Stray flux overheating transformer case?

[oldfieldguy]

Anybody want to comment?

A little history. Original transformer was installed in 2006. It failed catastrophically in 2010. The manufacturer said the cause was an external fire, although I pointed out this discoloration on that transformer's case, away from the obvious fire damage.

We replaced the failed transformer with an identical unit from the same manufacturer. A site technician called me about this obvious issue.

Facts: Neither the top oil nor winding temperatures are noticeably above normal. TCG was 0.34% Dissolved gas analysis shows that in the past two years CO went from 251 to 542 ppm and CO2 went from 1493 to 3065 ppm.

Transformer feeds a unit with an 7000 horsepower motor on a Siemens GM-150 VFD. When the motor is running, the transformer is about 80% of its 8 MVA base capacity. When the motor is not running, the load on this transformer is negligible.

The standard battery of tests (power factor, turns ratio, winding resistance, etc.) were normal a year ago.

I think I'm looking at an incipient failure. I'd like other opinions.

old field guy

 

[bacon4life]

Have you done thermal imaging on the transformer tank? At a factory heat run test on one of our transformers, I was very surprised that some of the tank wall near the bottom of the transformer was actually several degrees hotter than the top oil temperature. Turned out the manufacturer had installed stray flux shielding on the inside of the tank that ended about 2 ft above the bottom of the transformer.

 

[cranky108]

Not that I know the cause, but one problem with three legged cores is stray flux will travel through the transformer case, and cause heating of the case.

As you stated the load is primarly a large VFD system, and might that VFD present a harmonic componet? If there are simular units compare with them. But if not you may be looking at stray flux due to harmonics.

And as many VFD's are designed to filter harmonics, that filter may have failed, and no one knows about it.

Just a thought.

 

[oldfieldguy]

b4l-

Right now this transformer is 140 miles west of me. The technician on site is shooting it with an IR thermometer gun. He hasn't found much: "Bottom 86*F, Middle 93.5*F, Top 95*F cooling fins pretty much read the same." They're going to refine that with some hand-drawn circles so they measure the same spots.

This is an 8/10MVA 34.5-13.8 kV, Y-Y transformer. When my main compressor downstream is running, the transformer is at 75-80% of base MVA. When that motor is off line, the only load is a 13.8 - 480 , 500 kVA transformer, so the load is essentially zero.

I expect to get over there next week with an infrared camera.

I was at the factory when they untanked the previous failure. This tank is a simple metal plate construct with no apparent provisions for flux mitigation. $$$, you know...

The attached picture is from the failed transformer. Not apparent from the picture is that the windings behind that charred board are pristine, with NO signs of damage or heating, and the side of the barrier board nearest the winding show much less evidence of heating.

thanks


old field guy

 

[jghrist]

It looks like a 3-legged core. Y-Y transformers with 3-legged cores have no zero-sequence flux path and are prone to tank heating. That's why Y-Y distribution padmounts are almost always 5-legged cores. Maybe the tank heating occurred during a single phase fault but doesn't occur with the normal balanced load.

 

[prc]

First map the temperature on tank(by direct measurement by thermometer).Find the exact locations where the heatingis.

Did the press board barriers where you noticed charring was touching the inside of tank surface?(must be)The charring found on lal three phases or only one phase?

In three phase 3 limb units, normal levels of zero sequence load will not cause excessive heating. But single phase loading can over heat tank surface, but not line to ground (LG) faults.

Check with VFD supplier any chance of single phase loading, or excessive 3rd harmoics during operation.Any way tank heating will not lead to a failure unless winding is also getting heated locally.

 

[oldfieldguy]

Updating my original findings:

I and an engineer co-worker visited the site.

The attached picture shows a combination visible/IR shot of the transformer in question. At the time this shot was taken the transformer was at its low-load condition, with primary current in the single-digit range, and had been in this state for days. Ambient temperature was around 98F. Temperatures on the picture in are in degrees F as well. I find it disturbing that the temperatures on the end wall start out cooler at the bottom, get warmer, then cool slightly, before reaching to case top, where one would expect the highest temperatures to be found.

This is at 'no-load' state. In the vicinity of that 108.9 reading there is an installation of standard Sealtite vinyl-jacketed flex conduit held tightly against the case by a clamp. where that flex touches the case, the vinyl has not only melted but has charred and bubbled, as well. I understand that one needs 150 degrees C to decompose vinyl like this.

Reinforcing the charred vinyl, the paint on the case wall has heated enough to discolor to a light brown instead of the standard grey shade.



old field guy

 

[oldfieldguy]

Here's a visible light view of the paint discoloration and the charred vinyl jacket of the flex.

old field guy

 

[prc]

Since the bottom is getting heated up the issue is V/f is high some how.Then the flux will overflow from core and heat up the tank and inner winding.Earlier failure was in inner winding? Heating /burning innermost edges of winding? Disconnect VFD and check no-load operation.Better start from ambient condition so that tank heating can be found out easily as oil will still be cool.

 

[oldfieldguy]

prc-

The manufacturer's 'failure analysis' of the first transformer (2006-2010 - RIP) opined that failure was due to an external fire. This is partially true - the fire started when vaporized oil left the pressure relief valve and ignited in the open air.

I hate to tell my management that we just need to allocate funds for a new transformer every three or four years. They get silly about things like that.

There were no damaged windings in the original failure.

The recent pictures were taken with the VFD (Siemens GM-150) off line. The only load downstream of this transformer besides the VFD is a 300 kVA 13.8kV-480/277 transformer. When the VFD is not on line, that little transformer is supporting a few auxiliary loads like lighting and HVAC, around twenty amps at 480 volts, so the 8 MVA 34.5 transformer I'm concerned with is at essentially zero load. This is the condition at the time the photo of my 23 Jul 14 15:08 message was taken.

Your idea about starting at ambient is a good one, but now I have to tell the station to take that transformer offline for the weekend, an act that will have them running the backup generator.

Supposedly a manufacturer representative will meet with us on Friday, so I'll wait and see what he adds to the discussion.

thanks

old field guy

 

[stevenal]

Are the transformers K-rated? Review Cranky's and prc's posts above.

 

[tinfoil]

As has been pointed out already, you seem to have a transformer with a 3-legged core and no delta tertiary.

To quote "Power Transformers: Principles and Applications By John Winders" for transformers of this design (bolded by me):

"In large power transformers, third harmonic flux and stray 60 Hz flux from unbalanced voltages may leave the iron core and enter free space inside the transformer. This induces currents in the internal metal parts of the transformer and may cause severe localized overheating".

I know 8 MVA would not normally be considered a 'large power transformer', but this sounds a lot like what you are seeing here. Your motor load may be very balanced, but if there is any imbalance in the supply voltage or 'triplen harmonics' present, the residual flux after summing the 120°-spaced fluxes has to go SOMEWHERE.

 

[oldfieldguy]

tinfoil-

Thanks for the input. I do have a very balanced load here. When we use the VFD, we feed 13.8 kV into an isolation transformer that delvers 12-phase 2200 volts into the VFD's rectifier section. There's no place for unbalance there. One or more rectifier failures will take the drive down on its own protection in a few cycles, far faster than any neutral overcurrent element might act. I've seen that happen with other faults. The Siemens VFD trips much faster than any overcurrent protection.

I'm still waiting to drag a manufacturer rep out there to see what he says.

old field guy

 

[prc]

You are confirming that tank heating is there even without VFD.I have seen such problems with issues /defect in VFD.Do you find extra noise in transformers.Please check v/f both on HV and LV sides of the transformer to make sure there is no over excitation.

 

[oldfieldguy]

The feed to the transformer is directly from the utility grid that also feeds the local neighborhood. It is essentially a pure 60-Hz source with very few voltage excursions of short duration. If we had a poor power source, the protective elements of the VFD would catch them faster than any other protection element. Early on in the installation of this equipment, we had power quality issues and worked with the utility company to fix them since not only did the VFD trip on departure from nominal voltage, but it also tripped on rate of change. Since those early issues, we seldom have these problems.

The LV side does not present conditions for overexcitation. Under the low-load condition, my secondary voltage at the 480-volt transformer reads 492 volts.



old field guy

 

[cranky108]

Sounds like harmonics.

 

[squeeky]

Is the transformer in parallel with another one? If so. lift one earth.

The laminations should be earthed to the case. Have you put a clamp tester on the earth or star point? Do you have a power quality analyser? try look at your harmonics. If one is not availible try one of the inexpensive energy meters (Schneider) or a power factor controller which should have some harmonic display. I would not be happy with the big boards for the oil circulaion and cooling. Cut holes with a hole saw. I've never seen that before. The coils look fine.

What other gases were shown on your DGA?

 

[oldfieldguy]

Squeeky-

The transformer is not in parallel.

We haven't checked the core grounding yet. We may be in position to do that on Tuesday. I do not expect to find a problem.

The only gas besides CO & CO2 was hydrogen rising a bit, but well below any threshhold values. I suspect the hydrogen increase to be due to the internal heating that produced the CO & CO2.

I will also check the harmonics, but in its unloaded condition, absolute values of all harmonics are so low as to be undetectable. If the factory rep doesn't find anything on his visual inspection, we'll fire up the VFD and look at harmonics when the transformer is loaded.

old field guy

 

[Mbrooke]

Y primaries are a poor choice for a transformer, even on 5 leg cores. If ferroresonance is a low concern or mitigated by design a delta primary should always be used. I highly discourage fuse protection on Y grounded primaries as well since a blown fuse will induce currents into the tank.