Engine Driven Generator Vibration Amplitudes

[MachineryWatch]

I have been testing a V6 (4.2L) natural gas fuel, spark ignition, emergency generator running at 3600 RPM. The 1X RPM vibration amplitudes at all load conditions approach 1 inch per second on the generator bearing (spectrum peak). There is only 1 generator bearing on the outboard end of the generator.

The crest factor approaches 5 on this unit. This indicates that the energy in the waveform is not very close to a sine wave. This is common for internal combustion reciprocating engines. The amplitude observed in the waveform spikes is 5 g's (peak) which at 60 Hz is equivalent to 5 inches per second (peak) or about 250 mils (pk-pk)!!

These are prototype units and there has been some rotor/stator rubbing. The generator air gap is only 1 millimeter.

My data indicates that there may be some resonant amplification at the generator bearing. Amplitudes in the vertical direction are about 5X higher than in the horizontal direction, and there are some other indicators, as well.

(1)Are these extreme levels common on engine driven generators?
(2)Can the rubbing be expected with vibration levels like this and such a small air gap?
(3)Won't these high vibration levels significantly shorten the life of the generator bearing?
(4)Is the spiked nature of the waveform energy more or less damaging to the generator bearing and windings than a more sinusoidal waveform?

Skip Hartman

http://www.machinerywatch.com

 

[vanstoja]

I'm not a generator authority but I would expect to find electromagnetic forces in the air gap something like an induction motor's unbalanced magnetic pull(ump) which is an inertia-type, negative spring effect loading. That is, the more eccentric or the more bent the rotor becomes the higher the ump loading is with the only salvation from rotor-stator running contact being the bent stiffness of the rotor shaft assembly. If the air gap is undersized at 1mm (approximately 40 mils) then contact is that much more likely. You didn't say what your rotor or stator bore diameter is. My induction motor calibration says the air gap should not be substantially less than 15 mils per inch of rotor radius. If your 1X rotational waveform is severely distorted, I would expect to see higher harmonics of running speed in the spectrum. If they're not there, then all the energy is focused in the fundamental and 1X will be at its highest levels. Intermittent or continuous rubbing in the airgap could be disastrous, particularly the latter which may mean full annular rubbing known as dry friction whirl with precessional direction opposite to rotor rotational direction. In dry friction whirl the rotor is essentially trying to get outside the stator wall boundaries and you have a machining process ongoing at the rubbing contact area.

 

[electricpete]

Hi skip.

I think you were a factor of 10 off on your calculation.
At 3600cpm, 5g's => 5ips => ~ 25 thousandths of an inch (peak to peak).

Also, I'll point out that conversion applies for a sinusoidal waveform (not a spikey-impact waveform). The better way to determine the actual peak-peak discplacement would be a double integration of your velocity waveform. Since the time waveform spikes at 2600cpm are much narrower than a half-cycle of 3600 sinusoidal waveform, you will find the double-integration should give you a much smaller peak-to-peak result.

Also the relationship between rotor motion and bearing vibration deserves some consideration.

You're probably aware of all that, but I just wanted to throw in those comments.

 

[electricpete]

Two more cents - my limited experience with reciprocating machinery is that they all have hi crest-factor impacting-type of vibration on the engine. It's the nature of the beast. I have seen similar vibration magnitudes on our compressors, and we know they are ok because they have always run like that. Of course your machine may be different.

 

[FredGarvin]

Just to back EP up, a quick look at the handy-dandy slide rule shows just under 30 mils at 60 Hz. But again, like EP said, that is for sinusoidal motion only. Definitely not your case.

Just one thought, since you mentioned you've had evidence of rubbing, when was the last time you balanced the rotor? Is this a single phase generator?

EP....double integrate velocity to get displacement? Am I having a brain fart this early in the morning?

 

[MachineryWatch]

vanstoja - the bore diameter is about 12 inches on this unit so that puts our air gap significantly lower than your guidelines for induction motors. There is, of course, harmonic energy in the spectrum.

Pete - Thanks for pointing out my error on the displacement. I agree that the conversions are only accurate for a sine wave. Do you think the FFT more closely represents the amplitudes? Or is it somewhere in between what is displayed in the FFT (which seems artificially low in these cases) and what you get using the inaccurate (for this wave shape) waveform relationship conversion formula? I was using accelerometers so I have an acceleration waveform and I will try the Excel time domain integration we have talked about in the past and see what I come up with. I agree this type of waveform is common on recips. I just wonder if the amplitudes and impulsive nature of the energy are enough to damage bearings and reduce winding life significantly. The manufacturer also needs to meet various vibration specs that are out there, which it is currently exceeding.

Fred - This is a 3 phase, 80 kW generator prototype. The component manufacturers (engine and generator) each do their own balancing before the units are coupled. I am a big believer, in the fact that once these rotor components are joined you have a new rotor and the fact that independent component balancing was done may mean very little. No dynamic balancing has been done after assembly and I am recommending that the first step be balancing a couple of these units to determine what improvements can be made that way. On most rotating machinery you can get a pretty good idea from the phase relationships how successful balancing will be. I think that determination is harder to make in advance with the reciprocating engine. It is certainly worth a try before they start trying to redesign things.

Skip Hartman

http://www.machinerywatch.com

 

[electricpete]

Fred - you're right... I meant double integration of the acceleration.

Skip - Intuitively it seems like hi-crest factor waveforms must be tougher on almost any equipment than a corresponding sinusoidal waveform with the same overall velocity. One thing that comes to mind is fatigue... doesn't that depend roughly on the number of cycles times the peak amplitude? In that case you have the same number of cycles but much higher peak amplitude for the hi-crest-factor waveform. Likewise certaily plastic deformation will be higher with a higher peak stress.

So, it seems reasonable to me that the true peak would be a better indicator of the damaging potential of nonsinusoidal vibration than an rms value.

But that of course brings us to another philosophical argument... do we monitor vibration because the vibration itself is destructive of because it is a symptom of someting else destructive going on? Probably a little of both as you know.

As far as practical experience one way or another I don't have any. I do know that compressors have a more frequent overhaul schedule (3-10yrs) than pumps (10yrs to never), but not necessarily because the bearings are degrading.

 

[MachineryWatch]

Pete - You are right that the reason for monitoring is "a little of both". The purpose of my measurements on the genset really didn't have anything to do with monitoring, however. The vibration analysis was to help them determine the reasons for less than desired reliability of prototype units. They are really still in the design process for this unit. So my data acquisition and concerns during analysis were/are skewed to that perspective.

Skip Hartman

http://www.machinerywatch.com

 

[electricpete]

On the subject of why an impacting-type waveform (hi crest factor) can be more damaing than a sinusoial waveform of the same rms value:

(We already said the impacting waveform creates more plastic deformation and fatigue.)

Another aspect is that the impacting waveform has a spectrum which occupries many more harmonics and therefore has a much higher probability of exciting a resonance somewhere within the machine.

 

[GregLocock]

5 g rms 0-400 Hz is typical for an automotive engine block, but you seem to be implying that is is largely first order, that number is way too high. It sounds like a balance or alignment issue. Most of the energy for an I6 should be in the 1.5, 3, 4.5 order sequence at full throttle, and first order should be >>10 dB down on those peaks.

On an SI I4 engine you can see up to 20g at second order, due to the inertial problems with L/R ratio, so if you have an automotive generator it will probably be OK for a hundred hours, but I'm guessing that you have a shaft driven generator which may not be as tolerant.



Cheers

Greg Locock

 

[VibEngr]

Skip,

Anyway you look at it, the vibration levels that you measured on the generator bearing are too high. A few years ago, I did extensive vibration testing of several diesel engine generator sets. The single bearing generators had high vibrations for several reasons:

1. The overhung exciter was resonant at 4x running speed. Excitation from the engine that was coincident with this natural frequency caused high vibration. After 500 to 1000 hours of operation, the exciter would fail. You can easily check this with an impact test of the generator shaft with the unit down. We also used a strain gage telemetry system and proximity probes to take measurements during operation.

2. Depending on the mounting configuration, the end of the skid could be resonant causing high vibration at the generator bearing. We looked at rigid mounting and spring isolators. The best arrangement was rigid mounting to the skid and spring isolators between the skid and floor. The extra beam length was removed from the end of the skid to lower the vibration amplitudes.

3. High 1x vibration could be an indication of mechanical unbalance or an electrical problem. If the vibration occurs when the generator is loaded, and then immediately reduces in amplitude when then generator is de-energized, the problem is electrical. We found some loose connections in the generator. On the other hand, mechanical unbalanced will cause 1x vibration that varies with the speed squared.

To download other papers on vibration go to

http://www.engdyn.com/papers_main.htm

 

[MachineryWatch]

VibEngr - Thanks for your helpful comments. What were the speeds on the generators you were testing? This unit is running at 3600 RPM. The 1X RPM amplitudes do not change with load which leads me to believe that all the 60 HZ energy is from a mechanical source rather than electrical. We do not have an overhung exciter as you had. The unit has 2 isolator mounts under the radiator end of the engine and 2 isolator mounts under the outboard end of the generator. I think that a direct mount of the engine/gen assembly to a subframe and then isolators between the subframe and main frame is a reasonable start to get the vibrations down. This would allow for some stiffening of the total structure and add mass to the unit. It would also make it easier to add stiffeners if further structural rigidity is necessary. I did check the ends of the skid for resonance and the performance of the mounts There was no resonance and the mounts (isolators) seemed to be performing well. The problem with these units is that they are relatively small and the typical installation is to drop them on a slab with a forklift, or slide it off a trailer onto the slab. The frame rails simply sit flat on the slab with no attempt to actually secure it to the slab. This eliminates any ability to mount the frame itself on isolators. I am recommending that we make some attempts at dynamic balancing these units after assembly.

Greg-Thanks for your comments on typical engine vibration levels. You are right, this is a shaft driven generator with a single bearing on the outboard end. The highest vibration levels were measured on the generator bearing, not the engine. This is where my concern lies.

Skip Hartman

http://www.machinerywatch.com

 

[VibEngr]

Skip,

The engine/generator sets I was telling you about ran at 1800 RPM.

To confirm if your vibration is strictly due to unbalance or if a lateral critical speed may be involved, I would suggest performing an unloaded speed sweep. The amplitude of the 1x vibration should change with the speed squared.

If you find a peak or an increasing slope with a phase shift near the operating speed, you may have a critical speed problem. An impact test of the rotor with the unit down would help confirm this.

Regards,

Troy Feese

 

[MachineryWatch]

Troy - I'm hopeful that the manufacturer is working on a new electronic engine control module that will give us the ability to do a controled engine run-up and run-down. Impact testing was performed on the generator rotor when it was not assembled to the engine. Result was a response at 87 Hz. We may have to get some miniature accelerometers to do a more representative impact test with the unit assembled. Even with the miniature accelerometers it will be hard to place them where we can get a good response measurement. The only exposed place is the fan blades which, I believe will probably give us a distorted response. Thanks again.

Skip Hartman

http://www.machinerywatch.com