Loss measurement for transformer in operation 3

[sykimk]

Hi!

Are there any methods to measure losses of transformer in operation or to calculate transformer losses by use of rating, measured voltage & amperes and nameplate data?
Please let me konw the method or any reference books.

Thanks in advance for your advice.

 

[alehman]

Volts and amps alone isn't enough information. You will need to measure the difference between the input and output with a watt meter. Power loss of course depends on the load on the transformer. For larger transformers the manufacturer should be able to furnish factory test data with no-load and full-load losses. The nameplate may indicate this information also.

 

[busbar]

Possibly IEEE Std C57.12.90-1999 …Test Code for Liquid-Immersed Distribution, Power, and Regulating Transformers, and IEEE Std C57.123-2002 Guide for Transformer Loss Measurement

 

[jghrist]

I wouldn't expect much accuracy in the loss measurement with the input and output wattmeter method. Typically, a 1000 kVA transformer would have 14 kW load loss at full load and 1.2 kW no-load loss, total about 1.5%. If the wattmeter measurements (measuring full load) were 1% accurate, you could measure anywhere from -0.5% loss to +3.5% loss. Consider actual input of 1015.2 kW and output of 1000 kW. Input measurement could be as low as 1005.05 kW and the output could be as high as 1010 kW. That's with full load. If the transformer were only 50% loaded, the losses would be only 0.47% and the accuracy would be far less.

 

[dpc]

It might be worth spending some time to try to come up with the transformer test report. The test data on losses will be far more accurate than anything you would be able to measure on-line.

I agree with jghrist - power transformers are incredibly efficient and any attempt at loss measurement will be seriously compromised by instrument transformer error and metering error.

 

[alehman]

I don't disagree that there may be accuracy problems. Anyone have any ideas on how to get better results? What method does IEEE 57.123 recommend? (I don't have it and don't plan to buy it).

 

[pqsl]

Look at "Measurement of Transformer Power Losses and Efficiency in Real Working Conditions" IEEE Transactions on Instrumentation and Measurement. Vol.40 No.2 April 1991.

 

[edison123]

You would need a very accurate, low pf (<0.2 pf) measuring wattmeters and accurate instrument transformers to measure the losses exactly. An OEM test report is the ideal source and trafo losses remain almost the same in their life.

Any particluar reason, you want to measure the losses now ?

 

[pqsl]

Another aproach is in this link

http://www.hvtechnologies.com/download/HIGHVOLT/7-5-1e.pdf

but it looks like expensive solution comparing to Voltage and Current difference losses neasurement method.

 

[ScottyUK]

Just a thought: Rogowski coils can be made to far better accuracy class than CTs, and may be a solution to the problem of measuring current very accurately.



----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[pqsl]

To get high accuracy for P(out)/P(in) method high accuracy current sensors are necessary say 0.01% or less and that might be difficult with any conventional current sensors. Therefore P(out)/P(in is) inappropriate method to use with equipment introducing large error. Even 0.5% accuracy will add up to 1%. In addition voltage transducers also introduce more error to the measurement.

 

[rbulsara]

It depends upon what is your puspose in knowing this data.

If for metering or cost, the morale of story is that the losses are so small that its within meter's error margin and it will not be worth measuring, as you could pay 1-2% more or less even if you have a 100% efficient xfmr and a less accurate meter.

If you need to know heat gain for sizing cooling or ventilation, the rules of thumb, 1.5 to 2% of the rating, is more than sufficient.

However in theory wattmeter measurement is only viable method, all you have to get is accurate meter/sensors.

The tranformer manufacrture can always provide with the no load losses and full load I2R losses, just ask them. (It is another matter they may also come with those data by typcial Watt measurement)

No load losses are constant and you can always intrapolate I2R losses at any load from 0 to full load as R remains relatively constant. (Purists: do not throw in change in R due to temp....not worth it).

In absence of data you go with rules of thumb, say for sizing cooling or ventilation...

 

[ScottyUK]

Hi pqsl,

As I understand it, Rogowski-based sensors can be built with errors in the ppm level, far better than any conventional CT I have seen. I think the limiting factor in this case may be voltage measurement - I'm not aware of VT's in a comparable accuracy class and I do not think anyone would dare use a resistive divider for this application.

This is an interesting exercise.




----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[scottf]

ScottyUK-

It is possible to make a CT just as accurate as a Rogoski coil. The trick is to tune the CT to the exact burden being placed on it. If one knows the burden ahead of time, a CT can be designed to be almost perfectly accurate. The limit on the accuracy rating is only limited by the level of accuracy of the calibration equipment.

Same goes for VTs.

The reasons CTs and VTs are generally rated by normal standard classes is because the devices must maintain an accuracy class over a range of burdens and therefore must be designed such that the low and high burden rating are within class.

 

[ScottyUK]

If you need to know about CT's, ask a man who makes 'em...

Thanks for the info - I figured it was theoretically possible, but I've never seen a CT better than Class 0.1. Now I know better!



----------------------------------

If we learn from our mistakes,
I'm getting a great education!

 

[scottf]

Glad I could help...

Generally, 0.05% is about the best that can offered, which is about the best you can calibrate to, since I believe you try to keep 10 times better accuracy with the calibration equipment compared to the test sample.