high vibrations synchronous motor under reciprocating loads 2

[RRUUEE]

During operation we experienced high vibration on the pedestal at nde side on new installed 6MW Motor, single bearing type, running speed 450RPM.
Harmonic spectrum showed peeks at 7.5 and 22.5Hz that can be explained by bending exciatation 1st and 3rd Order.
The calculated bending frequency of the train is near to 3x order running speed
other peaks are around 120 and 240 Hz with sidebands
We're searching for explanations for the higher frequencies
These frequencies can be explaind by loose stator coils and other facts, but the motor is new
we found a previouss thread thread237-297373 that sounds similar, but outcome was not mentioned.
Could it be that synchronous motor reacts on reciprocating loads by high vibration frequency (120,240Hz) specially when strong bending results in assymetric air gap?
How to verify this?
Help appreciated
see also thread237-297373

http://files.engineering.com/getfil...67-4a65-99de-2ab2daeb05f4&file=resonancee.png

http://files.engineering.com/getfil...00496f9&file=hammer_test_after_stiffening.png

 

[waross]

"These frequencies can be explaind by loose stator coils and other facts, but the motor is new"
I have seen new equipment that was faulty.
I have found that the most difficult part of troubleshooting faulty new equipment is overcoming the chorus of voices proclaiming;
"BUT THE MOTOR IS NEW."
There may be nothing wrong with the coils, but don't neglect checking because "THE MOTOR IS NEW."


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

 

[electricpete]

Attached is a case study about a family of motors exhibiting 2*LF and low-level harmonics of 2*LF (actually only 4*LF…we didn't take spectrum any higher) which was clearly attributed to loose coils (based on waterfall of spectra showing change when motor was rewound or rewedged)

Slide 2 conveys the basic mechanism is that a fundamental force at 2*LF WHEN IN THE PRESENCE OF LOOSENESS creates harmonics of 2*LF (4*LF, 6*LF etc) in analogous manner than fundamental force at 1x due to machine unbalance etc creates harmonics of 1x in presense of looseness.

Slide 3 summarizes the history, also illustrated in slides 4-9. Coils were found loose on CD21 during post morten and 2*LF harmonic pattern had existed. Similar looseness seen on other motors from same OEM (slides 11-13). We suspected loose coils on sister motors which exhibited the same pattern. When those other motors were rewound the 4*LF disappeared as shown in waterfall diagrams. More significantly, when one of the motors (CD13 slide 10) was rewedged (not rewound), the 4*LF disappeared. There are fewere variables changing during rewedging…the only thing we changed was tightness.

Slide 14 - huge caveat
Caveats: * 2LF by itself (w/o harmonics) obviously does not signal looseness. * Harmonics of 2*LF can be caused by other causes: > magnetic force exciting looseness at other locations > magnetic saturation > 6*LF, 12*LF can come from electronic power supplies * MOST IMPORTATNLY


I have another case where running speed harmonics around 2*LF (on 22-pole vertical machine) were apparently caused by lower bearing loose in housing. I'll see if I can dig up those details.

I would approach any diagnosis carefully. Obviously you'd prefer to get some corroboration by other means before recommending drastic action based on symptoms which are inherently difficult to interpret.


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

 

[RRUUEE]

We will start a measurement campaign in the field to analyse more closely the origin of the higher frequencies around 120 and 240Hz
For us it's still not clear if the harmonics point more to an mechanical or an electrical issue (if such an differentiation is meaningful).
That's why we placed this issue here.

causes:
1. your reactions points to loose stators. Are the attached harmonics so typical?
2. can we exclude external sources or the electric net as possible orign at first?
3. can a bent shaft (not stiff enough) generate such an spectrum?
4. Actuall we have 2 motors with same issues. should we consider really a loose stator from the beginning of operation.
That points to a more fundamental design problem.
5. You mentioned magnetic saturation as possible source, how could this explain the vibration?
6. In an attachment of the mentioned thread237-297373: Synchronous Motor Vibration - Approx 120 Hz w/ 1X sidebands
it was mentioned that flat plates (inboard and outboard covers) are changed and affect the stiffness

measurement:
can you give advice for the measurement campaign?
We indentified some measurement points on the flat plats.
shall we tap test the rotor / Stator first?
how to measure dynamically excentic air gap?
what are the best locations for additonal measurement points

 

[electricpete]

My previous response was regarding your word description 120hz, 240hz.

Looking closer at your first attachment (spectrum while running), I don't see those particular frequencies standing out at all. You have a pattern of harmonics of 7.5hz (running speed) with highest harmonics at 3*7.5, 17*17.5, 29*17.5. (16 pole machine, 120hz would be 16*7.5, 240hz would be 32*17.5).

I'm inclined to think you are seeing harmonics of running speed which happen to peak near the resonance frequencies... not to far from resonance frequencies identified in your 2nd attachment (hammer test). I don't view it as electrical in origin. It may be the torque variations in the reciprocating load are exciting these harmonics (torque variation is coupled to radial vibrations in presence of asymmetric stator support which typically occurs in horizontal machines). What particular type of reciprocating load is this? Do you see these frequencies present on the driven machine?







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

 

[RRUUEE]

the recip load comes from a reciprocating compressor which typically generates a series
of frequencies acc. FFT


Mto= 120.155 kN Mt=Mt0+SUM[Mt(j)*COS[j*ABS(w)*t-phi(j)]]

ord. j Freq [Hz] Mt(j) [kNm] Mtj/Mto [%] phi(j) [deg]
1 7.5 33.727 28.07% 235.6
2 15 5.754 4.79% 62.5
3 22.5 13.745 11.44% 171.9
4 30 18.283 15.22% 287.2
5 37.5 2.478 2.06% 235.6
6 45 0.704 0.59% 340.0
7 52.5 1.053 0.88% 171.4
8 60 7.374 6.14% 276.4
9 67.5 1.629 1.36% 201.6
10 75 0.096 0.08% 60.8
11 82.5 1.065 0.89% 50.8
12 90 0.682 0.57% 161.4
13 97.5 0.575 0.48% 116.7
14 105 0.247 0.21% 178.5
15 112.5 0.540 0.45% 25.9
16 120 0.653 0.54% 125.9
17 127.5 0.441 0.37% 55.2
18 135 0.042 0.03% 261.0
19 142.5 0.257 0.21% 292.8
20 150 0.127 0.11% 45.9

also higher orders as the 17th are present.
but normally and also in this case higher orders (>12) are neglectable in height (Mtj/Mto < 5%).

 

[electricpete]

Interesting, I've always known that recips have a torque oscillation, but I haven't seen it quantified like this.
Out of curiosity, what is the source of these torque spectral values? Was it measured somehow? How/where

fwiw, I'm still inclined to think the vibration on the motor (at multiples of running speed) is a result of the torque variation of the load (at multiples of running speed). If you want a higher degree of confidence you can do some more investigation. Try uncoupled run of motor. Perhaps check for obvious mechanical looseness at bearings / housings / support structure since these can be alternate source of harmonics of running speed.

My attempt to respond to your individual questions:

1. your reactions points to loose stators. Are the attached harmonics so typical?
2. can we exclude external sources or the electric net as possible orign at first?
3. can a bent shaft (not stiff enough) generate such an spectrum?
4. Actuall we have 2 motors with same issues. should we consider really a loose stator from the beginning of operation.
That points to a more fundamental design problem.
5. You mentioned magnetic saturation as possible source, how could this explain the vibration?
6. In an attachment of the mentioned thread237-297373: Synchronous Motor Vibration - Approx 120 Hz w/ 1X sidebands: Synchronous Motor Vibration - Approx 120 Hz w/ 1X sidebands
it was mentioned that flat plates (inboard and outboard covers) are changed and affect the stiffness

1 - loose stator coils? I can't rule it out in theory but I think it's very low on the list. Loose stator coils (like any looseness) generate harmonics of an associated mechanical fundamental forcing frequency. In my case study linked above with steady torque, the forcing frequency acting on the coils was 2*LF (so we got 4*LF, 6*LF etc). In your case it could be coils responding to the varying component of the torque at 7.5hz. But there's nothing in the spectrum that narrows is to electrical cause and it could be many other things more likely/common than that. In particular simply coupling of the torque variation to radial movement via the machines asymmetric stiffness.
2 - external sources from grid. Very unlikely in my opinion… why would fundamental frequency 7.5hz be associated with the grid rather than the machine that operates at 7.5hz!
3 - bent shaft? Ordinarily would create 1x. Would only create harmonics in presence of some looseness or severe bearing problem. Seems unlikely to me although it never hurts to monitor what you can on the bearing (temperature, oil sample etc).
4 - Two machines with same behavior. You could interpret that as design problem as you suggest, or (my preference) you might interpret that as a sign that the response of the motor to this particular configuration (varying loading, mounting) is probably normal.
5 - saturation - saturation can increase CURRENT components on wye machine at frequencies of 5*LF, 7*LF, 11*LF, 13*LF. I can't say I've ever seen these show up in vibration or know exactly how they would influence vibration, but if they did see those frequencies (or double those frequencies), then I would consider saturation as a possible cause. I don't see anything like that in your vib spectrum.
6 - flat plates affecting response. Any flat plates attached affect the mechanical transfer function from forces to vibrations and in particular the resonances. You have already captured a bump test which is something like a transfer function in one of your attachments,

Again, bottom line it does not particularly alarm me as a symptom since the load torque is known to be rich in harmonics (and since torque is coupled to radial motion on horizontal machines with asymmetric radial stiffness).

The other question we might ask ourselves is whether these "known" load torque oscillations may be damaging to the motor. Certainly they may degrade coil tightness over time particularly if the winding was not well done to begin with. This might be something to factor into your long-term pm/inspection strategy, but nothing to worry about in diagnosing behavior that appeared immediately on new machines.



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

 

[RRUUEE]

for your curiosity: the torque oscillations were calculated by simulation program (not measured in the field yet).
The spectrum is more or less normal for a 4 crank machine.
Depending on the different masses and compression curves for each stage one can expect also 3rd and 5th order.

3 can we interpret looseness not only as loose e.g. bolts but also as weak support structure or weak shaft or weak guiding by not enough resistance in the oil film
4 indeed normal or should we say abnormal
7(new guessing) dynamically eccentric air gap due to high load and to weak shaft or resonance or operating near bending resonance.
8 Peaks only by coincidence near to frequencies 120 and 240

we will start measurement campaign soon
Thanks for your engagement

 

[CouplingGuru]

Tuning a reciprocating gas compression system can be quite tricky, there are many variables at play. You mention you recently installed new motor. Unless a new TVA was done with those motor characteristics your current resonance plot could be off. You may have a situation where you were operating between resonances with the old motor, but the new motor characteristics are causing the system to respond differently. In most cases, the TVA house will be able to determine if More or Less inertia is needed for the system as well as any coupling stiffness changes. I don't know your application specifically but in a broad sense those are the issues we face when tuning a motor ran reciprocating gas compression system.

When it comes to couplings we are always here to help.

http://WWW.PSCCOUPLINGS.COM

 

[RRUUEE]

thank for the Input. the mentioned points are right

But maybe you get me wrong, we havent replaced old motors, we're faced with the problem form the begin on.
new motor should tell only we can exclude wear phenoma.

 

[CouplingGuru]

Ok, I understand, I will assume that the TVA resonance plot is based off the current Motor, Coupling and Compressor combination.

There could be a couple resonance scenarios going on.

1.) If you see vibration resonance at speeds matching the TVA resonance plot, then you have a good analysis, and someone upstream made the determination that the vibration levels were acceptable OR the vibration levels are higher than anticipated, but still occurring at the point at which the TVA stated, in any case if a reduction in vibration is required, a new tuning of the system would be required to get the resonances to occur out of your operating speed range.

2.) If you see vibration resonance at speeds NOT matching the TVA resonance plot, then a few things could be going on. First, It could be a bad TVA. This is rare but humans can make mistakes. More common is some of the drive train components may not be matching stiffness's or inertia that are published. This is somewhat prevalent in couplings, especially if a very stiff coupling is being used, this is because the stiffness of rigid couplings can vary greatly through out their respective operating torque range.

In either scenario, Since this is a brand new install then I would tend to believe you have a resonance problem, that can only be corrected by re-tuning the system based off current measured operating resonances.
Granted I am not a motor expert, but I would tend to believe that a new motor of that magnitude would have been thoroughly inspected and tested prior to shipment, so I doubt that is the cause. But as Waross said ruling that out would be prudent.

When it comes to couplings we are always here to help.

http://WWW.PSCCOUPLINGS.COM

 

[RRUUEE]

was you said is true
we checked also for possible failures in model data or model of calculation.
inertias and masses are somewhat easy to check, the most unreliable part is the stiffness of the complex system.

what we can not explain the high peaks around 120 and 225Hz.
the sheer height of the signals there tends us to believe that the 3X cannot be the cause.

At the time beeing the system doesn't use a coupling.
But a relative stiff coupling might be a solution if there's something that can handle such high static and dynamic torques, Any ideas?

I consulted youre catalog types 3438 or 4048 are maybe in the range.
But I suppose these couplings cannot tolerate the bending stress by a single bearing Motor.

 

[CouplingGuru]

It is probably best to contact us directly and we can provide any insight we can. You are correct, we probably can't provide you a traditional disc coupling because of the single bearing motor, but may be able to find another rigidly coupled solution for you, like a torsion shaft. Since the current drive train is directly coupled that is the stiffest connection possible. Adding any coupling will most likely lower over all stiffness. Lowering the stiffness will decrease the frequency at which resonance occurs. Unfortunately there is a not a "one size fits all" solution. If your application has a wide operating speed range, it may be impossible to satisfy all conditions, in those cases our gas compression customers usually go with a torsionally soft couplng (rubber element) or an operating speed black out.

When it comes to couplings we are always here to help.

http://WWW.PSCCOUPLINGS.COM

 

[electricpete]

It seems the recent discussion is leading toward a conclusion that torsional resonance is the culprit.

The spectrum posted displayed shows many multiples of running speed, and there is not one frequency that significantly overwhelms the others.

To my thinking, this is not consistent with torsional resonance which would show an overwhelming response at one particular frequency (the resonant frequency) rather than significant components at many frequencies

To my thinking, this is consistent with harmonics of the 1X fundamental torque pulsation frequency, coupled to radial vibration by asymmetric stiffness (similar to what I see on practically any horizontal recip).

I'm skeptical of any conclusion that this is torsional resonance, but open to discussion if you think I'm missing something...


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

 

[RRUUEE]

Electricpete thank for your Input
we're still waiting for measured date form the field, before we can start with further Actions.

I'm lacking still a good and sound hypthesis able to explain our peaks in spectrumn.
So asymmetric stiffness looks like such an explanation term.

you mean you experienced asymmetric stiffness al lot?
But I hope that to a certain degree asymmetry shouldn't be problematic, right?
Otherwise we should heard more of such problems.

In our case indeed the motor shaft is relatively strongly reduced in diameter on one side.
Can you confirm that such asymmetry results to similar result as loose stators or soft foot do
(displaying of 2xLF) ?

 

[electricpete]

In the way I use the term, asymmetric stiffness refers to differing stiffness of the stationary components in the two radial directions.

It is not a problem. It is present in virtually all horizontal machines.

But it does provide a mechanism to couple dynamic torque to dynamic radial movement (a.k.a. vibration that you can measure with accelerometer). Imagine as an extreme case a motor on a very very tall thin stand. If you apply a torque to the stator, the base flexes and as a result the motor moves to one side. If you change the torque, the amount of flexing of the base changes and motor sideways movement changes.

Contrast that to vertical machine with ideal same stiffness in both radial directions. If you apply a torque, the supports may flex but there will be no radial movement. There are many ways to prove/envision that... one is simply a symmetry argument (which way should the rotor move in response to a torque, every direction looks the same).

Knowing that you have higher 1x-harmonic torque on a horizontal machine, it is not suprising to see high 1x-harmonic radial vibrations (due to the torsional-to-radial coupling created by asymmetric stiffness). If it were a vertical machine then you might have very high oscillating torques and not see anything on radial vibration (unless some other mechanism is present like for example gearbox).

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

 

[MikeHalloran]

I have to ask: Are there _any_ successful installations of this equipment at other locations, or is it all completely new?

Can you supply or vector us to photographs of the complete assembly?


Mike Halloran
Pembroke Pines, FL, USA

 

[RRUUEE]

The compressor type is recently installed in similar cases and run without problems, but with other motors. I can't tell if the now used motor type Has references under same operation conditions. I will check if we can share additional info

 

[RRUUEE]

Update:

measurement campaign finished
Report is created
no specially high excitations are found coming from compressor side as torsional or bending load with high frequencies
only raised noise floor detected
Problem identified as a resonating bearing housing
a fix will be tried out: implementation of an additional dampening element around the bearing housing

 

[electricpete]

Thanks for the feedback.

To my thinking if you have a resonance, you generally aren't going to see many different frequencies…. there would be one dominating all the others.

220hz is the highest peak in your spectrum. Is that your suspect resonant frequency?

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