Previous in Forum: Control Valve and Positioner   Next in Forum: Material specification
Close
Close
Close
8 comments
Rate Comments: Nested
Anonymous Poster

Reading a spectrum graph for vibration analysis

05/09/2010 7:26 AM

Hi all,

I am confused with the amplitude at 1x, 2x, 3x frequency. The 1x, 2x and 3x is the harmonic (or mutiples) of the original frequency, meaning that if the shaft is turning at 1500rpm, then 2x will be the amplitude at 3000rpm and 3x will be the amplitude at 4500rpm right? But how will the amplitudes at 3000 and 4500rpm be measured as the shaft is only turning at 1500rpm?

And how does one interpret this kind of graph, for e.g.: why does an unbalanced shaft wil have high amplitude at 1x frequency, while misalignments will have high amplitude at both 1x, 2x and sometimes even at 3x frequency?

Thanks.

Reply
Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.

"Almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, vote them!
Guru

Join Date: Jul 2007
Posts: 4448
Good Answers: 143
#1

Re: Reading a spectrum graph for vibration analysis

05/09/2010 8:32 AM

You've asked a complicated question, and I suspect you need to look at some fundamental wave characteristic information.

The keys to understanding vibration in a shaft are the end conditions and the exciting energy.

If the ends of a shaft are well constrained, as in bearings, then all vibration modes will be those whose wave length permits nodes (zero amplitude for all purposes) at those end points. If and end is not constrained (perhaps an overhung shaft for a grinding wheel), then that end will be an antinode and only those vibrations whose wave length permits that will be present.

Note that you can always force vibrations which don't satisfy the above constraints, but that leads to machine destruction and is pretty obvious.

Generally speaking, if you have a shaft with two bearings, you'll find the vibration to be one-half the fundamental frequency (sub-harmonic) where you have he half wavelength between the bearings (λ/2). You can also have the fundamental frequency (λ) where there is a floating node in the middle. The fundamental is also called the first harmonic. Multiples of the first harmonic are the second (2x), the third (3x), etc.

You can also have 3λ/2, with two nodes in the middle; this is 1.5 the fundamental frequency f0.

You can have 2λ, 5λ/2, etc. The key to which are allowed depends on the end conditions. The key to which are actually present, and in what amplitude, depends on the forcing function and irregularities (unbalancing factors) in the shaft; that is, if there is a resonance at, say, 5200 RPM, and you are only turning the shaft at 1250 RPM, you will not excite that resonance.

There is also the matter of dynamic vs static balance, but that is more than can easily be answered her.

I suggest a good book on rotating machinery.

This is a very facile explanation. People spend a lifetime on this subject, and you'll need to do a bit of study.

__________________
"Well, I've wrestled with reality for 35 years, Doctor, and I'm happy to state I finally won out over it." Elwood P. Dowd
Reply Score 1 for Good Answer
Guru
Engineering Fields - Electrical Engineering - Been there, done that. Engineering Fields - Control Engineering - New Member

Join Date: Dec 2008
Location: Long Island NY
Posts: 15600
Good Answers: 981
#2
In reply to #1

Re: Reading a spectrum graph for vibration analysis

05/09/2010 9:40 AM

GA TVP45. The only thing to add is an explanation of what the dynamic signal analyzer in producing a Fast Fourier Transform (FFT) does.

A FFT analysis produces a graph of the sine wave only frequencies required to recreate the original wave form. This sum of frequencies is the aptly named Fourier Series. So as the Wikipedia link picture shows, a square wave can be represented as a sum of sine wave oscillations. (You'll notice the squiggles that appear immediately after an edge get smaller and higher in frequency as higher frequency sine waves get added.)

So the higher frequencies the OP is seeing comes from the fact that the vibration is not a sine wave in shape at the measurement point.

__________________
"Don't disturb my circles." translation of Archimedes last words
Reply Score 1 for Good Answer
Guru

Join Date: Jul 2007
Posts: 4448
Good Answers: 143
#3
In reply to #2

Re: Reading a spectrum graph for vibration analysis

05/09/2010 11:04 AM

Yes, you're quite right. I have this neat little waveform generator box that I try to use in electronics class. You can add up to 8 frequencies at different amplitudes and get rough triangles and square waves. Even when they see it, students seldom believe it. Besides, every time I try to show somebody a Fourier series, I always either lose or add a π. Why the heck can't Congress pass a law changing that to just 3 so I don't have to keep track?

__________________
"Well, I've wrestled with reality for 35 years, Doctor, and I'm happy to state I finally won out over it." Elwood P. Dowd
Reply
Anonymous Poster
#4

Re: Reading a spectrum graph for vibration analysis

05/17/2010 11:36 AM

Thank you for the replies, but if you don't mind, can you please explain to me why are harmonics important? I mean in this case, the shaft is rotating at 1500rpm, it will not rotate at 3000rpm or 4500rpm, so why is harmonics important, and how does one even measure the amplitude at harmonic frequencies in the 1st place?

And secondly, "why" does unbalanced shaft only have 1x amplitude and not both 1x and 2x? Is there any physical reason for it, or is it just base on experiments?

Thank you!

Reply Off Topic (Score 4)
Guru
Engineering Fields - Electrical Engineering - Been there, done that. Engineering Fields - Control Engineering - New Member

Join Date: Dec 2008
Location: Long Island NY
Posts: 15600
Good Answers: 981
#5
In reply to #4

Re: Reading a spectrum graph for vibration analysis

05/17/2010 12:06 PM

A very good question!

A perfectly balanced shaft will not generate any vibrations, but as we all know nothing is perfect. The easiest frequency to excite is the fundamental frequency because the least amount of spring constants must be flexed and these springs will have the lowest spring coefficient being the longest shaft lengths. The next harmonic will require twice as many springs to be flexed with shorter, stiffer shaft lengths, and so on.

Now depending on where along the length of the shaft the imbalance resides will partly determine the ratio of the energy distributed to all of the possible frequencies. This is why on a guitar, the closer the string gets picked to the end of the string the more harmonics in the fundamental note exists.

Simarly to the guitar amplifier being able to output the harmonics of the fundamental note, the FFT analysis itself will measure and display the higher harmonics that exist.

__________________
"Don't disturb my circles." translation of Archimedes last words
Reply Off Topic (Score 5)
Anonymous Poster
#6
In reply to #5

Re: Reading a spectrum graph for vibration analysis

05/17/2010 1:43 PM

Thank you for trying to explain it to me, but I guess I am a bit slow :(

I am still not understanding as to why 2x frequency is important and how did we even managed to measure the amplitudes of the harmonics in the first place since the shaft only stayed at 1500rpm?

And in terms of "it is easier to excite the fundamental frequency", wouldn't it be just because fundamental frequency is at a lower rpm, which mean less rotational force is required etc...? But I still don't understand why or how a misaligned shaft managed to have a amplitude at both 1x and 2x frequency while unbalanced shaft only managed to have at 1x? It felt almost as if it is because misaligned shafts is easier to "vibrate" than an unbalanced shaft, but then I would have thought that misaligned shaft should just have a higher amplitude at 1x frequency...

And another question arise is that if 2x and 3x frequencies are important, why not 2.3, 2.6 or 3.4x frequencies important as well? Even though I can understand that the maximum amplitude will occur at harmonics due to resonance only occurs at harmonics, but that doesn't mean the rest of the frequencies are not important?

Thank you very much for your patience!

Reply Off Topic (Score 5)
Guru
Engineering Fields - Electrical Engineering - Been there, done that. Engineering Fields - Control Engineering - New Member

Join Date: Dec 2008
Location: Long Island NY
Posts: 15600
Good Answers: 981
#8
In reply to #6

Re: Reading a spectrum graph for vibration analysis

05/17/2010 2:14 PM

Ok, one last try because you're on the edge of an important understanding.

First a little understanding of what and how a FFT analysis works. This is a well understood mathematical analysis of a large collection of data samples taken at a regular interval of time. (All of the data points obtained by a digitizing oscilloscope.)This analysis takes these incremental data points and translate them from a voltage versus time information, into a an amplitude versus frequency display. There are a variety of mathematical limitations and distribution effects that happens when a frequency that actually exists in the data taken does not exactly correlates to a specific frequency that can be displayed but this gets into a very complicated side discussion. The general rules of thumb are that one wants to sample as fast as possible and to use as many samples as possible for reasonable resolutions. The raw mathematics of this process is typically at least a half semester in a undergraduate engineering program. So for now let's just accept that it works.

Now onto why an integer times the driving frequency is so important to examine. The FFT analysis will display the amplitude of all frequencies that exist in your data and not just multiples of the driving frequencies. So the frequencies that are not multiples of the drive frequency are not there because of the driving frequency. There present from another source. So to see if the mechanical driving signal is finding a non-linear load, like a damaged bearing, then one would expect to see more high frequency multiples of the fundamental frequency to appear. This way one can test and ispect the operational quality of the whole machine without a physical teardown and failures prior to becoming a catastrophic failure can be detected.

__________________
"Don't disturb my circles." translation of Archimedes last words
Reply Off Topic (Score 5)
Guru

Join Date: Jul 2007
Posts: 4448
Good Answers: 143
#7

Re: Reading a spectrum graph for vibration analysis

05/17/2010 1:44 PM
__________________
"Well, I've wrestled with reality for 35 years, Doctor, and I'm happy to state I finally won out over it." Elwood P. Dowd
Reply
Reply to Forum Thread 8 comments

"Almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, vote them!
Copy to Clipboard

Users who posted comments:

Anonymous Poster (2); redfred (3); TVP45 (3)

Previous in Forum: Control Valve and Positioner   Next in Forum: Material specification

Advertisement