Data-Driven Methodologies to Assess Distribution Transformers Losses: A Case Study on a University Campus

Abstract

In 2021, the Brazilian Electricity Regulatory Agency (ANEEL) identified that Distribution Transformers (DTs) accounted for one-third of the US$1.75 billion annual costs due to electrical losses in Brazil. Therefore, considering the importance of the financial burden of these losses, this paper presents two data-driven methodologies devoted to estimating DT losses. The methodologies differ when considering the power factor angle of data measurements at the DT low-voltage terminals and the DT windings' connection type. Besides, the approaches are assessed considering simulated and actual data obtained from smart meters installed in DTs at the UNICAMP campus. It is disclosed that errors among the methodologies can reach up to 190% if the proposed correction factor is disregarded. However, further studies show that these errors can be restrained by around 5% when applying the developed factor, reaching more consistent results. Lastly, the paper revealed that estimation errors are influenced by the time step resolutions of the data measurements, particularly for overloaded transformers.