mapa del sitio home contacto

Automatic clustering of Electricity Distribution Utilities using Multivariate Data Analysis
Nombre del Congreso
XII Symposium of Specialists in Electrical Operation and Expansion Planning – SEPOPE
Río de Janeiro, Brasil
Fecha: 2012
In recent years, the need of developing new methodologies to obtain tariffs which take into account incentives to distribution utilities has come up. One option is to determine maximum prices based on comparison with efficient utilities that present related technical and administrative characteristics (i.e. Yardstick Competition Scheme), which must be determined by the regulator. Since it is not useful to carry out an individual comparison, the aim is to determine groups of distribution companies which allow characterizing utilities according to those parameters that determine their distribution costs while reflect their technical and administrative efficiency. This paper presents a novel approach for addressing the problem of how to structure the distribution utility groups according to similarities in their characteristics (variables). The stated proposal applies Multivariate data Analysis techniques, including Principal Component Analysis (PCA) and Cluster Analysis. The method is applied to the twenty electricity distribution companies of Ecuador, which have their own characteristics associated to their size, load, and technical and administrative way of being operated. The aim is to determine the adequate groups which allow characterizing the utilities according to specific parameters such as assets, number of customers, service area, load, losses, among others. A total of seventeen variables have been used to form the data matrix. First, univariate normalization is carried out to the data matrix in order to avoid the influence of different measurement units of original variables. This constitutes an important aspect of the proposed method. After that, PCA is applied to the normalized data matrix in order to reduce the dimensionality of the original data, maintaining as much as possible the variation presented in them. PCA reduces the problem from seventeen variables to only five principal components (PCs) and also allows determining the weight of the most important variables in the characterization of electricity distribution utilities. The results of the analysis constitute the groups of utilities, set according to similarities that consider, at the same time, the physical size and the technical and administrative efficiency of each firm, which will be the base for the tariff calculation under the scheme of Yardstick Competition.
Cepeda Campaña, Jaime Cristóbal
Furlan, Norma Lucía
Tascheret Graffigna, Carlos María
Andreoni, Alberto Mario