The identification of systematic error in determining the average temperature of power transformer insulation obtained by comprehensive analysis of an authentic on-site condition assessment database

The paper describes an analytical approach adopted to improve the insulation condition evaluation procedure. Electrical testing of power transformer insulation may provide reliable information about its state and aging grade. Inconveniently, the majority of dielectric parameters used in this evaluation are temperature dependent; hence, the accurate measurement of the temperature is decisive for the final assessment result. The on-site tested transformers are usually put off operation and disconnected few hours before the tests initiate and are still slowly cooling down during testing. Non stable average insulation temperature may lead to a considerable systematic measuring error. Aiming to avoid this, an authentic several years collected database was taken and a suitable emendation method applied.