Originally posted by jank:

quote:

It was interesting to see the results of the RIC test on a motor with 22 bad bars out of 51 from the PdMA site. As the PdMA states, there is some kind of anomaly on the RIC test.
In case that you see the results of the RIC test only on a motor similar to the one from the PdMA site, would you make a call?
I can see 3 waves that are virtually exact sine waves with total harmonic distortion probably lower than 5%. It seems to me to be a pretty poor resolution considering that 43% bars are broken or cracked. One has to think what would the RIC test look like if only 2 or 3 bars were broken?

I have to admit, I have little experience with the RIC test but the above question adresses the core issue: How sensitive is this test?

I have read the paper and could not find a good reason for making a call the author made based on the slightly distorted sinwaves.

David

Good points.

On figure 3 the current signature shows pole pass sidebands only 36 db below the line frequency - an easy call from current signature.

But I think everyone agrees the on-line current test under load is more sensitive than the off-line tests.

The important comparison is RIC vs single-phase test (both off-line tests)... which would you have more faith in?

I have been studying Jan's graphs for awhile trying to figure out what they mean. I’ll put it in my own words, and you tell me if I’m wrong.

What we see is impedance vary with voltage, and at low voltage it’s unbalanced and as voltage increases it becomes balanced. (The low voltage power frequency region where the impedance is unbalanced is comparable to the RIC test on a voltz/hz basis.)

In a linear system, impedance doesn’t change with voltage magnitude.

For a magnetic system, the impedance changes at the low end and the high end and is constant (linear) in the middle. So these test voltages (and presumably the pdma operating at similar volts/hz) must be in the low end where the exciting amp-turns are not much greater than the coercive amp turns (Hc) and the resulting flux density is not much greater than the residual flux density (Br). In these ranges the results are heavily influenced by the non-linear effects of Br and Hc. (Also we might expect the excitation current to be non-sinusoidal in this range.)

I think this is a similar conclusion to the fact that residual magnetism is known to intefere with the RIC test?

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

UPDATE ON THE ORIGINAL MOTOR
Yesterday, I observed the disassembly/inspection and test of this motor.

(As you remember, it had 45 dB pole pass sidebands stable over time)

Single phase test showed no significant variation in current as rotor was rotated.

2000 amps was injected from end to end of the shaft. We had about 6 hotspots. The worst was at the bottom of the motor 140C at 10 minutes. Several others 60C-100C. rotor core average temp ~ 30C. Green flux paper indicated there was normal current flowing in all these bars.

We pulled off the fans and steel endplate from the rotor for better inspection (slides 1 shows rotor before disassembly, 2 and 3 after disassembly).

At the location of the 140C hottest spot (slide 4), we could see a crack which ran along side of bar (between bar and slot in end-ring).

We saw several other cracks. The worst crack is shown in slide 5 (there was not hotspot at this location).

It seems like those cracks could result in an open circuit of the end-ring. I’m not sure how much the current injection tests current flowing in this pat (as opposed to core test). We couldn’t do core test since the holes in the spider weren’t big enough to insert the test cables.

I don’t have the thermographic images yet. Should have those on Monday.

Any thoughts? The shop recommends that we melt out all the old solder and re-solder. I think that’s just a few thousand dolars and a few days work. But based on time and to lesser extent budget considerations, the plant would rather get the motor back and reinstalled, save the money this year, and schedule to pull the motor again in 3 years to do that repair. Since it has been stable, that seems reasonable to me.

Any thoughts at this point on the severity? Has anyone seen anything like this?

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

LPHD_Compact.ppt (278 Kb, 22 downloads)

Here is a repeat of the earlier powerpoint with some thermography shots added.

Slides 4,5,6 are the hottest spot, "location 1" at bottom of the rotor (reached 140C at 10 minutes). The hotspot shows up better from the side of the rotor (through gap between end of core and end ring) than from end of rotor. We did verify normaly current flow in this one with green paper.

Slides 7 and 8 are the worst crack by visual inspection, "location 2" at the top of the rotor. It didn't show as much of a hotspot from the side view. But from the end it's a strange pattern that I can't make sense of. Any thoughts on that? (slide 8?) We didn't use green paper at this location (I wish we would have).

We are going to do the logical thing and repair it now (melt out the solder and resolder), rather than putting it back in service and pulling it out again later for repair.

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

LPHD_Compact2.ppt (528 Kb, 16 downloads)

There are several good reasons not to take the thermography on the rotor too seriously.
Passing the current through the shaft means creating magnetic flux circulating in the back iron of the rotor around the shaft, similarly to the flux in the current transformer. If there is a conductive path around this flux, the current will circulate through this path.
If the cage is not insulated, the conductive path always exists. The path consist of: 1. Shaft, 2. Lamination on one end of the rotor, 3. Rotor bars, 4. Lamination on the other side of the rotor, 5. Shaft. The circuit is complete. In other words the path around the flux is quite complicated. Also, this path has never been designed to carry any current at all! So the current will find few “good” spots to transfer from the lamination to the bars or the rings. But the spots are not good enough; there is enough resistance to cause the spots to heat up. The thermography picks those spots as the “trouble” spots. But in fact those spots have nothing to do with the “week” spots of the cage.
Let’s examine the posted example. If I measured correctly, the length of the iron is about 20”, and the diameter about 30”. The number of bars is about 66. From the data in the original posting one can calculate the current in the bar to be about 477 Amps(at full load). If the injected current through the shaft was 2000 Amps, then the current per bar could not have been higher than 33 Amps (2000/66). Yet according to the thermography it caused a noticeable heating.
Obviously, at the full load the watts in the “fault” would have to be considerably higher: (477/33)^2 = 209 times higher. And during the start-up (5x nameplate current) the situation would be even more serious. The watts in the fault would be (477 *5/33)^2=5223 times higher. If one calculates that the watts during the thermography reaches 100 W, then during the startup it would be 500 kW! The bad spot would probably evaporate.
I have tried the method on a rotor with aluminum cast bars. The end rings were touching the lamination. At about 4000 Amps through the shaft, the lamination on either end of the rotor were glowing red-hot. The extremely poor conductivity of the silicon steel lamination, transferring the huge current to the end rings resulted in this enormous heat build up.
I do not use this method any more.
jank

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

Those are great comments. I have been wondering about the path for current during the injection test and also during a rotor core loop test.. I'm still thinking about that.

The thermography did take us straight to the location of several cracks in this case like the one on the bottom at location 1. The crack is located between the bar and the endring and it extends the entire length of interface between bar and endring around the corner from the side view to the end view. Do you think from the visual images that the crack appears to be a problem?

Looking at the thermography at location 2, it looks like current might not flowing in the copper bars, although I'm not sure. But again it is different at this location than at the others, so it gives a clue of places for a closer visual inspection. But visual inspection is also of somewhat limited value.

It seems like we could get a better thermography test more directly testing the endring to bar joints if there were some means to clamp directly to the end-rings, but they are not very accessible for a clamp. How about if we had an ajdustable copper strap that you could cinch up tight all the way around the end ring to apply current?

Any other suggestions?

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