Residential Load Clustering Contribution to Accurate Distribution Transformer Sizing

Abstract

Data mining on customers' consumptions information, its clustering, and feature extraction are found to be part of the necessary studies for distribution system planning such as demand-side management, loss reduction, electricity pricing strategies, etc. Therefore, accurate determination of the customers' consumption pattern has an essential impact on the planning studies of the distribution company. Optimal distribution transformer sizing (ODTS) is one of the critical planning studies in a distribution company which can significantly contribute to the minimization of the distribution transformers total owning cost. In the ODTS planning process, the first step is to evaluate the transformer load curve over the planning time horizon. In this paper, we present a method for analysis of residential customer smart meter data to develop accurate normalized daily load curves for each pattern of consumption. Then, an ODTS software is implemented to calculate the optimal distribution transformer power rating. Two case studies are employed to compare the ODTS software results using the estimated transformer load of this work, versus using the peak load of customers connected to the transformer. This comparison showed that ODTS output in the former case results in a reduced power rating of the distribution transformer and also a reduced total owning cost (TOC).