Beating Frequencies 1

[rasmumm]

I have a severe "beating" phenomenon that is affecting the overall 1X vibration levels on a 1775 rpm motor that is in close proximity of a rather large generator. I am quite sure it is "beating" because when I adjust the span and # of lines on my meter I can differentiate between the running speed of the motor and the 1800 rpm peak. My questions are: 1) How detrimental is this to the health of the bearings (they are journal type), and 2) Has anyone had any success dampening out the effects of the beating frequency (the levels are elevated on the motor in all radial orientations) Thanks

 

[electricpete]

Beating as you know occurs when there are two separate vibration frequencies which are close to each other.

I don't think the vibration seen by the machine is any worse than it would have been if the two frequencies were far from each other. The only difference is that if the frequencies are close we can more easily detect it by listening/feeling and more likely to see the oscillation on our vib instrument (higher frequency oscillation from wider-spaced non-syncrounously-related frequencies gets filtered out).

One way to confirm the influence of the generator is to take a measurement while the 1775rpm machine is shut down and check if 1800 rpm is still there.

The most destructive aspect may be the fact that your 1775 is subject to vibration while shut down, which can result in false brinnelling damage of the bearings.

 

[GregLocock]

OK, I am on very dodgy ground here since I have never measured bearing vibration in my life, but from a fatigue point of view I can see a possible issue with that argument, although given the numbers in this case I think it is fine in practice.

If the two frequency components are of comparable amplitude then the time signal will rise to a peak of twice the individual value in each cycle of beating, in this case at 25 rpm.

This will DOUBLE the maximum strain seen in the structure. Fatigue accumulates roughly as number of cycles*strain^6 or so in steel, in some regimes (from memory in MIL HDBK 5), so each cycle of beat will do 64 times as much damage as the individual frequencies would have.

So, depending on the frequencies involved it may be possible to see a problem with fatigue that would not have existed if they had not been beating.

In this case I am sure that the relatively small number of cycles more than compensates for the additional amplitude. This only happens at 25 rpm so the damage in one minute is 25*64 or 1600. The damage from the individual frequencies would be 1775+1800, or 3575.

The other reason I am on dodgy ground is that in reality all any pair of sine waves 'beat' to some extent (that is at certain times they will constructively interfere, at others destructively), it is a subjective call when you have beating.

On a related note, what is happening to the vibration power (proportional to amplitude squared) in this system? Very odd.




Cheers

Greg Locock

 

[electricpete]

Greg - I think we said the exact same thing.

I said: "I don't think the vibration seen by the machine is any worse than it would have been if the two frequencies were far from each other."

In other words, the only thing that makes beating different from other combinations of frequencies is that the two interfering frequencies are close to each other giving low-frequency interference which is easily detectable by sound, feel, and meter response. The machine doesn't care about that low frequency aspect, just humans do (imho).

 

[unclesyd]

rasmumm sated he has journal bearings which can be affected by low frequency vibrations. We had similar problem of beating frequencies on two very large pusher centrifuges. These were new machines and we lost both within a month due to journal bearing failures. The passing point on the frequency spectrum amounted to a second critical speed on the slower machine. If one machine was running as full speed and the other machine started up there was point that the vibration frequencies started beating. The results were dramatic as the slower machine started shaking and took the building with it. If tighten up the high vibration shutoff switch the machines would never reach speed if the other machine was at full speed. The machines were on the fifth floor of a steel framed building and the effect was dramatic, especially if you were on the fifth floor.
The bearings on the first machine had large number of longitudinal grooves as if the shaft trying to run in an orbit. The shafts were 6" with .0055 gap in the bearings (2). The solution was to put some extra logic in the control loops to help on timing of the machines. A change was also made in the bearings in that the oil groove was changed and the clearance was changed to .0045" and the oil pressure to the bearings was raised. There also some changes in the speed of one machine.

We also have a large fan room with probably (30) 250-300Hp motors driving belt driven Buffalo Fans. All motors are on isolated pads with none of the concrete pedestals the same size. None of the fans operate at the same speed. There also a difference in the size of some of the blowers, number and size of the blades. I was told this was done to stop the frequencies beating togather since this machinery is all on the same floor. Everything is this system has journal bearings.

 

[electricpete]

I guess I stand corrected. Thx Unlesyd.

If I try to understand what happened I conclude that the external vibration influenced the motion of the shaft within the bearing.

The first question that comes to my mind: Is this any more likely if the external vibration is at a frequency close to running speed than if some other frequency far from running speed? It seems plausible that if the shaft is orbiting at 1x, then some external vibration close to 1x is more efficient at transferring energy over a period of several cycles compared to vibration far away from 1x. (I vaguely remember some result from linear systems that energy transfers more easily among modes with similar resonant frequencies).

Would be interested to hear other comments on that.

We have several machines that experience beating vibration and I can't say I have ever heard of that scenario. One thing I have witnessed personally is false brinneling damage on races of rolling element bearings. Vibration on bearing housing was ~ 0.17ips with the machine shut down.

 

[unclesyd]

electricpete.
I'll see if they will give me a little more information on the vibration frequencies where this occurred. The centrifuges are quite large and are inherently unstable during operation due to varying loads and speeds. The two machines that were involved were identical in that the rotating units were interchangeable. Each component of the rotating assembly is dynamically balanced (Schenck) each time it is assembled after repair. I don't have the numbers they balance too but it quite low. The center of rotation is constant though the center of mass changes due to product loading and the movement of the pusher plate to clear the product. The pusher plate moves while the machine is coming up to speed. The normal vibration spectrum, while operating, on these machines is just a little less than chaotic.

Our plant is on sandy soil with layers of hardpan (clay) we have a lot of beating vibration throughout the site. We have seen this on many a non-rotating bearing. I have seen a few bearings with discernible false brinneling. They were probably more but were run and turned into a basket case. These bearings were mainly in the Fan Room I mentioned in the previous post.
The rotating equipment people call this type vibration "sympathetic vibration”.

 

[bentlytim]

I do appreciate everyone's responses to rasmumm's original statement and would like to add my two cents. My comments below are making the following assumptions:
- The motor is mounted solidly to the same foundation as the generator.
- The beat significantly changes in amplitude over time. If you stand at the machine and monitor it for several minutes, you will see the amplitude rise and fall, not just stay at a high level continuously.
- The beat has always been an issue and not something that has recently occurred. If this equipment has been operating this way for several years without motor bearing failures, it may be lived with. If this problem has recently developed you need to determine what has changed before making any 'corrections' that may not actually correct the real problem.

First, you stated that you considered this a 'severe' issue. By severe I assume that you mean the amplitudes reach what you consider unacceptable levels. Are the two separate (1775 and 1800) frequencies at what you would consider acceptable levels, or could they be improved (i.e. balancing). I suggest this first as a possible way to reduce the 'driving' frequency amplitudes.

Second, the way I look at a beat frequency is that the two independent driving frequencies are close enough in speed that they come in and out of 'phase' with each other. When they are in phase the amplitudes can easily be high enough that real damage is done in that time frame.

Third, if balancing is not feasible or cannot reduce the driving frequency amplitudes low enough, can the motor be isolated from the generator pedestal? Installing vibration isolators on the motor may reduce the influence of the generator.

If the motor is already on vibration isolators, they may be worn out and need replaced.

 

[electricpete]

I think those were some good practical comments, Tim.

I have minor disagreement with your 2nd item. It is along the lines of the original question and my original response. Just because there is noticeable beating doesn't mean it reaches any higher amplitude than when you have several frequencies of comparable amplitudes which are spaced far apart.

We have generally two separate indicators of machine severity. One is the overall value of the vibration. Another is the peak time waveform value.

If your overall value is going up and down due to beating then you should take the average value (not the peak) and compare it to criteria for overall vibration to determine severity. If you are concerned about the peak you should look at the twf peak and compare it to twf peak criteria to determine severity. In my mind the peak value seen on your meter during beating has no significance.

But as mentioned, the beating itself is also a symptom and deserves to be investigated.

BTW - is the turbine vibration seen on the motor very directional (much higher in horizontal than vertical). In such case resonance may also be a factor. You haven't told us anything to indicate resonance, but in our plant the machines that we noticed turbine-frequency vibration on are the machines that have resonance in that neighborhood.

 

[electricpete]

I should clarify the resonance referred to in my last sentence were structural/support resonances. We bumped the machine bearings and saw ~1800cpm.

 

[electricpete]

In our case the running speed of the machine was not close to turbine speed. Your case a little different. Resonance near running speed would be a problem in itself.

 

[bentlytim]

electricpete
Thanks for the additional input. I like the reminder to determine if there is any directionality to the condition.

You are correct in ensuring that everyone understands the beat itself is from two different symptoms. The beat itself could be a third symptom of a deteriorating or changing condition (i.e. resonance from changing structural stiffness).

However, I probably was not as clear as I needed to be for item 2. I do defend that the increase in beat amplitude is a real vibration as the signals go in and out of phase. Although depending on your equipments 'OA' calculation design, the overall number may not be a good indicator of the severity condition. Using the average of the max and min amplitudes to determine severity is a reasonable approach to determine what energy is being input into the system by the condition over time. However, I would try to use the average from the time waveform and not the spectrum to ensure accuracy.

You are correct that the original symptom signals do not change. But the combined signal (i.e. beat frequency) amplitude does change as the symptoms go in and out of phase. If you look at a time waveform signal with two components that are relatively close in frequency (such as this case with only 25 cpm difference), they are going in and out of phase, you will see the combined waveform as one curve and the one curve will increase and decrease in amplitude instead of a waveform that is complicated and more consistent with its peak amplitudes.

Yes, we do need to ensure that we are consistent with the calculations regarding the original time-waveform signal. You must be sure that we are looking consistently at true peak, rms, or other method as consistently as possible.

Thanks again and hope this makes things clearer rather than just adding to my typical confusion.

 

[GregLocock]

I agree, the actual instantaneous strain does increase, so the system is being stressed more highly.

I find this completely confusing, we are putting the same energy in, yet according to that estimate I did in my earlier post we are actually doing less damage? This is odd, to say the least.

Cheers

Greg Locock