Transformer Turns Ratio Test, increased then decreased

[MrJLH]

While looking at a 2000kvA 15kV primary to 480V secondary NETA turns ratio test from 2018, the testing data indicates a percent error above 0.50%. The transformer was retested in 2023 and now the percent error is around 0.23% for all phases. NETA states that the acceptable percent error is 0.50%.

Any insight into why the decrease? Perhaps a bad test in either 2018 or 2023?

Thanks

 

[junkhound]

Insight? Do a quick reverse engineering xfmr design for min # of primary turns, calculate turns ratio for 480V. Calculate what a 1 turn error would show in percentage. Highly unlikely for a turns ratio error. Anything less than a 1 turn error is primarily due to leakage flux . The transfromer in question sure did NOT change turns ratio. Thus, leakage flux changed.
So, leakage flux is the 'insight'. Was a cover panel removed for the 2nd test, a slight leakage path, removal = less percent error. Xfmr making less hum now? = change in coupling factor.

 

[David Castor]

What's the accuracy of the test? Was the same instrument used? I doubt that this difference means anything.