What is causing my equipment to arc over ?

[Scrummer4]

I am trying to find the phenomena (plural) which are causing my equipment to arc over. We are talking about a 400 Vac, 12 pulse, 50 Hz rectifier. We had all 3 phases of the delta rectifier arc to the nearest earthed panel. The equipment has been in service for about a month, and is one of 4 identical equipments. Clearances between the 400 volt bars, fuses and panels are all compliant with the relevant standards. 11th harmonic filters are fitted.
This has only happened once at this 2 MVA installation, but a similar situation has been seen with similar equipments, not only ours, but also competitors, who I believe are also struggling to find the causes.
We are considering transients caused by system switching, resonance triggered by such switching, PFC interactions, and similar causes. But my underlying question is “How is the (more or less) 50 kV generated which causes this arc ? Then, how do we prove, disprove, and test for these apparently random conditions ?
Any ideas, references to papers, websites, etc., will be more than welcome.
Hope I haven't rambled on for too long, but it is a complex situation to explain.
Thanks.

 

[magoo2]

Is that a typo? 50 kV! I'm not sure how you'd get that high a voltage, but I wanted to make sure you didn't mean something else.

Also, how much insulation to ground (or air gap) do you have?

 

[wbd]

What kind of environment is this equipment located in?

 

[ghostbuster7]

Try hanging a high speed powerline monitor(1 microsecond response or better) on it.The captured voltage waveform shape should get you pointed in the right direction.

We had a similar situation and found intermittent flashovers were occuring on the utility faulty insulators.This caused subsequent flashovers to occur on electrical equipment connected on the secondary side.

Good Luck and good hunting.

 

[Skogsgurra]

I had this in a steel rolling mill once. The reason for this were deep commutation notches coming from another thyristor rectifier on the same grid. This rectifier did not have commutation reactors installed. Not on armature, nor on excitation rectifiers.

These notches sometimes turned the thyristors off and caused the magnetic energy stored in the commutation reactors to flash over to whatever was closest, which - as seems to be your case, too - happened to be the steel in the cabinet.

The snubbers are designed to take care of the energy caused by reverse recovery snap-off, but not the energy released when a conducting thyristor is forced off. The energy (W=1/2*L*I^2) is then orders of magnitude larger and the resulting voltage can be very high. Thyristors are very good isolators when the carriers have been removed by a reverse pulse.

We installed commutation reactors on that other equipment and got rid of the problem. Please note that also smaller, single-phase thyristor rectifiers need commutation reactors. Especially if they supply heavily inductive loads, like field windings. Those supplying resistive loads are usually not critical.

Installing a transient recorder may be an interesting thing to do. But there is a real risk that the transients will destroy the instrument. They are seldom designed to withstand more than 4 or 5 kV.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

PS

Is this in Italy or Spain? I have seen this problem (no commutation reactors) with equipment from both countries.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

There is another possibility. It is the delta rectifier that has this problem - not the wye rectifier? Right? You could try to arrange a symmetric star connected set of resistors between phases and ground the star point. Use something like a few kiloohms and corresponding power rating, say 100 W. That effectively bleeds off any charge that might accumulate. It seems that the wye neutral may already have a high-impedance (or low impedance?) connection to ground.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

Sorry. Forget my last post. The rectifiers are connected via the DC bus - aren't they? So, any existing ground on the wye side also acts on the delta side. Or???

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Scrummer4]

Thanks for all the ideas.
magoo2. No, it’s not a typo. If we say about 25 kV per centimetre, and this arc “jumped” about 2 cm, we have about 50 kV. Like you, I don’t know how I could get that high a voltage, which is why I’m looking for all the ideas I can.
wbd. Environment is an air conditioned room, in southeast Asia.
Skogsgurra. Rectifier is running at only about 20% load, so current to be commutated would be only about 200 Amps. We don’t have comm inductors; load is predominantly capacitive (DC Filter).
I had also thought about the isolated delta aspect. Our configuration is a little unusual in that the star rectifier is supplied through a (star) auto transformer (step up 400 to 480 V), while the delta is an isolated winding. Yes, the two rectifiers are connected at the DC bus. We don’t have a neutral coming in to the rectifier; 3 phases only. Although the filter uses star connected capacitors, the star point is floating. The DC bus is floating, so the only “reference” we have is through the “virtual earth” of the neutral (which is hopefully earthed back at the HV trafo), about which the 3 phases rotate.
In one of the other problematic installations, where they had many similar flashovers, they use a 3 winding trafo, with both delta and star isolated, and they had failures on both star and delta sides. On this site, when they disconnected all the filters (in this case delta connected filter capacitors), the problem disappeared.
I’m thinking heavily as I write this. The 50 kV I mention above is between the lines and earth. (The flashover is to earth.) It is not between the lines, or line and neutral. Somehow, could we have a floating 3 phase system. Floating maybe for just an instant ? The isolated delta could be floating even further from earth. The filter has capacitors with a (small) capacitance to earth, but there are 10 caps per phase. Is this line to earth capacitance resonating and generating this high voltage ?
I don’t know if this is relevant. Power factor of the lightly loaded rectifier with IP filter will be unity or leading. However, the other site was more heavily loaded and would have had a lagging PF.
We have a number of problems solving this type of problem. First is that the failures are catastrophic, so must be eliminated as quickly as possible. This may mean that we “solve” the problem without knowing the cause of the problem. Second, is that the failures are random, with long time periods between them (one only on this site, so far). Third, both the client and the utility company are being very uncooperative, denying any part in this phenomena, and pointing the finger at us.
All we have is ideas, with no proof of anything :-( !
Keep the ideas coming in, please.

 

[ghostbuster7]

If this is a true delta systemn a 480 volt delta system, G.E. has documented in writing over 30kv. being developed on an arcing ground fault within the factory.Over a 2 hr. period,this arcing ground damaged over 20 motors.

I would still recommend to record the waveform transient.A simple high voltage divider network could be used to reduce the voltage to the monitor if this is an issue.

If you can't spend the time ,install high quality surge suppresion system with surge counters ($5k-$10k).If the count is high,continue the investigation.

 

[Scrummer4]

Thanks ghostbuster7. No, it's not a true delta connected 480 V system. It should be a 400 V star system with the N connected to E at the HV transformer. The 480 V comes from the rectifier input trafo which is a combined auto (star) and isolated delta. Only the delta feeding the delta rectifer is isolated, and it was this which flashed over. Could be the same thing ? Is the GE experience documented ?
There is supposedly a 300 kA TVSS already fitted in the IP disribution cabinet. I've already asked if this has a strike counter, but the answers are slow coming. I'm thinking about doing some measurements in the factory where we may see voltages higher than expected, but not high enough to flashover.

 

[Skogsgurra]

The situation where I saw high voltage was when thyristors were switched off close to zero amps, but not quite. The snubbers can handle transients caused by currents that are around 100 mA (thyristor holding current) so turning off a thyristor when it conducts only a few amps will cause a high transient voltage. The actual load has very little to do with this phenomenon.

The energy in the stray inductance in the transformer is proportional to current squared, if the thyristor I(hold) is 0.1 A and there are 2 A when current is interrupted, you will have (2/0.1)^2 = 40 times the energy that the snubbers were designed for.

I would install high-power varistors across phases and perhaps also from DC poles to ground. That would be "hiding" the true problem but, given that the failure is catastrophic, it is sometimes necessary to do things without knowing ecactly what is happening.

Use a 680 or 750 V type with a high energy rating. Some are shown in this data sheet:

http://www.epcos.com/inf/70/db/var_01/02210226.pdf

You can even use a 1100 V type, remember; you are not trying to protect against mild transients, you are trying to limit the transients to acceptable values. And then a 750 or 1100 V device is OK.


Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Scrummer4]

Gunnar: Thanks for the explanation. I see what you are getting at now. I'll have a look at the photos and see if the snubbers Caps look blown. Yes, varistors may help.
However, I can't help feeling that it is something to do with the fact that the delta is totally isolated.
To explain a typical 12P rectifier better, even when there is an isolated star winding for one of the 6P bridges, the star point of that isolated winding is floating. So in that case we have two totally floating secondaries, and on that other site, we have seen flashovers on both sides (star and delta).
I'll try and measure the line to earth voltage on the isolated delta, but I'll need a high Z meter (electrostatic ?), which I'm sure we don't have.
Then there's also the apparent effect of the filter. Being on the primary side (which should be an earthed neutral system, though not using the neutral), how can this affect the voltage on the isolated seconadry winding ?
Thanks for all your ideas.

 

[Skogsgurra]

Guess who's got one? Two, actually. One Gossen 6 kV 1%, teflon insulated for minimum leakage current. And another one that goes to 30 kV. Have not seen it for years. But it is there, somewhere.

I am not so sure that you need to measure that voltage. Interesting, yes. But I am not so sure it is a static build-up. But, then again, why not?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Scrummer4]

I have a feeling that it is the fact that the delta secondary is floating which is key to the matter. I wouldn't be expecting to measure 20 or 30 or 50 kV between these line terminals and earth. All I would be looking for is perhaps a trend, that this voltage creeps up, maybe to 1 or 2 kV. That would show that there is the potential (excuse the pun) to have a high voltage breakdown here. Maybe a small cap from each of these lines to earth would be enough to discharge these high voltages, effectively giving this floating secondary an earth reference.
But where does the filter apparently fit into this equation ?
And, we have hundreds of these equipments in service. Why only these very few, catastrophic failures ? There's something else !

 

[Skogsgurra]

Scrummer, you do not mean capacitors. If the static build-up is slow, capacitors won't help.

Are both wye and delta windings floating? I thought you said that the wye secondary is solidly grounded? In a 12-pulse rectifier, both secondaries are connected via the diodes and the DC bus. That would prevent static build-up.

Honestly. If you have costly breakdowns, just connect varistors as I said. That will take care of most problems except a direct flash-over from primary to secondary on the transformer. You don't have that? Have you?

The root cause may be a very interesting thing. But getting rid of the problems seems more important right now.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Scrummer4]

Gunnar,
Yes, I was thinking of capacitors, trying to give the delta bridge a 50Hz earth reference. I can see that varistors would do the job in a different way, actually conducting only when the voltage gets too high.
Wye secondary is by an auto-transformer, so the actual earth connection is back at the MV transformer star point. The neutral wire does not come into the rectifier cabinet: only 3 wires + earth.
I know what you mean about getting rid of the problem. However, we have a little bit of time; there is a lot of ducking and diving going on on site, and we are not getting any answers to our questions, so I have a little time to "ponder" on the real cause.
Thanks for your interest.

 

[itsmoked]

The auto transformer is not a variable transformer is it?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[Scrummer4]

No, it's not a variable transformer. It is the star primary which has an over-wind to increase the star secondary voltage. Star point is floating. No neutral in the rectifier.
What are your thoughts ?
Thanks.
Scrummer

 

[itsmoked]

I was just thinking that if it was, for some reason, variable the brush point could be intermittent causing the same spiking but it was a long shot.

You might also consider arc tubes. They are little glass tubes with carefully crafted electrodes accurately spaced. They arc at a specific voltage. If you acquired some that were, say, half the gaps seen in your system they would obviously be the arc point instead of to bad locations. These likely won't work if you have low impedance as the follow on current would probably cause catastrophic failure. These arc tubes are for shunting of high voltages like TV picture tubes and such.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com