Detection of forced cooling faults in power transformers based on winding temperature indicator and load data

In this paper, time series decomposition and thermal models are used to identify WTI issues; specifically an absence of sensor load dependency and detecting when the WTI is tracking top oil. Following this, two different transformer thermal models are used to calculate WTI temperatures which are then compared to WTI measurements under forced-cooling conditions to identify cooling issues, especially a reduction in cooling performance. A discussion of the differences between two different transformer thermal models, IEC 60076-7 and one purposed by Susa, is provided