Phase Shift Transformer Heating and other issue

[MattSW]

Gents,

We went to an oil and gas site recently and were asked to evaluate their distribution network. They had blown up a couple of 1000 KVA vfd and each time had been reported that the vfd internals saw 1000+ volts. No mention of where, but on some recording the vfd engineer said he saw that voltage. I never saw the report or anything, but that's what I was told.

The incoming power comes through a 1000 KVA step down from 12470 to 480 through a Wye-Wye xfmr. This splits between a wall of pump controls, mostly small vfd and a large 24 pulse phase shift transformer. The 24 pulse xfmr feeds a 1000 KVA vfd that feeds an 800 hp pump injecting salt water.

Our techs had ran a DGA on the 24 pulse xfmr and found some gassing that pointed towards heating. When we discussed the findings with the site boss he mentioned that a few years ago the 24 pulse xfmr was hot enough they said it glowed at night. On the nameplate of the 24 pulse xfmr it says unbalanced voltages will cause excessive heating of the transformer. When prodding further they said they used to run all the smaller pumps at 70 hz and the larger pump at 63 or more hz but the smaller pumps kept dropping out and faulting with a missing phase fault. So, we figured that excessive loading had cause the phases to become imbalanced which caused excessive heating of the 24 pulse xfmr. So, my first question is that I recognize this can happen, but why does the imbalance cause excessive heating? And can you calculate roughly how much of a temperature increase it can cause, in other words, if I have a 5% voltage imbalance my temp will increase xx%?

Second, when assessing the site I put an amp clamp around one of the grounds on the transformer and it showed 11 amps. While walking the site I put it on every ground I could find and most were between 5 and 10 amps, up to 14 amps. While looking the main step down xfmr I noticed the Xo wasn't ran to the actual ground it just had a copper bonding strap that was about 3" wide and very thin running to the xfmr case right behind the bushing to a nut that was painted. At my previous job commissioning 1 Mw gensets if we didn't use an NGR on the site we had to take neutral to ground with same same ampacity as the phases. Is it not like this on power xfmr? In discussions with a EE it's kind of thought that the site has a weak or open neutral and that if we take the Xo bushing to ground it should alleviate some problems. Any thoughts on this going forward? Other thoughts what might cause the issue with the ground current?

Thanks

 

[petersonra]

Gents,

We went to an oil and gas site recently and were asked to evaluate their distribution network. They had blown up a couple of 1000 KVA vfd and each time had been reported that the vfd internals saw 1000+ volts. No mention of where, but on some recording the vfd engineer said he saw that voltage. I never saw the report or anything, but that's what I was told.

i think i would want to see the report just to know what they mean by this statement.

The incoming power comes through a 1000 KVA step down from 12470 to 480 through a Wye-Wye xfmr. This splits between a wall of pump controls, mostly small vfd and a large 24 pulse phase shift transformer. The 24 pulse xfmr feeds a 1000 KVA vfd that feeds an 800 hp pump injecting salt water.

So you have a 480V feeder to a bunch of small VFds and another to a 24 pulse xfmr.

Our techs had ran a DGA on the 24 pulse xfmr and found some gassing that pointed towards heating. When we discussed the findings with the site boss he mentioned that a few years ago the 24 pulse xfmr was hot enough they said it glowed at night. On the nameplate of the 24 pulse xfmr it says unbalanced voltages will cause excessive heating of the transformer. When prodding further they said they used to run all the smaller pumps at 70 hz and the larger pump at 63 or more hz but the smaller pumps kept dropping out and faulting with a missing phase fault. So, we figured that excessive loading had cause the phases to become imbalanced which caused excessive heating of the 24 pulse xfmr. So, my first question is that I recognize this can happen, but why does the imbalance cause excessive heating? And can you calculate roughly how much of a temperature increase it can cause, in other words, if I have a 5% voltage imbalance my temp will increase xx%?

If it really did glow at night why didn't they do something about it then? my guess is someone is exaggerating. if it truly glowed one would think that there would be some evidence of the overheating such as scorched paint.

The missing phase fault could be a hint. maybe you have an open phase somewhere upstream of the two feeders. I am a little surprised that on a transformer of this size there is not a thermostat to alarm on high temperature.

If you have a missing phase coming into the 24 pulse xfmr, 1/2 of the 24 pulse xfmr has to supply 100% of the power required. It would get warm. One would think though that the VFD would note this problem and trip.

Second, when assessing the site I put an amp clamp around one of the grounds on the transformer and it showed 11 amps. While walking the site I put it on every ground I could find and most were between 5 and 10 amps, up to 14 amps. While looking the main step down xfmr I noticed the Xo wasn't ran to the actual ground it just had a copper bonding strap that was about 3" wide and very thin running to the xfmr case right behind the bushing to a nut that was painted. At my previous job commissioning 1 Mw gensets if we didn't use an NGR on the site we had to take neutral to ground with same same ampacity as the phases. Is it not like this on power xfmr? In discussions with a EE it's kind of thought that the site has a weak or open neutral and that if we take the Xo bushing to ground it should alleviate some problems. Any thoughts on this going forward? Other thoughts what might cause the issue with the ground current?

I think you should probably have a connection to a GE on the secondary of the MV xfmr. It is an SDS after all. But there should be no current to earth through the GEC. but if it is not wired correctly, you might be splitting neutral current through the ground wires and the neutral wires.

Thanks

I think you should start by getting the one line or 3 line diagram for the SDS and verifying that the installation was actually done according to the prints. Your engineer should also review the prints to make sure that it was designed correctly in the first place.

 

[Jraef]

So the 24 pulse transformer is on the LV side to feed an 800HP drive? wow, what a bad design. Typically on something that big, they will do the 24 pulse transformer and step down in one swell foop; i.e. 12.47kV primary, 4 separate 480V secondaries shifted 15 degrees each from each other. But oh well, that's neither here nor there in this case.

Yes, when you have a phase shifting transformer for reducing harmonics, it is indeed very sensitive to voltage imbalance. That's the main drawback of that solution, but until recently there was no better alternative for something that big, so it was just an accepted consequence and typically, line voltage balance would be monitored to shut the drive down if it get out of tolerance. To give you an example from an 18 pulse drive study I was part of, at 75% load and less than 1% imbalance, the line current harmonics was less than 5%, the desired effect. But at just 2% line voltage imbalance, it jumped to 30% I-THD, which means the transformer was having to handle 30% more heating effect that normal. So that can indeed cause a transformer to significantly overheat if allowed to continue like that.

If they "tapped off" other smaller 6 pulse drives from one of the 4 separate outputs of that transformer, that would immediately be a severe current imbalance and trigger the heating effect. the line voltage imbalance causes a disproportionate current imbalance, which is the real culprit. In this case they were bypassing that indirect effect and just going straight to a current imbalance. That was a HUGE no-no! Someone had no idea what they were doing... It's possible that they now have a short in one (or more) of the windings in the transformer. That would lead to some VERY undesirable effects on the VFD behind it, especially on the rectifier diodes and if they short, the VFD DC bus jumps and kills the output transistors. It would all happen in a flash-bang event on something that size.

 

[MattSW]

Thanks for the quick replies.

I asked for for a one line already and they sent me some construction drawings, that’s it. I don’t really believe it glowed because this was supposedly over a year ago and I don’t think it would still be running. They had a an outside firm design the site and no one can answer any questions about who did it or where the rest of the plans are.

Dga showed temp faults and a low power discharge.

I think they they did the step down then phase shift because of the other loads on the site. The 24 pulse xfmr feeds only the 24 pulse drive.

What you said about the diodes shorting did ring a bell. When the drive fails it is failing the igbts. Could an arc inside the 24 pulse transformer cause the igbt to blow? Or what else would cause the igbt to fail on the vfd?

Thank you all so much for the help

 

[Jraef]

There are lots of things that could make an IGBT fail, a severe DC bus ripple and spikes due to damaged diodes is just the scenario that fit your other described issues. Usually it's something happening on the load side though, especially if the IGBT failure is the only symptom. Short circuits on the output, extremely fast and large step changes in load, things like that. If it's happened more than once, I'd be going over your output circuit with an ultra fine toothed comb. And adding a load reactor...

 

[Ingenieur]

if the xfmr glowed and operated like that for a period of time it is likely damaged
an oil immersed xfmr got so hot the enclosure/tank glowed?
or had an ir heat signature?

how is the xfmr case bonded to the ground system? 2 bonding jumpers?
the x0 to case bond should be sufficient
easy enough to messure: v for x0 to gnd and R x0 to gnd

 

[petersonra]

one would think that a system like this were the voltage variance between phases is so critical that the VFD would have protective trips to prevent issues with the xfmr.