Stator failure 6kV synchronous motor 1

[henkv]

Recently we have had a ground fault in the stator of a 11MW synchronous motor. A condition measurement(PI,PD, surge) 1 month before the event showed no indication of deterioration. The motor is directly powered by a VF drive.Cable length is too short for voltage doubling. Measured rise times do not exceed 5000V/us. Motor is only 3 years in operation and winding temperatures never exceeded more than about 50 degrees C. The failure occured while in "idle" mode, taking about 1% load. So far we have no indication of a cause.

 

[Steventyj]

Was the idle mode a common mode of operation or was it associated with something unusual such as an area shut down?

Have you had a chance to give a visual inspection of the stator to rule out the obvious causes such as contamination/moisture?

If you do not look for electrical faults in your VA analysis it may be helpful to have someone review your VA readings for anything unusual, but not bearing related.

Please keep us updated if you discover anything.

 

[swgrmfg]

henkv:

A few questions....

1. How far away from the neutral point did the fault occur?

2. A machine this size should be equipped with a 'Rotating Machine Surge Protection package'. Is this one protected? If yes, is the package located at the motor or far from it?

3. You said that the machine was in idle mode, do you normally shut it down from idle mode? If yes, what was the power factor at that time? the current? and what type of breaker is it that opens the entire circuit? Is there a breaker between the drive and the motor?

GH

 

[freddycor]

henkv:

What kind of load move this motor?
Do you know if not have problems with the peaks of the drive?
Any problems with the drive before the failure?

Alfredo Corona
Mexico

 

[electricpete]

All good questions. A few more:

Since you say motor was tested (PI, surge) one month prior, I assume this was a maintenance outage. How long after machine restart did failure occur? Was some form of space heating used to keep the machine dry?

I would also go back and take another look at the surge test results to make sure there was no slight indication of failure detected/caused during that potentially-destructive test.

 

[GusD]

Hi henkv

Stator shorted to ground after some tests were performed shortly before the failure.
You mention that PI,PD and Surge tests were performed ,but you don't mention at what voltage levels were these tests
performed at.
The PI (polarization Index) test is usually performed at much lower levels than the operating voltages.It measures the dialectric value of the insulation but does not stress motor to its operating levels.The same could be true for the surge tests.As for the PD (partial discharge) test,it too, may not stress the insulation to its operating levels.
Regardless of what tests one performs ,unless you apply
opearting voltages that stress windings (such as in HI -Pot tests)Phase to phase or phase to ground insulation cannot be guaranteed (99%at least).
Understanding that TEST voltage levels for used motors would be slightly lower than FOR new motors (2Xnominal Voltage +1000 )is the rule for new motors.
The other suggestions regarding possible low ins.values due to Hi-humidity or others should not be discounted without further investigation.

GusD

 

[henkv]

Thanks for responses all.
-Motor is part of a steel rolling facility with a load cycle of say 3 mins full load, 6 mins idle mode(15% speed, 5% current, power factor ??)It is a 12 pole motor running at 170 rpm nominal speed.
-Inspection after failure showed no moisture. Some contamination was found, analysed in 2 labs, no cause for failure. There were also no traces of corona, loose wedges or dust from vibration problems.
-We have crowds looking at the measuring results, maybe one lead to an early indication coming up.
-Fault occurred in 1st coil, 1st turn, from motor terminals.
-No surge protection package installed, measured rise time within expectation from drive side and motor design spec, there is no breaker between motor and drive.
-We have regular bi-weekly shutdowns that we can use for condition measurements. Space heaters are in use. Motor had been in operation for a month, after the measurement before the fault occurred. We have an on-site motor maintenance team, which is experienced in large motors.
- We have had no problems with the motor/drive before the fault.
- Because of motor confuration - coils in parallel and in series, surge tests will give little information.
- Hi-potting will not be done as a maintenance test, other measurements are done at nominal voltage

 

[induction]

1 turn first coil is a sign of start up surge in a motor with the many poles. Is it a vandergraf roller?If they rewind them the should use high feq wire. The class of insulation is much higher than normal wire it would had the start up better. I agree with the person who wrote that P.I. tests use low voltage and may not show moisture. A baker surge tester would do the job better than most test I have seen it can stress the insulation properly. At a steel plant I did P.M. the oils and humidity was high. You may not be able to see it but it is there. Try the wire change if possible if it requires rewind.

 

[electricpete]

1st coil first turn is a sign of possible surge, although I'm not sure why would be limited to those associated with startup. Also 1st coil has highest voltage stress to ground.

Is there sign of turn or ground failure? If it was a turn insulation failure, there is normally a lot of copper melting. Is the location of failure on end turn or slot?

You say there are difficulties with surge testing on this particular motor, but surge testing was conduted. Is it possible the failure/degradation begun during surge test but was not noticed at the time.

 

[mirbaghri]

henkv
You say that the PD test was done on this motor. I think that this test was done in offline situation and it's results didn't show any problem.if this is true it must be recommended that offline PD test doesn't able to detect all of the insulation defects. because the failure was occurred at 1st coil of the motor winding, I think that if there is any void or tree channel in insulation it would be recognized ( if suitable PDA instrument ws used).therefore insulation delamination may be occurred in your motor's winding.
Insulation delamination occur when thermal cycling applied on groundwall insulation because of different thermal expansion coefficients between copper and insulation or even different layers of insulation.
In stable thermal condition, e.g. in trips periods, gaps which is made by delaminaton dissapeared or very decreased and therefore insuation resistance and dissipation factor were affected notsensible

 

[henkv]

Induction, could you please explain what a vandergraf roller is? I don't think ours is. This motor drives a so-called 4-high mill through a 1:2 gearbox. We do use the Baker surge tester. Will look into the high freq wire, but I know that the wire for this motor has been insulated for surges according to IEEE 522. I am not quite sure what kind of startup you are referring to, as the motor is drive controlled, there are no large currents when we start running. The last time it was shut down was 3 weeks before the failure.

Electricpete, We assume it started with a turn short in the slot, turning into a ground fault. We have found melted copper/iron over about 40 cm length in one slot, but no large holes were seen. The motor was operating at a low current. We did not have difficulties measuring, but interpreting the surge test. On a motor with many parallel and series circuits, the response on the surge pulse does not really tell you a lot.

Mirbaghri, We have been told that an online PD measurement for HV PWM machines is not available yet. If you know a manufacturer with a working system, I would be very interested. If the motor ran hot, thermal cycling would certainly be considered as a cause, but as it is running very cool, this seems unlikely. Insulation resistance and dissipation factor had not changed in the last measurement before failure.

 

[electricpete]

Lots of possibilities. Once again the point regarding the surge test was that it seems possible that a turn insulation failure occurred during the surge test (this is a potentially destructive test), but was not noticed at the time.

The fact that it was a turn insulation failure at the first coil is consistent with this theory (as well as surges from other sources).

Assuming that you plan on rewinding the motor, during the teardown you may have opportunity to isolate the coils and hi-pot and/or surge test other coils to failure? That may give you a rough feeling for general strength of your ground and turn insulation (but of course may not detect problems localized to one coil).

Regarding mirbaghri's suggestion of thermal cycling.... it seems it may not limited to motors which run abnormally hot. Long length stators with a lot of up/down cycling may be susceptible.

 

[henkv]

We do have a plan in place where we will test the rest of the stator both electrically (surge, PI, PD) and then mechanically to see if we can find root cause. So we will have a comparison with other "first" coils to see if they have aged more than average. But we are interested to find out if other combinations of 6kV motors and VFD's have had similar problems. We do have a very frequent load/no load cycling. With the mechanical check of the coils we will find out if that cycling has caused serious delaminations. But then ... delaminations should not result in turn to turn faults.

 

[electricpete]

Good points. Sorry I don't have any experience with similar failures.

Out of curiosity, how did you measure your surge environment. That seems like a challenging measurement to make. Did you characterize the peaks as well as the rise times?

Also I assume your guys are looking at the existing motor insulation system design. Clearly dedicated turn insulation would be required for this application (I have seen several large motors without dedicated turn insulation... just used the strand insulation as turn insulation).

Presumably the machine has passed many surge tests including high levels when new. But also isn't the theory that surges can degrade insulation over time. I have also heard that mylar tape insulation can degrade over time under voltage stress.

 

[jbartos]

Suggestion: It appears that it may have been a voltage spike since:
1. First turn is involved
2. Test have been performed recently
3. Turn to ground fault
4. Idle mode of operation implying moderate temperature, no interturn stresses due to high current, etc.
5. Light load conditions, implying higher voltages
6. The highest chance for the weakest insulation spot, i.e. the lowest insulation and surface resistances to ground
etc.
The more thorough analysis of the fault and consequent damages is needed.

 

[induction]

vandergraf roller is a self contianed unit, the motor runs in oil with the acctual gears within the roller itself. Ok it is not a vandergraph. That changes it a little. Are the heaters on the winding or in the endbell. Ok in such a slow speed motor such as that one the are many recurculating current cuase by the number of poles. Oh re-read your last statement directed at the connection. Big difference, if has blow before was it in the same spot?
More so than than high speed motors low speed motors suffer from high core loss. I have seen where in a case such as yours that the motor had been repared and not core loss tested and had iron loss. It may not look like much for the size of the core but the rating are much lower for slow speed motors. That siad, ask them to test the core and and see if there is a hot spot there. That speed in my experience calls for either the coil itself to be wraped every coil with nomex as phase paper. Pure nomex as slot cell. I test a motor in your instance "time frame" use a megger or P.I. tester for the general tests only surge test maybe once a month at most it really should not hurt it but not worth the chance. Really wish I could see it could give you a better diag. Could you email me a picture?

 

[henkv]

Coils of this motor have both wire insulation and 2 layers of mica tape turn insulation.
After long discussion, decision was made to replace whole stator and not rewind. So we eliminated the risk of finding damages to the iron package. When inspecting with a boroscope through the ventilation ducts, we did see some damage to the iron package, but could not quite assess the extend of the damage.

Electricpete, not quite sure what you mean with measuring the surge environment. Our approach is to take "fingerprint" measurements when the new motor arrives and then compare further measurements to this fingerprint. Only if a clear difference is noted will we start to analyse the cause for this difference. So far for these motors we did not find significcant differences. Surgetest was done 2 times after the fingerprint with about a year in between measurements.

Jbartos, we are certainly considering the spike as a possible cause, but cannot find any evidence for an external event that would be able to damage a well insulated motor and not the sensitive pc's/drives/I/O alongside it. I tend to disagree with your point 5 as the motor was running at low speed at the time of the fault, so the VFD would only have given it a 15%rms voltage. Yes we are planning a thorough root cause analysis, but still, the fault may have destroyed the evidence.

Induction, I am not sure how I gave an impression of a recurring fault, but this fault is the first and only we have had on these machines. If you post your email address here I can send you a picture, but that will not reveal much more then a simple description here: a 4 meter by 4 meter stator housing with a 2 meter inside diameter for the rotor. And on the insde of the stator a minute black spot of 10 cm wide, 40 cm long.

Oh yes, I have to come back on what I said on space heaters. We do have them, in the end bells, but do not use them. This is because there is a water to air cooler on top of the motor that is always on, with water that is always 27 degrees C. So there is a constant flow of warm air circulating through the airducts, keeping the motor nice and dry at 27 C (well above dewpoint).

 

[electricpete]

It sounds like you've looked at the space heaters closely but I don't undestand because I'm not too familiar with water-cooled motors. Does it have forced arir even with thte motor secured? You cool it with hot water? Also 27C is 80F, not necessarily that much higher than ambient unless in an air-conditioned space (is it?). Also the entire winding must be at that temperature... have you verified winding temp above ambient temp during shutdown using winding RTD?

By surge environment I was not referring to the motor surge withstand, but the power system surges... specifically your statement "Measured rise times do not exceed 5000V/us" How was this measured? (what instrument, what location?).

 

[jbartos]

Suggestion/clarification to henkv (Electrical) May 1, 2003 marked ///\\Jbartos, we are certainly considering the spike as a possible cause, but cannot find any evidence for an external event that would be able to damage a well insulated motor and not the sensitive pc's/drives/I/O alongside it.
///Spikes do materialize alone or with surges in form of a surge peak.\\ I tend to disagree with your point 5 as the motor was running at low speed at the time of the fault, so the VFD would only have given it a 15%rms voltage.
///Agree. I meant the high input voltage to VFD. Many power electronic devices have input voltage range/window.\\ Yes we are planning a thorough root cause analysis, but still, the fault may have destroyed the evidence.
///Yes, however, certain traces of the fault remain, e.g. location, magnitude of the fault, some conditions under which developed, etc.\\\

 

[henkv]

Electricpete: Yes the air is forced through the cooler into the motor. Unfortunately the weather here in Holland is not that warm so 27 C is always a couple of degees above dewpoint. And yes we have checked this on the RTD's. The water is used for cooling lots of installations and is designed to be around 27 C. We once had a 2 day outage in winter (not that cold here) and at the end the water was still 27C!
Surge environment was a 200MHz scope with a fast HV probe measuring right at the motor terminals during rolling on the twin motor (different VFD)of the failed one. Later this was repeated on the VFD that drives the failed motor, when we had a replacement in place.

Jbartos: on your following point:///Agree. I meant the high input voltage to VFD. Many power electronic devices have input voltage range/window.\\\We have designed the supply side to be within range for both full load and no load conditions and we may have like a 10% increase on no load. I agree that this will have an effect in a worst case scenario.

I do appreciate everyone's input. We have 5 more motors like this and all are very critical for production. So we would love to come to a conclusive root cause.

Henkv