An old post floated to the top with discussion of the so-called swirl frequencies which are Fp sidebands around 5*LF (and maybe 7*LF?) in the current spectrum as a means of monitoring for broken rotor bars
http://maintenanceforums.com/eve/forums/a/tpc/f/7161085912/m/8501091271

We don't look at those frequencies. What is the purpose of looking there? What new/different info do we get by looking at those frequencies that we don't get by looking at simple pole pass sidebands around 1*LF?

I have never heard about “swirl frequencies” either. I was looking up the dictionary, and the swirl means “to whirl in eddies”. As I understand current analysis, the broken bar causes the variation in the stator current in general. Since the current contains higher harmonics, there is no wander that those harmonics are affected also, it cannot be otherwise. So the sidebands should appear around those harmonics as well. The point is that those harmonics are substantially smaller than the main current, hence the signal to noise ratio is much lower.
It is my opinion that looking for those frequencies will not provide any more information. Similarly as e-pete I do not look for them either. But I would love to find the origin of the theory behind those frequencies and why they are called “swirl”.
jank

They were first discussed in paper presented by a General Electric engineer whose name escapes me at the moment. I should be able to get a copy of paper and will post it for review.

Marty

Thanks guys. The GE guy famous for rotor bar detection research is Kliman.

Marty's comment reminded me to look through some articles I have filed away (since I know I had read something by GE/Kliman but I didn't realize it said anything on this topic).

I found the attached article "Broken Bar Detector For Squirrel Cage Induction Motors" by Kliman and Rao of GE. It is printed on GE letterhead and identified (last page) as document GER-3609 6/89 (500). No copyright listed anywhere, so I feel pretty good about posting it here.

I think it does indeed shed some light on my questions.

The terminology "swirl" is described on the left hand side of page 4/11 (above figure 8). Something to do with the nature of the flux anomaly in airgap directly above the broken bar.

I am not so much interested in the terminology swirl as in the significance of looking at sidebands around 5*LF.

I think the important features we are supposed to deduce about the 5*LF may be described in Figure 11 (bottom of page 5/11). It shows a broken rotor bar curve (1) which is relatively flat (or in fact increasing) as we increase harmonic number. The other defects discussed (2,3,4) all have comparable magnitude at the first frequency, but far decreasing magnitude at the higher frequencies. So perhaps corroboration via the sidebands around 5*LF are supposed to help us avoid making a call based on sidebands at 1LF only which could be from 2,3,4.

Note that even though the curve of Fig 11 appears to show us harmonics (not sidebands), I believe it is showiong us sidebands because:
1 - the label is db down (as would be expected for sidebands)
2 - on top right of page 4/11 he tells us in reference to figure 9 that "each line is actually two lines separated by twice slip frequency but not resolved on the printer at this scale" (I would presume similar logic applies to figure 10 and 11 as well).

Also Figure 11 represents airgap spectra (vs current), but I'm guessing they would be similar.

Do you guys think I am interpretting this right? He is saying that eccentricity type defects 2,3,4 might cause pole pass sidebands around 1LF but not around 5LF -> confirm rotor bar by looking near 5LF ?

Has anyone ever had a rotor let's say 35 db sidebands around 1LF which turned out to be a false alarm?

If Marty or anyone has other articles shedding further light on this, I'd be interested to see them.

This message has been edited. Last edited by: electricpete,

BrokenBarDetectorKlimanRao.pdf (1,734 Kb, 68 downloads)

Very interesting discussion.

EPete, I have often come across motors with PPF sidebands less than 35 dB down that on physical inspection did not reveal any rotor bar damage (see attached example). Spoilt my name out there for sure. Some motors have been showing strong PPF sidebands for years. The only clue is that the entire family shows the same pattern. I could never see anything different in their construction though. One theory was that the sidebands were due to bar caulking, I am not convinced.

See the attached paper. On Page 3, he states the effect of eccentricity is more pronounced around 5xLF as compared to that of the broken bars. This is in contradiction with GE. I will start modifying my set-ups to look at 5xFL for such & see what I get.

Regards,

Aditya

New_Fault_Detection_Techniques_For_Induction_Motors.pdf (1,190 Kb, 48 downloads)

Can't do two attachments in one post I guess.

I have attached the high PPF sideband example here.

Regards,

Aditya

PPF_sidebands.doc (184 Kb, 40 downloads)

Is Swirl a slang termminology for Sideband? If it is, then is it related to amplitude, frequency or phase modulation?

Walt

I have attached a PdMA document that details the different parameters they use to diagnose rotor defects. If you scroll to the bottom of page 7 you will see a brief description of what we are looking at as well as some example data.

From my experience it is not present 100% of the time when you have a rotor problem but when it is present you are pretty much guaranteed something is going on.

Noah, please jump in with your expereiences.

Regards,
Marty

Fault_Zone_Rotor.pdf (753 Kb, 41 downloads) Diagnosis of rotor faults

Thanks for the info Aditya and Marty. In Aditya's example it is discouraging to see a motor with those sidebands not showing broken bars. Makes it seem important to understand these other possible discriminating factors.

I am still reading thru Aditya's article. Interesting idea about superimposing an alternate frequency power supply. As I gather the main purpose was to help separate pole pass sideband from the main peak (?). If that's the case (and I may be understanding it), seems like a lot of effort when we can usually achieve the same thing for most motors at 50% load with higher resolution on the FFT. I agreed with your interpretation it seems to conflict with the GE.

Walt - the swirl effect is just referring to the fact that the vector representing the flux disturbance from the broken rotor bar is spinning as it orbits around the motor as mentioned in that GE writeup.

quote:

from Aditya in another thread:
Do you have an idea about whether the presence of these sidebands around the line frequency harmonics is more likely to be eccentricity or bar damage? As mentioned earlier, GE & ABB seem to contradict each other on this point.

The pdma writeup posted by Marty W also mentioned that the pole pass sidebands below 5*LF are supposed to be a way to help confirm pole pass sidebands.

I have read in another location that the existence of several pole pass sidebands below 5*LF with none above is the result of combination of FM (from torque-> speed oscillation) and AM. If that is true (and I'm not sure it is, since it is sort of opposite of how the sidebands around 1*LF behave... they are one-sided on the left in absence of FM and only appear on the right in presence of speed oscillation)), then the FM part would seem to be unique to rotor bars, not eccentricity.

I'd like to go back and review those GE and ABB articles again to make sure I read them right. At the moment, I'm not sure whether GE was talking about multiple harmonics of LF having sidebands... or else multiple sidebands of pole pass frequency around 1*LF. The latter interpretation seems to fit in line more with the discussion in that article that there is a sharp (rather than sinusoidal) variation in the airgap flux at location of the broken bar, traveling at pole-pass frequency with respect to the main flux. In contrast, rotating eccentricity should give smooth sinusoidal 2-pole disturbance to the flux and single pole pass sidebands. At least that's how it seems to me at the moment

I was looking at the sidebands of the higher harmonics from the case shown on this board under “Classical Rotor Bar Problem”. Just by coincidence I took one current reading with Fmax=1000 Hz. The attached ppt file shows that the pole pass sidebands are not only around 60 Hz, but also around 3x, 5x, all the way to 11-th harmonic. I am still not sure that it shows anything more than the basic current signature. Even the 13-th harmonic could have been (with little imagination) included on the Power point presentation.
jank

SWIRL_FRQUENCIES_.ppt (1,296 Kb, 42 downloads)

It's been a while since i have had time to post anything. Swirl is the term used by PdMA to describe the peaks in the low res spectrum @ 300 hz. Swirl by itsef has little significance. With that technology the high res test must have the sidebands @ 60 hz first, then use the swirl to confirm a suspect rotor. Also the inrush test will show fluctuations during a start. Offline test is a RIC with flat spots in the sine wave.

That said i never condemn a rotor/ motor with one technology.

There must be sereral indicators in the vibration spectrums as well, rotor bar frequency with 2Xlf sidebands, and a beat frequency in the time waveform.

I have good results with rotor issues if all tests show indications of a problem.
Stators are another issue.