Predictive testing of single-phase dry transformer 4

[electricpete]

This is same question as :
thread238-87685

However it occurs to me I made it more complicated than necessary by mentioning ferroresonant transformer.
Let's just assume it is a standard dry type transformer.

On a low voltage single-phase transformer, we experienced failure from turn-to-turn short which was believed to be due to thermal aging. We want to figure out what kind of testing we can do (on-line or off-line) to have maximum confidence that the sister unit transformers will operate reliably and we can detect similar degradation ahead of time. What periodic testing of dry type transformers is appropriate if maximum reliability is desired?

Megger testing is a given. 500vdc.

TTR – (ok let's assume I can't do that test and move on).

Doble test?

* DC Hi-pot or step voltage test? I know this is fairly common on rotating machines but less so I believe on transformers. We did test one to 2kvdc with no failure. However we are not sure whether the test itself may degrade the reliability of the future reliability of the transformer (we’re not worried about failing a transformer during a test, just worried about passing the test and failing later as a result of the test).

Excitation current?
Winding resistance?
I think excitation current and winding resistance pose some challenges because on 3-phase unit we compare phases. In this case we would have to compare sister units or trend over time. Would it be beneficial?
Tear-down visual inspection and re-assembly (somehwat intrusive and may introduce problems).

Infrared?

Any thoughts and opinions at all would be appreciated. I have some experience testing large oil-filled power transformers but none at all on the small transformers. Also interested in any relevant links or standards references.
Thanks.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[jbartos]

Suggestion: A no-load transformer starting current signature can be compared to subsequent ones taken at regular time intervals. Any deviation in the signatures can be analyzed by comparisons. E.g., the higher current would mean that the parallel branch in the transformer electrical equivalent circuit is somewhat deteriorating. Additionally, transformer leakage current signatures also reveal the transformer insulation deterioration.

 

[busbar]

Peter, this is kind of a frustrating problem given the “features” of ferroresonant transformers that would seem to mask a shorted-turns problem.

The NETA 2001 Maintenance Testing Spec §7.2.1.1 Transformers, Dry-Type, Air-Cooled, Low Voltage, Small isn’t a lot of help. Suggested there is primarily {1) resistance measurements through bolted connections, (2) insulation-resistance tests winding to winding and each winding to ground, with calculation of dielectric absorption ratio and/or polarization index, and (3) turns ratio tests.

Their normal excitation current and unloaded heat dissipation are high, making baseline data not much to go on. DC resistance is not suggested by NETA as a test for windings—just bolted connections, probably because of the ‘needle in a haystack’ character, and significant change in resistance from temperature variation. Turn ratio is very ‘dynamic’ for a ferro. (For their ‘wonderful’ regulating capabilities, they seem to be more of a PITA for me.)

 

[alehman]

Periodic IR may be worthwhile, particularly if you suspect hot spots, perhaps due to a systematic manufacturing problem.

NETA adds power factor, winding resistance, overpotential and excitation current for medium voltage transformers (7.2.1.2).

Surge testing may be an option. This is typically used for insulation evaluation of motors. Connect windings togather in pairs (1-2,2-3,3-1), inject an impulse on each pair and watch the resulting ring waveform. If there are differences in the insulation, it will show up as offsets in the waveforms. Not sure how well it might work on transformers.

I wonder if partial discharge monitoring would be useful.

One other rather generic statndard -
IEEE standard test code for dry-type distribution and power transformers, IEEE Std C57.12.91-1995
Abstract: Methods for performing tests specified in IEEE Std C57.12.01-1989 and other referenced standards applicable to dry-type distribution and power transformers are described. This standard is intended for use as a basis for performance, safety, and the proper testing of dry-type distribution and power transformers. This standard applies to all dry-type transformers except instrument transformers, step-voltage and induction voltage regulators, arc furnace transformers, rectifier transformers, specialty transformers, and mine transformers. The following tests are covered: resistance measurements; polarity and phase relation tests; ratio tests; no load losses and excitation current; load losses and impedance voltage; dielectric tests; temperature tests; short-circuit tests; audible sound level measurements; and mechanical design tests.

 

[CarlPugh]

The transformer that failed should be torn apart and analyized.

If the insulation (other than at the short circuit) is detoriated then there is a thermal problem.

If there are loose turns, then the transformer may not have been impregnated properly.

If there is a corona problem, then there should be places in the insulation where the insulation is thin. One way to check this is to hold the insulation up to a light and look for thin spots. This is a difficult inspection to do and should be done very thoroughly.

To answer your question, I am now aware of any tests that can be performed on a dry type transformer that will give an accurate indication of how long the transformer will operate.

A visual inspection is sometimes helpful in locating hot spots.

The factory usually does a double induced voltage test. This is generally not done at the user'e site.

 

[electricpete]

Thanks for the great responses.

alehman - what voltage is recommended for dc hi-pot? (you mention mv transformers - I assume hi-pot could also be applied to low voltage 120v transformer)

All -

If transformer we opened for visual inspection is it likely that tank circuit (of ferroresonant transformer) can be easily disconnected for performance of turns ratio test?

I believe surge test really only tests the first few turns. At least that's what we're taught for motors. Does the same logic apply to transformers? If it does it seems like it is a little extra risk (of creating a failure) and effort without testing a significant portion of the transformer.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[alehman]

NETA is vague on the subject over overvoltage tests on transformers:

"AC overpotential test shall not exceed 75 percent of factory test voltage for one minute duration. DC overpotential test shall not exceed 100 percent of the factory RMS test voltage for one minute duration. The insulation shall withstand the overpotential test voltage applied."

The value of surge testing certainly decreases with the number of turns. I suppose it depends on the parasitic capacitance and resistance of the windings. For low voltage windings there are relatively few turns.