Direcionadores de custos de uma distribuidora de energia elétrica

Abstract

After the wave of liberalizing reforms in the electricity sectors of many countries, various studies and regulatory agencies applied models for benchmarking and cost functions for the distribution of electricity. A fundamental question to be defined in this context is which variables are determinants of costs in electricity distribution. This dissertation analyzes the major processes of electric power distribution, which are run on a regional basis, to assess the statistical significance of supposed cost drivers. The survey was conducted using data from Cemig D, which is the largest electricity distributor in Latin America in terms of the extension of the network and the number of consumers. A strategy of mixed methods of research was adopted, and made a sequential study beginning with interviews and meetings with company experts to understand the processes and activities required to deliver the service and to survey for potential cost drivers in each process . Subsequently, the data for the potential cost drivers variables were analyzed for its statistical significance in determining the costs. In the largest dataset we applied the method of Best subsets to obtain the best model. In other cases correlation analysis for the construction of models that best represent were applied. The selected models were estimated by ordinary least squares (OLS). For the low voltage maintenance costs results showed that the most important variables are the most common in the literature, which are the extension of networks, the number of consumers and consumption. However, the separation of these products by their urban or rural location is essential, since the production technologies in these environments are quite different. For the high voltage maintenance costs the model considers the total length of power lines as the independent variable. The proportion of the structures according to their nature proved to be significant, as the percentage of unpaved roads. For the maintenance of substations, the most significant variable was the power delivered, followed by decomposition of the number of transformers in the voltage levels, the average duration of interruptions and the number of large customers. For the cost of billing the number of consumers separated in urban and rural locations proved to be significant.