unusual LTC DGA results 4

[electricpete]

This is an 84MVA 25kv/13.8 kv delta/wye transformer.
On the high-side, it has a Westinghouse type UVT tapchanger. This tapchanger type uses vacuum switches for breaking/making and preventive-auto-transformer for limiting current in bridging positions.
The LTC info shown on the nameplate is included in slides 1 and 2 attached.
The LTC compartment is separated from main tank by a barrier board (no gases showing in the main tank).
The LTC compartment has a dehydrating breather.

(I am told that) we normally operate in 12L position with very few changes other than every month we cycle from 12L to 9L to wipe the contacts to avoid coking. (That doesn’t quite make sense to me because that wouldn’t excercize the reversing switch.)

Below is a summary of the DGA (also included as slide 3 in case the formatting below gets garbled):
[tt]
Date H2 O2 N2 CH4 CO C2H6 C02 C2H4 C2H2
5/17/2011 16 10641 54428 18 185 25 2801 10 0
7/5/2011 Change in plant conditions forced LTC into 3L-9L for approx 24 hrs
7/13/2011 1037 7597 64113 3700 447 916 2809 5667 61
8/5/2011 1817 2821 68339 5611 450 1459 3298 9924 141
8/8/2011 1700 3707 68124 5698 454 1506 3496 10006 140
8/9/2011 1665 3235 65462 5419 431 1421 3041 9431 130
[/tt]

Beginning 8/5 we have done several thermography and ultrasound surveys and show no differences between this unit and an identical sister unit (no indication of ongoing fault).

Note in that chart that we performed an unusual evolution on 7/5 which likely pushed the LTC into an unusual tap somewhere between 3L and 9L for 24 hours, and then returned to normal POSITION. Based on the timing of the jump (could have been 7/5), and the fact that the level seems to have stabilized (now that we are in normal lineup), with no indication of ongoing fault, it is theorized that whatever arcing was going on only happened during the 24-hours when we were in the unusual lineup.

Questions:
Does this theory make sense to you?
Do you have any suggestions of the cause? (reversing switch, selector switch, bypass contacts, other?).
Any additional troubleshooting you recommend?
Why do you think the O2 went down.... seems quite a bit lower than usual?

(There may be some more details available regarding dielectric strength and other oil tests... will check on those tomorrow)


=====================================
(2B)+(2B)' ?

 

[electricpete]

A few peripheral questions that arise in addition to above:

1 - I tended to think LTC was sussceptible to coking if contacts carry current for long period of time without switching. To me, our reversing switch contact would fall in that category. But would not seem to fit the present scenario because in the normal current carrying position there seems no gassing. It was suggested by another engineer that coking also occurs on contacts not carrying current which remain in that condition for a long period of time. The latter is unfamiliar to me, is it a possibility?

2 - How high a level of combustible gas in oil is tolerable before it should be considered there could exist an explosive hazard in the air above the oil?

=====================================
(2B)+(2B)' ?

 

[electricpete]

If anyone has inputs this morning, it would be timely for me.

Referring to the drop in oxygen.,,
1 – it is distinctly different than historical levels.
2 – the drop in oxygen began when the jump in combustibles began.

Not believing in coincidences, we want to explore how/why the drop in oxyden could be related to the increase in combustible gases. Two alternatives I can come up with:

1 – LTC tank had obstruction that caused inability to suck in through the breather (but did not inhibit ability to breath out thru 1psig regulator/relief valve). This resulted in repeated vacuum operation which sucked gases out of solution (although nitrogen did not go down very much). Or perhaps interfered with operation of the vac switch somehow since it is an enclosed unit and sees forces depending upon external pressures. LTC does not have any pressure gage which would alert to clogs in the breather.

2 – Combustion of combustible gases sucked up oxygen. Perhaps this occurs when there is arcing going on.

Both of these sound a little far out to me, but interested if anyone has heard of either of these or any other ideas about the peculiar drop in oxygen.


=====================================
(2B)+(2B)' ?

 

[DanDel]

Hi 'pete. Looks like the tap change on 7/5 caused your increase in H2 and C2H2, which would be typical for internal arcing. It's tough to analyze DGA in a LTC since all the real bad gasses are caused by it's normal operation (switching contacts - arcing).

Your other combustibles are coming from a heating issue, I think. Since they are leveling off, you may be OK.

I think the O2 went into the CO and CO2 production.

Keep sampling to determine the TDCG rate, which should remain negative. You could always schedule a shutdown for an internal inspection. I would be more worried about the CH4/C2H6/C2H4/CO production at this point, indicating overheating.

 

[electricpete]

Thanks Dan.

I think the O2 went into the CO and CO2 production.

fwiw in this case the CO2 didn't go up, but the CO did.
I can accept that O2 and oil are input to a reaction and CO plus other gases come out of that reaction. It is related to what I called combustion, but that was probably not a good term.

Still, I'm not too sure about it. The reaction that produces the gases is likely localized to one small area of the compartment. I'm not sure whether it makes sense that it could suck up more than half the O2 in the entire compartment.

Have you ever seen this type thing (reduction in O2 accompanies increase in combustible gases) before?

Your other combustibles are coming from a heating issue, I think.

I tend to think it is a contact problem of some kind which caused both overheating and localized arcing at the contact. The reasons are:
1 - The CH4/C2H6/C2H4/CO (overheating) increase and C2H2 (arcing) increase all came at the same time.
2- C2H2 is not normal for this LTC where interruption occurs inside vacuum switch.

The decrease in O2 is not anything that worries me. It is something that I am interested to understand since it may be a clue about whatever is going on that caused this dramatic increase in combustibles.

We got another sample result drawn 8/10/11 - roughly the same, against suggesting it is stabilized.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Two corrections:
1 - I was wrong to say CO2 didn't go up. Both CO and CO2 went up. But the increase in CO and CO2 is far less than the decrease in O2.

2 - I was wrong to say CO or CO2 is product of decomposition of oil. C2H2, C2H2, C2H6 and H2 are products of decomposition of oil. CO and CO2 are (of course) only products of decomposition of cellulose.

Now a new thought. Let us try to test the theory that O2 is consumed in production of CO and CO2, perhaps during cellulosic degradation.[TT]
O2 CO C02
10641 185 2801
7597 447 2809
2821 450 3298
3707 454 3496
3235 431 3041[/tt]
PPM represents a volume fraction, which (based on ideal gas law) is proportional to moles of each gas. We had 250ppm increase in CO which might account for 125 ppm decrease in O2, as well as maximum 700 increase in CO2 which might account for 700ppm decrease in O2. If we put those two together, we might account for 825 ppm decrease in O2. But we have a 7,000ppm decrease in O2 (almost a factor of 10 higher). I realize some CO and CO2 may remain bound in the insulation or released to the the airspace, but it makes me wonder.


=====================================
(2B)+(2B)' ?

 

[electricpete]

So let me say there are 3 candidate theories for drop in oxygen:

THEORY 1 – LTC tank had obstruction that caused inability to suck in through the breather (but did not inhibit ability to breath out thru 1psig regulator/relief valve). This resulted in repeated vacuum operation which sucked gases out of solution (although nitrogen did not go down very much). Or perhaps interfered with operation of the vac switch somehow since it is an enclosed unit and sees forces depending upon external pressures.

THEORY 2 – Oxygen consumed during decomposition of oil. Although the gases that we normally associate with oil decomposition do not contain oxygen. Perhaps there are other created gases we don't know about during oil decomposition? Or perhaps it is something like combustion of produced gases in presence of arcing.

THEORY 3 - Oxygen converted to CO and CO2 during decomposition of cellulose insulation.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Theory 1 suffers from inability to answer the question: why didn't N2 decrease also?

Theory 2 and 3 do better on that question, but have their own questions discussed above.

=====================================
(2B)+(2B)' ?

 

[prc]

pete, I faced a similar case some 4 years back in a 100 MVA 220/66 kV transformer with resistor type tap changer on LV side.Tap selector was in the main tank and arcing contacts were in separate diverter switch chamber.

DGA of main tank oil was monitored for one year and the pattern of gas formation was exactly as reported by you.ie oxygen was coming down from 8550 ppm to 3154 ppm. Methane went up from 0 to 508 ppm, Ethylene went up 2 to 627ppm, Ethane went up from 1 to 235ppm.N2 was nearly constant ie 35232 to 33764 ppm. Transformer was in continuous operation for nearly 7 years after first energisation.On internal inspection, coking(pyrolitic carbon) was found in reversing switch contacts and also in some other contacts of tap selector.

It is true that coking is generally found in the current carrying contacts that are not operated. Some 25 years back, I saw coking in a series of identical transformers.It was only in the reversing switch contacts then.But in later case, it was found in othere contacts also.

So in your case, I suspect coking or misaligned contacts.This was triggered during the tap changing operation done on 2011-07-05 when reversing switch operated..The problem is continuing.Better to immediately take shut down, drain out oil from tap changer chamber and inspect contacts as situation may lead to contact melting and failure. Increase of thermal gases is quite rapid and immediate action is called for.The decrease in oxygen is due to coking action.It consumes oxygen to form carbon oxide deposits.

But I have no answer on why coking happens?I could not find any explanation in literature.

 

[Ceast]

See the attachment.

 

[Ceast]

Test contact resistance. When was the last time the LTC was opened up and inspected or repaired?

 

[DanDel]

'pete, I don't think you can perform an empirical chemical analysis based on the DGA values from an oil sample.

The general opinion seems to be that an internal inspection is needed. I would imagine you would find a 'cloud' of carbon around one of the contacts or connections inside, which is from the arcing/acetylene production. If you're lucky, you should find the weak connection causing the overheating by using contact resistance test methods as mentioned by Ceast. The key here is to not disturb anything until 'As-Found' resistance readings are taken for each connection. You should find the bad connection(s), then fix it (them) and then retest to verify the fix.

 

[backpack96]

I would suspect a defective contact. In some cases, a winding resistance test will help identify a problem. I would not expect to see C2H4 in a vacuum type tap changer if properly venting.

 

[electricpete]

Results of inspection showed two moving contacts and two stationary contacts damaged.

=====================================
(2B)+(2B)' ?

 

[electricpete]

I think there are only two moving contacts, so it should be "both moving contacts".

The stationary contacts damaged are the ones corresponding to the tap 11L. That is the location where the tapchanger sits 99% of the time.

=====================================
(2B)+(2B)' ?