Hi,
Will you please help me to locate casting voids in rotor? Which test to carry out?
Also how to confirm core lamination problem in rotor?

I have already expressed my opinion on casting voids in the thread “Validity of the RIC test”. In my opinion, if the cast imperfection is not such, that one or more bars are totally open, you will not be able to find out.
I have presented the experiment with the “rotor bar” 76% open, which resulted with ~ 9% increase in the resistance of the bar. How do you want to find 9% difference among, let’s say, 58 bars? It is even my view, that there is no need to look for such faults, because they are irrelevant for a successful operation of the motor.
Think, for example, how is the crossection of a conductor reduced in any switch. The whole crossection of the incoming/out coming conductor is reduced to one or few points in the contact area, and yet the current will get through because the length of the “fault” is so miniscule. The heat developed in the reduced crossection will dissipate very quickly to the surrounding areas (copper and iron); hence the impact is negligible.
If there is a problem with the torque of your motor, it seems to me one should look elsewhere. The casting voids are a long shot indeed.
jank

There are many ways of finding the casting voids in rotors, these have been discussed in a recent thread 'motor current sidebands'.

Please check out the URL http://www.us.framatome-anp.com/ultracheck/pdf/ppm.pdf. I wonder why Jank says such voids can't be detected, have found them many times. It is another story that I too feel that the motors are not likely to fail overnight due to these and have trended such voids for years.

Regards,

Aditya

Thanks.

Sorry for late comming.

actually the color of the rotor has got changed to blakish blue on whole periphery at both the ends. The rotor is AL diecast and with full enclosed slots i.e. bars not visible.
And the rotor is laying outside.
how can i test?

Blackish blue does not sound good.

One good test is clamping onto the ends to send current through and monitor with infrared. With a fab rotor, you would clamp on the end-rings. With a cast rotor, I'm not sure exactly what you clamp on to (can anyone clarify that for me?)

Also there is growler test, and visual inspection.

quote:

I wonder why Jank says such voids can't be detected, have found them many times.
Regards,
Aditya

It would be interesting to find out, why you think that you found the voids in the bars.
I have read the article (your link to P/PM technology) with great interest. What I am skeptical about is the fact, that they find a problem and they try to fit it to the present knowledge. I do not think it is the proper approach. A good experiment in this field should look like this (in my opinion):

1) Test the motor and find it in good condition.
2) Introduce a void in one bar, assemble the motor, and retest it.
3) Compare the original results with those with proven voids.
4) Draw the conclusions from this experiment.

Testing the motor, finding the sidebands and later find the voids in the bars means nothing in my opinion. I would be very surprised if you did not find any voids in aluminum cast rotor. The void in the picture represents about 10% of the total bar area; a virtual nothing considering the number of the bars.
If an experiment is to be valid, the initial conditions have to be known exactly.
Looking at the picture of the void in the Boeing article I am pretty sure that the rotor would work flawlessly for another 25 years.
jank

Jank,

I have attached the test results for two identical motors tested on the same day. These are 160 kW, 415 V, 3000 RPM motors with cast rotors.

Do you feel that inspite of the obvious defects in the first one, both would have given the same performance for the same time period?

I am going down to this plant next week & will collect data on these motors again. I believe the client has rectified the motor.

Regards,

Aditya

Rotor_Bar_Case_Study.doc (160 Kb, 31 downloads)

Good discussion.

That is an interesting article and case study in the P/PM link . 250hp 1800 rpm motor with pole pass sidebands 43db down below line frequency in the current spectrum. Post-mortem of the rotor shows:

#1 - a void in on bar that goes axially the length of the bar from one end ring to the other (figures 9 and 10).

#2 - A dog-leg in the bars due to laminations apparently shifted prior to casting (figures 11). Shifting laminations would presumably affect every single bar in the motor and figure 12 taken 180 degrees opposite figure 11 appears to confirm that it affects all bars.

As far as detection - #2 would apparently not cause any assymetry since it affects all bars, and I don't think it would show up in current signature analysis sidebands. I would tend to think that #1 resulted in the assymetry that caused the oscillation. What else could have caused the 43 dB down? I am aware that there are some correlations between number of broken rotor bars and dB down.

For example - see equation 2 on page 3/8 of the attachment which reads as follows:

n_broken=2*R/[10^(db/20)+pp]
where
R is number of rotor bars = 46 (from Fig 8)
pp is number of pole pairs = 2 (4-pole motor)
dB = dB down in curent AT FULL LOAD

For dB = 39, the formula gives 1 broken bar
For dB = 43 (observed in CSA), the formula gives 0.6 broken bar.
For dB = 53, the formula gives n_broken = 0.2 (estimate of fraction void seen during disassembly)

Note - I think since there is a relatively low number of rotor bars, a small defect in rotor bar shows up on CSA more on this motor than it would on bigger motors with more rotor bars.

Jan - is it your objection, that for 20% broken bar, you expect to see higher dB difference such as 53? (or conversely for 43db, you expect to see 60% broken bar)?



I tend to think the correlations are not that exact. If this motor has 43dB, there must be some reason and the 20% void appears to be a logical explanation. What else would it be?

As far as severity - I think #1 is not so bad... a fractional increase in resistance of one bar. That bar will run hotter but it will be spread across the length of the bar. Unless there is a portion of the bar with very narrow cross section, it will not likely degrade. If it does degrade, we will likely have plenty of warning (if this were the only fault).

But in my opinion #2 is more of a concern. Here's why. We have a localized high-resistance defect on every single bar in the whole motor where it takes the dog-leg. Maybe it's not enough to cause any of them to fail. But in the unlikely even that the weakest one does fail, then as well all know it shifts the load to it's neighbors... but the neighbors are already in a similarly bad shape. It is not difficult for me to imagine that there would be relatively little time between opening of the first bar and the whole rotor becoming open-circuit on this motor. Seems like a bad situation - very little time between detection (first rotor bar break at the dogleg) and when motor would completely fails. This is not a likely scenario, but the consequences are much worse than typical rotor bar degradation that gives a warning.

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

t31chapter07.pdf (194 Kb, 23 downloads)

Adytia,
Two motors side by side is a golden opportunity for comparison, and one should take as much advantage of that as possible. It is apparent that those two motors operate quite differently (I assume an identical application). I tried to perform a current analysis from the attachments. On the motor with large sidebands it seems to be that the sideband is 27 dB (am I correct?) down from the peak. I do not expect this motor to come close to the other one. I would assess the 27 dB rotor to be in a pretty rough shape.
At the same time I have a hard time to imagine that the fault is caused by the porosity only. Seeing the porosity on other dissected rotors, it appeared to be randomly distributed. In may opinion a random distribution will not result in large sidebands, but more as an increase of average resistance of the rotor bar.
By the way, why is the fluctuation of the RMS current so large in both cases?
E-Pete,
I have to get back to your postings later.
jank

quote:

Originally posted by electricpete:
...is it your objection, that for 20% broken bar, you expect to see higher dB difference such as 53? (or conversely for 43db, you expect to see 60% broken bar)?

I have posted some current signatures before in other threads showing rotors with one broken bar. In one case the decibel ratio was something like 36 dB and in another case 50.5 dB.
The available formulas probably do not look deep enough. They merely look at the number of rotor bars broken. I do not know what is being taken into consideration in those formulas.
Without any proof known to me, I believe that the size of the sideband should also reflect the inertia of the system.
In my view (and I might be totally wrong), the larger inertia should help getting the motor through periods of lower torque. In other words the inertia would prevent the motor from necessity of “acceleration” periods after the slow down during the lower torque period. The acceleration period always sees a higher current; it would not be only the change of the impedance of the elements of the equivalent diagram.
Any thoughts about this?
jank

Electricpete,

The formula for computing the no. of broken bars does not give good results. For the data that I had attached, there are 2 poles, 39 bars and the dB difference is 29.24. This gives the no. of damaged bars as 2.52. I know that the damage was much more severe. The entire rotor had to be repaired and the diecast bars were replaced with copper bars.

I had a paper with another formula but cannot locate it right now. I do know that the Framatome software also takes into consideration the no. of PPF sidebands, the load percentage and the amplitude of the running speed peak in the demodulated current spectrum.

Jank,

I did not understand your question about the current fluctuation. Do you mean the modulation in the bad motor? I have attached some zoomed views of both motors' current waveforms, would appreciate it if you could explain the point.

Regards,

Aditya

Current_fluctuation.doc (389 Kb, 13 downloads)

quote:

Originally posted by Aditya:
Jank,
I did not understand your question about the current fluctuation. Do you mean the modulation in the bad motor?
Aditya

The line representing the RMS current is about 20 Amp thick, (even in the good motor), something between 170 and 190 Amps.
I have just recently monitored current on 600 Hp motor. The upper frequency was set to 1000 Hz, the lower to 10 Hz. (CSI 2020 analyzer, monitor mode). The RMS current was just a simple horizontal line. I am probably missing something (???). No big deal.
jank

Jank,

The time waveform that is shown is not the raw current data, rather a time domain representation of the demodulated current signal. Framatome refers to it as RMS Envelope, Baker call it Torque Ripple. You can see some fluctuations in the raw current signal also, but it is not as evident.

Regards,

Aditya

quote:

Originally posted by jank:
Without any proof known to me, I believe that the size of the sideband should also reflect the inertia of the system.
In my view (and I might be totally wrong), the larger inertia should help getting the motor through periods of lower torque. In other words the inertia would prevent the motor from necessity of “acceleration” periods after the slow down during the lower torque period. The acceleration period always sees a higher current; it would not be only the change of the impedance of the elements of the equivalent diagram.
Any thoughts about this?
jank

Jan - you ask some thought-provoking questions about the role of the inertia and how it affects the current signature.

I did some research to try to understand better. I learned something but still have some questions. I got the attached article from www.spectraquest.com. It says:

quote:

The cyclic variation in current leads to a torque pulsation at twice slip frequency [pole-pass] and a speed variation which is a function of system inertia. The speed reduction causes a reduction in the component at f1(1-2s) and a component at f1(1+2s) appears.

If I understand correctly, he is saying
#1 - I the absence of speed modulation, there is no USB (it appears because of speed modulation)
#2 - speed modulation causes the LSB to lower and the USB increases.

#1 doesn't exactly make sense to me. I would have thought that the torque pulsation causes current amplitude modulation which gives rise to a USB and LSB (which contradicts #1).
#2 makes sense to me. I have shown mathematically on the forum before that speed pulsation causes a USB and LSB. It's certainly plausible that the phase relationship of the FM LSB and USB when added to the AM LSB and USB can cause one to increase and the other to decrease.

Now I looked at some data and found some info that supports #1. I need a new post for a new attachment..... (see next post)

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

Spectraquest_Thompson.pdf (953 Kb, 18 downloads)

Attached is a current spectrum from an 8000hp 1200rpm vertical motor with a humongous flywheel (I think 8' diameter and 1' thick). With that flywheel we have tremendous inertia and you can imagine there would be no speed variation whatsoever. So with no speed oscillation on this motor, what kind of current spectrum do we see? A LSB with absolutely no USB. (The speed is 1188rpm, so LSB would be at 3528cpm and USB would be at 3672cpm) It looks like #1 was right after all!

Any thoughts to explain why the torque modulation without speed modulation causes only LSB (item #1)?

Which sideband do you prefer use to evaluate severity?
* LSB only? (I have heard people recommend this, but I don't understand why).
* Whichever is higher (LSB or USB) - that's what I do.

Abbreviations:
LSB = Lower Sideband
USB = Upper Sideband.

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

RCP_LSB_NO_USB2.ppt (87 Kb, 19 downloads)

I see that item #1 is mentioned several other places. I found this explanation in "Current signature analysis to detect induction motor faults" by Thomson and Fenger in IEEE Industry Applications Magazine, Jul/Aug 2001 Volume: 7, Issue: 4 on page(s): 26-34 ISSN: 1077-2618

quote:

A full theoretical analysis of a three-phase induction motor operating with broken rotor bars can be found in [10]. It is well known that a three-phase symmetrical stator winding fed from a symmetrical supply will produce a resultant forward rotating magnetic field at synchronous speed and, if exact symmetry exists, there will be no resultant backward rotating field. Any asymmetry of the supply or stator winding impedances will cause a resultant backward rotating field from the stator winding. Now apply the same rotating magnetic field fundamentals to the rotor winding; the first difference compared to the stator winding is that the frequency of the induced electromotive force (EMF) and current in the rotor winding is at slip frequency and not at the supply frequency.

This sounds like a reasonable explanation for #1.

Are we trying to detect casting voids and core lamination during motor manufacture or after sales?

Casting voids and core laminations in boiler plates are detected using ultrasound test (UT). The manufacturer is responsible to produce them according to the material specs.

For the most part we're talking about detecting defects after sale.

UT is sometimes applied for this type detection. But then you have to inspect the whole rotor (not knowing where the defect is ahead of time). The injected/induced current test can localize voids

So what are the aceptable limits of casting voids and core laminations to be met by the motor manufacturers? I thought customers don't have to worry about this. Well, before this post.

There are plenty of cast aluminum rotor motors which are delivered with casting porosities. Then it gives an indication of a rotor defect on the current signature analysis test. There is some question whether or not porosity defects can degrade in a manner similar to the way a fabricated-rotor high-resistance bar/end-ring joint could degrade. For the most part, I think the cast rotor defects don't degrade over time, but I know there are exceptions that have been identified here on this board.