Hydrogenerator Vibration & Noise

[StanBognatz]

I've got a unique situation with a recently re-stacked hydrogenerator. Vitals are:

# poles = 56 ; # stator slots = 360 ; line freq = 60 Hz ; rpm = 128.57

Old design had 4 stator core splits ; New design uses a continuous core stack

Vibration taken directly from the back of the core iron shows a 4.3 g-pk component at 720 Hz, with the next largest response at its first harmonic of 1440 Hz at 0.45 g. The adjacent keybar showed similar but slightly higher radial responses of 4.8 and 0.9 g-pk, respectively. Tangential responses on the keybar were 0.9 and 2.4 g, respectively.

Through each spectrum 120 Hz harmonics are present, but they are comparatively low level, typically 0.05 to 0.09 g.

The 720 Hz vibration creates significant audible noise, with levels that increase directly with load. At FSNL and with excitation on, we have about 75 dbA at 20' outside the generator enclosure. At 35MW load, we have 90+dB, with the predominant 720 Hz component.

Sister units to this problem unit operate at full load with noise levels in the 75 dBA range; no access for vibration measurements at this point.

Our client is not accepting the unit for operation based upon the noise being generated. We have had several theories regarding the source of the noise, with possible solution. But unfortunately, none of them has been very viable from an operational perspective.

I would appreciate any commentary on the issue.

Thanks,
Stan

 

[Strong]

Stan,
720 Hz is 6x 120 Hz. My guess is that there is looseness that causes the high harmonic, and there must be a resonant structure near 720 Hz. I use a pressure gradient microphone to scan all radiating surfaces and any air vents. You can try a conventional pressure mircophone, but it is not as directional. You could use a light weight accelerometer to measure where the vibration "feels" high. If you find and correct the resonant structure and the sound and vibrations become normal levels, then you are done, otherwise you may have to track down where the looseness (rattle) is occuring and correct that. An ultrasound meter may also be helpful to detect location of the "rattle" noise.

Walt

 

[StanBognatz]

Thanks Walt.

I've been suspicious of looseness based upon the 120 Hz harmonics, but the group seems to feel that with the brand new re-stacked core that things are 'tight'. One area that is in question is the fit between the keybars (vertically mounted between the stator frame and core iron around the periphery of the core), and whether there should be any clearance there or not. Do you have any experience there?

 

[Strong]

Stan,
I don't design generators, but based on designs that I have seen, I think all joints should be tight. Several years ago I worked on a project that had two new identical generators side-by-side. One had a loud noise, similar to your description, and the other did not. The manufacturer said all was OK. My recommendation to owner was to demand an extended warranty covering this issue.
Another project involved a large (450 MW) generator with very high vibration and sound at 120 Hz. The technical description (mostly spin control) was unbelieveable, because they were trying to avoid retrofit of a fix on several machines in service or on order. Good luck dealing with the Big Boys on this problem!

Walt

 

[electricpete]

Based on the fact that your next highest harmonic is at 2*720 hz, I would say the force causing this vibration is more likely at 720hz, rather than 120hz.

I will have to think awhile about possible electromagnetic sources.

If tighteness of the wedges were a concern (not in my opinion...I would expect to see 120hz + harmonics in that case... in fact have seen it before), then a tap test could be done if the machine were disassembled.

What's a keybar?



=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[StanBognatz]

Thanks electricpete. The 'keybar' is a piece of square steel tubing, ~ 2" x 2", than runs vertically along the outside of the back of the core iron. It is welded to the stator frame and, for lack of a better description, abuts the core iron. Relative tightness between the core and the keybars is current topic of discussion. There is no mechanical fastening to the keybars. All attachement of the core to the stator takes place at the stator core studs clamps - which have flat plates cantilevered onto the back of the core iron and are tightened via 2 nuts each. The studs themselves are welded to the stator frame.

 

[electricpete]

4.8 g’s at 720hz. The magnitude is around 0.4 ips.

300 slots sounds awfully big (how many horsepower and what voltage?). I don’t have experience with as big as machine as I think yours, but that sounds pretty high for long-term reliability of a big machine. Over time the vibration affects the coils in the slots, their semi-con coating (assuming they are above 7kv or so), the strand insulation (I don’t think those big guys have turn insulation), the core insulation etc.

You have 56 poles for each of three phases, so a total of 168 pole phase groups. To put 168 groups into 300 slots, you need a fractional slot winding (probably 132 groups of 2 plus 36 groups of 1. In the motor world, it’s not uncommon for fractional slot windings to have much much more electromagnetic noise than other motors. We have two families of large fractional slot winding motors at our plant and they are by far our noisiest motors at the plant.

I am pretty sure what you are seeing is electromagnetic related and likely inherent in the design of the machine.

Some info that may or may not help to pin it down:
1 – Does vib change with change in excitation.
2 – Does vib change with change in load.
3 – Is there a regular pattern of rise and fall of vib around periphery of the backiron. That may tell you the spatial harmonic of offending field and may provide a clue.
4 – Check for current unbalance in stator. That can increase electromagnetic vibration.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Again I don't think looseness by itself would cause this problem. Normally that would show up at harmonics of 120hz.

I would get the OEM involved in analysing the electromagnetic harmonic performance.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[StanBognatz]

360 slots, 56 poles hydro-generator, and yeah it's big, about 23.5 FEET in diameter.

There is indeed a fractional slot winding (6 3/7 I believe).

We feel it is likely electromagnetic as well, and inherent in the design. But we've got a high noise (95 dB) situation, and a customer unwilling to accept the unit. So, we are trying to determine what exactly is the mechanism that is generating the 720 Hz. That's where we're hung up.

The vibration and noise increase directly with load and field current, but not with excitation (as reported to me).

We are currently outlining a battery of tests for the next site visit, and current balances are in there, as is consideration of using an external DC excitation to see if ripple effects are contributing to the 720 Hz activity.

Any other thoughts from an electrical testing perspective?

Thanks for the help.

Stan

 

[electricpete]

I see some of my questions were already answered but I wasn't paying attention: 35MW. Increases with load.

The 4 pieces I assume were 4 90 degree quadrants? If so that seems like it could be due to:
A - ease of manufacture:
B - intentional part of the electromagnetic design.

If A, I would assume the 4 quadrants would be fitted together tightly as tightly as possible. If B I would assume maybe a small gap.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Finally, even if the electromagnetic design were unchanged, there may have been a core resonance created which now provides a large response to an electromagnetic force component that has always been present.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[StanBognatz]

Originally it was a 4 piece ('segmented') core design as you noted. The new design is a continous stack (no joints).

 

[electricpete]

Following up on the last point... even if traditional looseness is not the cause, you might be able to "tune" the stator by adjusting tightness of radial core support at those keybar locations

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Did there appear to be any intentional gap or insulation in the 4-piece segmented design?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

I would see the main troubleshooting effort would be to trial and error with varying types of adjustments of radial support to the core to see if you tune out the resonance:

-keybars wedged in tight
-keybars loose or removed.
-install another temporary support between keybars for extra stiffness

Also thinking some more about determining the spatial harmonic, assuming it's a rotating wave rather than a standing wave, you couldn't determine it from vibration magnitude, but you could determine it from phase measurements around the backiron.

If noise is the issue more than vibration, is it possible that some kind of treatment to absorb/damp the sound would help? (I suspect this might be a last-resort option if you can't limit it at the source).

There are a number of other checks you might consider if they're easily achievable, but don't have any apparent tie to the problem: airgap check, winding resistance test, pole drop test.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

I found this item that by the title appears it may be relevant (although a quick skim didn't tell me anything)

http://home.houston.rr.com/electricpete/HigherHarmonicsCalcOfSyncGenerators.pdf

Let me know after you download it so I can delete it and free my website space back up. Thx.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[StanBognatz]

I got the article, thanks. Didn't get too much out of it.

We are looking at re-tuning the core by linking mechanically to the frame, but IMHO it is way too trial and error. I doubt if anyone will be able to model and predict the effects.

The rewind contractor is standing firm that the vibration levels are acceptable, and I essentially agree except for the noise being generated. I'm personally leaning toward damping materials on the frame sections and sound absorbtive materials on the interior of the overall generator enclosure.

 

[tmoritz]

Did you make any changes to the rotor or are you using the original rotor? How old is the rotor? How many arms does the rotor structure have? Did you decrease the air gap? I'm thinking the rotor might be flexing or relaxing resulting in a sinusoidal variation in your air gap. That would cause a higher frequency vibration in your stator, wouldn't it? A decrease in air gap and/or increase in stator rigidity would accentuate the problem. If this is a plausible suggestion then you might want to use an air gap sensor to measure the clearance. Take a look at Bently-Nevada's web site or Vibrosystm for air gap sensors.

Regards,


Tom Moritz
Mechanical Engineer
US Bureau of Reclamation

 

[StanBognatz]

Thanks Tom. The original rotor is being used, made circa 1950 (not real sure on this). I do not know if air gap was changed, that is an excellent question and one I took for granted. The stator is definitely more rigid as we went from 4-segment to continous stack. There are not any air gap sensors installed, and I've already recommended that. Do you have any good contacts at USBR that might be able to chime in on this post??

Thanks,
Stan

 

[tmoritz]

Stan,

Can't help you out with a point of contact on this subject. I'm relatively new to Reclamation and most of the people I've met are mechanical engineers by trade.

Just a few more questions, you obviously used a FFT spectrum analyzer to measure the vibration spectrum. Can you tell me what frequency span and how many lines of resolution you used? The slot pass frequency for this unit is 771Hz, not to far off from the 720Hz you stated in the original posting. From what I've read (I've got limited knowledge in electromagnetics) slot passing excitation is a result of differences in reluctance of the stator iron and the winding. Doesn't seem like this would be an easy problem to fix.

Good luck,

Tom Moritz
Mechanical Engineer
US Bureau of Reclamation