Dry Type Transformer Temperature Rise 1

[gordonl]

I've recently had a 12MVA dry type ventilated transformer fail. The transformer was a 34.5kV to 4160V which fed a 4160V starter lineup which feeds several pump motors. Transformer is 150C rise design, insulation class 220, manufactured in 1996. The transformer was part of a unit substation with a fused primary disconnect switch, and a secondary main circuit breaker. There are no drives, soft starters, or other sources of harmonic currents on the bus. The bus was extensively monitored, and the max voltage harmonics on the primary were 1.3% THD and current harmonics were less than 1% THD. The transformer load was 7.7 to 10.8 MVA running with a maximum 15 minute average of a little under 9MVA. The primary volatge seldom exceeds 103% of the 34.5kV nominal. Room ambient max 30C.

There was no diferential protection, only primary and secondary overcurrents and the fuse in the disconnect. The primary timed overcurrent covers 6 substations, so has a relatively high setting.

The transformer suffered interturn secondary winding fault, after failure there were no grounds present on the primary or secondary but when the center winding was dismantled the low voltage windings closest to the core were cooked. The actual wire insulation was baked and falling (flaking) off, and the fiberglass inner tube was reduced to only the fiberglass, particularly at the top of the winding. (No epoxy)

We have a 10MVA and 7.5MVA which were purchased at the same time as this transformer, so a thermographic study of these transformers was done and we found that they were running with the core above the center winding over 220C. (Upto 239C measured)

We beleived there is a design flaw in these transformers so we sent the 10MVA unit to the manufacturer for a heat run test. (not originally preformed) We witnessed the tests and found the following results:

111% excitation no-load, center core 1" from winding top reached a steady state temperature of 186C, and the center coil low voltage winding on the first wrap at 15" from the top of the coil reached 98C. (Amb. 24C) Low voltage excitation loss rises 37.9C (This was calculated by the manufacturer using the measurements of winding resistance after shutdown)

full load test, (secondary shorted, full load primary current) center core 1" from winding top reached a steady state temperature of 152C, and the center coil low voltage winding on the first wrap at 15" from the top of the coil reached 202.6C. This RTD peaked at 209C, the excitation tests were done first, with the load test done second, I figure it reached 209C as the core cooled from the excitation test, and the windings increased form the load test. (Amb. 21.8C) Low voltage load loss rises 105.2C. (This was calculated by the manufacturer using the measurements of winding resistance after shutdown)

Combining the average temperature rises from the resistance measurement method using the ANSI formula gives only a 128.2C rise which is within the 150C rise design.

Now for the question, is there a way of combing the RTD measurements from the excitation rise and load rise tests to prove the hot spot is exceeding the 220C rating of the transformer? It would seem to me that the peak reading of 209C corrected from the 23C ambient to 40C ambient is out of limits already, as well 202.6C at 21.8C ambient.

I tried combining the LV winding RTD values using the ANSI formula for combining the average temperature rises and calculated a rise of 219C. Is this calculation valid?

The manufacturer has requested that the transformer be sent to a third party test lab for full voltage, full load testing at our expense.

Thank You

 

[jbartos]

Suggestion/comments marked by ///\\\:
There was no differential protection,
///The 87T differential relay is needed for the internal phase to phase fault protection.\\ only primary and secondary overcurrents and the fuse in the disconnect. The primary timed overcurrent covers 6 substations, so has a relatively high setting.
///The protection against the external overloads and overcurrents shall not exceed the transformer thermal curve or limit. A protection against voltage surges, e.g. arresters, appears to be needed.\\The transformer suffered inter-turn secondary winding fault, after failure there were no grounds present on the primary or secondary but when the center winding was dismantled the low voltage windings closest to the core were cooked.
///Most probably because of lack of 87T relay.\\Now for the question, is there a way of combing the RTD measurements from the excitation rise and load rise tests to prove the hot spot is exceeding the 220C rating of the transformer? It would seem to me that the peak reading of 209C corrected from the 23C ambient to 40C ambient is out of limits already, as well 202.6C at 21.8C ambient.
I tried combining the LV winding RTD values using the ANSI formula for combining the average temperature rises and calculated a rise of 219C. Is this calculation valid?
///The temperature readings cannot be combined arbitrarily. The industry standard, which the manufacturer follows, has to be followed to have the test reconstructable and objective for you as well as for the manufacturer. The hot spot shall not exceed the insulation class temperature limit. This generally means, if one is raising the ambient temperature toward 40C at transformer rated parameters, and the hot spot is projected to be 220C (it may be one temp. sensor inside the winding) at 35C, then the product is not meeting the standard. The Insulation Class 220 consists of materials that can be capable of operation at 220C. This means that ambient temp 40C and temp rise 180C at individual specific spots are permitted and 220C is temp limit no matter whether it is a hot spot or continuous homogeneous temperature throughout windings. However, some design margins, e.g. 10C or so to allow for some inter-turn shorts, are highly advisable for the type of equipment you are testing. Also, the dry type transformers have different testing procedures from the liquid filled transformers. See for example IEEE C57.12.01-1989 IEEE Standard General Requirements for Dry-Type Distribution and Power Transformers Including Those with Solid Cast and/or Resin-Encapsulated Windings.\\

 

[gordonl]

Thank you for the reply, I would like to add the following comments/questions to your response.

I realize overcurrent protection won't pickup interturn faults until after the fact, but I'm approaching this problem from the point that the overheating caused the fault and not vice-versa. When the facility was installed differential was not included, and is unlikely to be added any time soon.

Because of the high temperature readings on the other transformer cores with the thermographic camera, and the high readings during the test at the factory I don't think 87 would have prevented it.

Do the test standards require any monitoring and checking of the RTD temperatures during a heat run test? Are there any standards that do require this test?

Thank You,
Gord

 

[jbartos]

Suggestion to the previous posting marked ///\\\:
Because of the high temperature readings on the other transformer cores with the thermographic camera, and the high readings during the test at the factory I don't think 87 would have prevented it.
///Yes, if you know that the transformer has a thermal problem because of its engineering, design and manufacturing, then the thermal protection has a high priority. The 87T will protect the transformer from more serious damage due to interturn shorts which can cause heat generation throughout the windings in the transformer instead of at one heat spot caused by the deficient production. However, considering the transformer cost, the 87T relay is normally the standard protective device since the interturn shorts happen; especially, when the transformer ages.\\Do the test standards require any monitoring and checking of the RTD temperatures during a heat run test?
///Yes, the posted standard states in 5.11.3.1. The average winding-temperature rise of dry-type transformers above their ambient temperature, when measured by the resistance method and tested in accordance with the applicable provisions of ANSI/IEEE C57.12.91-1979, shall not exceed the value in Table 4A.\\
///Yes, that is what they are for. How would the manufacturer satisfy the buyer?\\Are there any standards that do require this test?

 

[gordonl]

The transformer passes the average temperature rise test using the resistance method despite the fact that the RTD's are overtemperature. The resistance method does not involve the RTD's but measures the resistance of the winding before and after the heat run test to calculate the average temperature rise. These values have no relation to the RTD readings.

 

[busbar]

FWIW, an outline of ANSI/IEEE C37.91-1985 IEEE Guide for Protective Relay Applications to Power Transformers is at:

http://standards.ieee.org/reading/ieee/std_public/description/relaying/C37.91-1985_desc.html

Section 5.2 states: “Current differential relaying is the most commonly used protection for transformers of approximately 10MVA …and above…”

 

[gordonl]

Thank You for the link. I am not trying to argue the merits of differential, the mill was installed turn-key without, and I've no money to change it. This transformer is likely marginal on the economics for differential protection given it's total value of less than 100K US.

But my question at this point would go to anyone farmiliar with transformer test procedures and requirements:
Is there a standard for proving that the hot spot temperature in a dry type transformer is no more than the rated margin provided between the insulation class (220C) and the average rise of 40C amb + 150C rise = 190C.

 

[jbartos]

Suggestion: Standards do make distinctions between the average temperature and the hot spot temperature.
220C is related to the hot spot temperature and it may bear the temperature rise 180C at 40C ambient, 190C at 30C ambient etc. at the hottest spot of the winding conductor and insulation. 220C is the insulation temperature limit. This holds true for other electrical equipment, e.g. motors, etc. The related industry standard covers insulation temperature in essence. However, when applied to the transformers, the transformer standards will refer to this one too.
The hottest spot location may vary from the transformer to transformer.
The average temperatures are lower and not related to one hot spot. The hot spot temperatures, e.g. 220C, exceed the average temperatures.
The 220C-insulation temperature limit at the hottest spot may be created at the conductor at 220C caused by 80C max limit of the transformer iron core. This would result in the conductor temperature rise 220C-80C=140C above the transformer iron core temperature limit.
The standards do not tie that this way within their small or condensed space. Therefore, some standard interpretations are needed. All this content is from the ANSI/IEEE referenced standards and their inside cited references to other standards (i.e. continuing saga that is normal to make standards thin).
If you need some specific statment from the standards, not explicitly appearing in the standard, then a registered professional engineer/firm is in a capacity to provide it for you. They will probably charge for it some small fee.

 

[gordonl]

jbartos,

Thank You for your time and insight, it's appreciated.

 

[djs]

If I recall correctly, the IEEE standards suggest the highest temperature point will be the middle coil, at the top of the winding, next to the core. This is determined by basic thermodynamics. An RTD placed at this point (on the coil) should read close to the max temperature. The actual max temperature will be actaully inside the coil itself. On many dry transformers an rtd is embedded at this location for temperature monitoring and alarming.

 

[electricuwe]

gordonl,

as requested in an other thread I would like to give your my comments on this problem.

My personal experience is only related to rectifier transformers used for electroplating, electrolysis or induction heating equipment, but I guess I can also give some helpful hints to your problems with transformers not significant loaded with harmonics.

It is obvious that top of the center coil is the hot spot in a cast resin three phase transformer. But beside the thermal aspects already discussed in this thread there is a further issue that has to be considered:

At the end of the coil there is a significant radial component in magnetic field, which penetrates the conductors of the winding. Depending on the dimensions of the conductors this field causes eddy current losses in the conductor. For rectifier transformers these losses are very important because these losses increase with the term Sum(Square(h*Ih)) which is also called K-Faktor.

But for a cast resin transformer of 12MVA these additional losses caused by eddy currents may be important even if the transformer is loaded only with sinusoidal current. Some manufacturers use continously transposed conductors to cope with this problem.

The eddy current losses are also a topic in liquid filled transformers ,but usually the risk of overheating in a liquid filled transformer is much less severe. The oil can move away the heat generated in small areas much more effective than heat conduction in a cast resin transformer.