sinusoid coming to the same peak each cycle of 2*LF. Do you think? (To me there is some difference... pure sinusoid matches resonance with sinusoidal excitation while sinusoid which wandered higher and lower might have suggested some kind of impulsive excitation like torque from recip).

Why does the first spectrum of the first attachment show a 7200cpm magnitude of around 0.1 ips, while the second spectrum shows 0.57 ips.

What location are these readings taken from?
If taken from same location, what changed?
What are the horizontal and vertical magnitudes at the bearing housing?
By chance are there any prox readings taken?

Just trying to get the full picture.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Why does the first spectrum of the first attachment show a 7200cpm magnitude of around 0.1 ips, while the second spectrum shows 0.57 ips.

The attachment shows data for a "normal" (6,600V motor), and a "high vibration" (4,160V) motor.

That explains it. Still interested to know where it was measured etc.

=====================================
(2B)+(2B)' ?