Effectiveness of Using Simplified Methods to Estimate Transformer Loss of Life

Transformer loss of life is an important factor for the asset owner / operator to consider as it can be used in asset replacement planning and can help reduce the likelihood of unanticipated failures. Loss of life for a transformer is correlated strongly to the hottest spot temperature, which is a function of the loading on the transformer and ambient temperature.
Accurate measurement of transformer load and the ambient temperature is not usually available to owners / operators of distribution transformers (due to cost). Instead, assumed profiles are utilised, and load profile simplifications such as the two-step calculation provided in IEEE Standard C57.91 are employed. This work looks to establish the applicability of such load profile simplification methods (‘root-mean-squared’ and ‘two-step’ methods are considered), as well as the viability of averaging load and ambient temperatures into thirty-minute profiles from original minute-by-minute profiles. The analysis presented is conducted on load profile data of 81 installed distribution transformers in the United Kingdom for a period of 137 days.
It is shown that, as a general rule, the simplification of the load profile to a ‘root-mean-squared’ or a ‘two-step’ profile does not appropriately match the calculation on a minute-by-minute basis. It is noted that only 11.9 % of the root-mean-squared and 12.2 % of the two-step model results fall within ±10 % of the original calculated value. Averaging of loads and ambient temperatures to thirty-minute profiles saw a larger portion of results matching the original, with 30.2 % within ±10 % (rising to 55.7 % within ±20 %).