Transformer Sheet Windings 3

[gordonl]

We've experienced a failure on a high current transformer with low voltage sheet windings. Anybody have any experience?

Transformer details:
34.5kV primary
1150V Secondary
13.3MVA
Aluminimu Windings
High Voltage layer wound
Low voltage sheet wound
54" sheet divided into 3 18" strips
Age 5 years

Any comments or possible contacts would be appreciated.

Thank You,
Gord

 

[electricpete]

I've never heard of a sheet winding. Is that anything like pancake coils used on shell form transformers?

By the way, IEEE57.125 on transformer failure evaluation is a darned good document with lots of technical info. As I remember there was alot of info about the way that different types of windings tend to distort under through-fault conditions, and the resulting weak points of each different style of winding. Do you have access to that document?

 

[electricpete]

I dug out my copy of the IEEE and I see there's a section on sheet windings.

By the way, it mentions that sheet windings are sometimes used in rectangular-wound transformers.... is this one of those? If so expect very low tolerance for through fault and don't bother to rewind it (buy a new one).

Here is some info from section B.1.2.7:
This winding consists of a single layer of copper or aluminum spirally wound on an insulating form. A layer of insulating material is wrapped between the conductors. Thermosetting adhesives can be added to the insulating material to bond adjacent conductors. The width of the sheet extends to the full height of the coil. Typically used in the low-voltage windings.

Inward radial forces in sheet windings are resisted by the rigidity of the inulating forms and layered insulating papers. During assembly, tension is maintained on the sheet condustor and laered insulation to eliminate voids and to provide a rigid winding to resist outward radial forces. The sheet winding is stronger than a circular wire wound coil in te axial direction under tension, but it is weaker than the wire wound coil in the axial direction under comprssiion. The thermosetting adhesives add rigidity to both radial outward forces and axial compresive forces. Oil duct spaces are designed into the windings to allow circulation of the cooling oil. The sheet wound coil may minimize axial forces by allowing current redistribution during thru-faults.

 

[busbar]

gordonl-- can you give us some details about the installation? Load character? Single or 3ø? High- and low-side winding configuration? Oil, air or cast insulation? Point of failure?

 

[alex444]

Scenario is pretty grim any which way you look at it .... and can be bad or real bad depending on how much money you will dishing out.

I know that winding repairs can be done in place, assuming the damage is minimal and accesible .. otherwise you are going to either re-wind or replace the transformer.

... but before you rewind or replace, you need to determine if this failure was due to a manufacturing defect or a transformer missaplication ... so you don't make the same mistake.

One outfit that could probably do this type of repair .. or at least point you in the right direction would be S.D. Myers webpage ...

http://www.transformerconsultants.com

... my experience with them has been very good.

Given the amount of info you provided, hopefully this will be of help.

Good luck

 

[gordonl]

Thank You electric Pete, that was interesting.

Sorry for the delay in replying here are some more details:

The aluminium was through every thing and we're opting for a total rewind. (Any comments there?)

They were circular windings as described in Pete's post (not shell), oil filled, 3phase, the load is a cycloconverter driving a 9200kW motor.

The windings seemed to have failed at the joint between the two sheets. The sheet is the full weight of the winding but split in three parts because of the availability of the sheet. The sheets are joined at the beginning and the end by the low voltage leads, but through out the winding the sheets are simply butted against one another. There was 10 mils of kraft paper insulation between layers and it seems to have been cut by the edge of one strip at the point where two of the strips meet.

A transformer designer from another plant has suggested that the problem is the fact that we have three sheets, and two would have been much better. When thinking about it, the center sheet of the three wouldn't have room for thermal expansion.

At a few points in the winding one sheet was riding up on the sheet beside it, which seems to be where the paer was cut.

The load on the transformer in very cyclical, the load is about 200% of nameplate, with a cycle of about on 45 seconds off 2 minutes. The transformer didn't seem to have a thermal problem (top oil max needle 30C), and when we disassembled one of the unfailed coils there was no insulation damage except for the creasing of the paper as described.

Because of the nature of the load we suspect there is some movement in the coil which may have caused the cutting action of the sheet.

Has anybody specified large current transformers with cyclical loads like this before? What winding configurations were used?

We're having the tranformer rebuilt with copper winding, an extra 10 mils of nomex between layers (along with existing kraft paper), and adding a small gap between sheets so they don't ride on each other. Any comments here?

Anybody read any articles on transformers for large drives which you can provide a reference for?

Any general comments on the above would be appreciated.

Thank You,
Gord

 

[electricpete]

Another aspect of the load which might bear some scrutiny is the harmonic content. I remember seeing many power plant electrostatic precipitator transformers that generated monstrous amounts of combustible gases due to the harmonic nature of the load, even though the oil temperatures didn't run that hot. Presumably the harmonics cause hot spot at certain metal parts within the transformer. The transformers I was involved with included at least 5 different transformers that all had the same behavior... also as far as I know they all have been running with those high gases for decades and no failures yet.

 

[gordonl]

The harmonics are quite high, over 20% ITHD on load, but we have not seen any gassing. We had a gas analysis done about a month before the failure, which gave us no warning.

 

[electricpete]

I believe if you have 20% THD in current, then your total heating will be about 140% (Irms^2) of what it would have been without harmonics.

I assume that you haven't had temperature alarms in the past since you would have told us. Is the unit equipped with winding temperature detector which simulate hot-spot... and give alarm or highest-temperature attained (red needle)?

If you have such a device, I think it would work properly to estimate the overall heating effect of harmonics. But if you are relying on oil-temperature alone, that is a dangerous proposition since the oil may not have time to heat up fully during the few hours of load.... and winding temperature may be higher than you think.

Also possible that 3rd harmonics are circulating in a delta winding?

How does loading compare to nameplate?

Are there adequate lightning arresters? Are they equipped with surge counters? Any history of recent through faults (especially in the last month since dga sample).
Just thinking out loud. I'm sure you're already looking at these angles and more. Questions are easy... answers are tough.

 

[electricpete]

It may be possible that the load generated voltage spikes which damaged the low voltage winding. This is a more believable scenario particularly if the failure initiated as turn-to-turn (not turn to ground) and occured in a part of the winding very close to the low voltage bushing. It was very common in the early days of asd's for turn-to-turn failures to occur at the line terminals of motors... now we use inverter-rated motors to prevent this. In the case of motors it turns out the cable length is very important in terms of wave behavior of steep-voltage fronts. In your case maybe a parallel capacitive or series inductive filter might help (getting ahead of myself... of course it depends on your investigation).

Is a sister unit in operation where you can monitor the voltage? If so you can compare that waveform to standard bil-testing waveforms.

Did you identify whether failure started as turn-to-turn of phase-to-ground? Did the insulation look to be in good condition or overheated, loose, spongy? Winding clamping looks tight? Windings have shifted? Location of the failure near the terminal? Any other observations during teardown?

 

[electricpete]

Also any anomalies from electrical testing before and after the failure (TTR, winding resistance, power factor, excitation current)?

 

[electricpete]

Some more info to amplify/correct some of my previous comments:

#1 - Computing heating in the presence of harmonics is quite a bit more complicated than my estimate of 1.2^2. I looked up IEEEC57.110-1998 and see there are a ton of details in there. I was suprised to see that Eddy current losses increase with the square of frequency. (although core hysteresis losses decrease with increasing frequency as I expected). A few other factors: they mention the possibility of a dc component of the load current which may push transformer into saturation (although I have a hard time understanding how a transformer can feed any dc component). Also sometimes zero-sequence triplen harmonics induced by the load can circulating in a delta winding.

#2 - In judging the severity of any surges you might measure... it may not be fair to compare them to BIL test current waveforms.... those waveforms are intended for infrequent application. The voltage spikes on transformer secondary may be several times per second. I know that in the case of motors there is a clear difference between repetitive-surge withstand capability and one-shot surge withstand withstand ... the repetitive surges eat away at certain types of mica paper over time.... I'm not sure if the same applied to oil-soaked paper insulation used in transformers.

 

[gordonl]

Thank You for your interest Pete, I'll try and answer your questions.

This transformer was delta - delta so I don't think the circulating triplen harmonics applies.

The failure occured about a quarter to half way into the coil. (The good coil with burn spot was quarter way in)

The designer suggested sheet windings were good at withstanding impulses because of the inherent high turn to turn capacitance. (Please comment on how relevant you feel this is)

There are reactors between the drive and transformers which would help attenuate any spikes. I don't have potential transformers on the secondary of the transformers, so I don't have a way to monitor voltage easily.

There were no signs of overheating in the coils.

The designer briefly reviewed the losses of the unit, which included increased eddy losses in the windings due to harmonics (over half of total loses) with a K factor of 6. Do you know what value of Ithd this would correspond to?As for test results I have two oil analysis available prefault and factory test results. When we found all the aluminum in the tank we didn't bother with testing, because of the magnitude of the damage they wouldn't have meant anything I beleive.

The RMS loading is in spec.

The fault was turn to turn (never progressed beyond)destroying about 15 out of 19 turns on the low voltage.

I think I better describe what we found on tear down inmore detail. We first put the faulted coil on the lathe (we could see damage out of the bottom of the coil). The entire high voltage was OK. The outside papaer layer of the low voltage had some burning (minor) The seemed to have started at the junction between the second and third sheets of a turn. (Each turn is made of three sheets side by side 18" wide each) Guessing by the severity of the damage at different levels, the fault occured around 1/4 to 1/2 way into the coil. At the center of the layers there was a large hole where the aluminum at "burned" out. Because of the large amount of damage at the fault, and the distortion of the low voltage coil due to the fault currents we couldn't draw any conclusions, so we tore down a good coil. (centre)
High voltage was good. In the low voltage we found spots were the paper was badly creased and starting to cut. This corresponded to places were the sheets of one turn were riding up on each other causing pressure points on the paper. About 5 turns down we found a burn spot (about an inch across) which occured at the joint between two sheets and seemed to be caused by a turn to turn fault, with arcing above and below to each layer of the sheet.

Because of this we beleive it was a mechanical/winding problem. I've done extensive long term monitoring of the load on the high voltage side, and have done 6 high speed captures of current when the load hits the drive, with no spikes. (All measured at high voltage)

Pete, I have pictures and DGA available before the fault which I can e-mail/fax if you want. Please e-mail me at gordonlees@yahoo.ca with your fax and e-mail data if you want to see these items.

Thanks Again,
Gord

 

[electricuwe]

To gordonl (refered to your post dated Dec 17th 2001)

K-Factor and THD are two integral factors calculated from the harmonic spectrum for different purpose. You cannot convert one to the other without knowing the harmonic spectrum. THD can be helpful to calculate the additional heating of conductor as long as eddy-current in the conductor is not an issue. K-Factor is used for the same purpose if eddy-current is an issue.

For more information search for papers written by Linden W. Pierce in the IEEE Transactions on Industy Applications.

To my mind the fact that the additional losses due to harmonics are more than half of the total losses indicates to me that the design of the transformer is quite poor. This fact leads to two problems:
- usually the additional losses a not equally distributed in the whole winding but concentrate at certain locations.
- If in some operating points the K-factor of the load current exceeds the value of 6 total losses and hot spot temperature increase very steep. You should check the K-Factor of the current in all operating points if you think that the problem might be related to thermal problems.

 

[gordonl]

Unfortunately I don't have an IEEE membership or access to a decent library. I had problems with some dry types and I had a very good article by Mr. Pierce on that topic, but unfortunately I have nothing on this problem.

I don't beleive I have a thermal problem though, there were no signs of insulation degradation when we tore it apart, only the one burn mark where a turn to turn fault was occuring.

 

[gordonl]

electricuwe,

Please have a look at this old thread of mine concerning dry type transformers if you don't mind, if you have any experience it would be appreciated.

thread238-10757

 

[gmazumder]

i think what you are talking about is also referred to as the Al foil transformers. These transformers are known to have high fault withstand capacity and are good performers.
you have provided lot of details in the thread and the replies and i need to know more details.

the high voltage winding i presume was conductor wound or it is also foil wound?

did you find any spacers within the turns, looks from your description that the mechanical fault intrepration is ok but the same can also occur if the transformer has faced major faults also.

remember these are high fault withstand capacity transformers.

any way need more details, periodic DGA results would be of help.

regards
gautam mazumder

 

[gordonl]

The high voltage was conductor wound.

The secondary is as you described aluminum foil (sheet). The low voltage had 19 turns, sarting from the core and working out: core - spacer - 3 layers of aluminum with .01 mils of paper in between - spacer - 4 layers - spacer - 4 layers - spacer - 4 layers - spacer - 4 layers - spacer - high voltage.

We had annual DGA analysis done with no signs of gassing, the last was a month before failure.