Forecasting models for the electricity consumption of the cement industry in Brazil

Abstract

The consumption of electric energy in the Brazilian industrial sector has been investigated over the past few years. This interest is related to sector development, energy planning, and energy efficiency. Thus, forecasting models are important for decision-making. The objective of this work is to compare different forecasting models applied to monthly data on electric energy consumption in the cement industry in Brazil. Therefore, the Holt-Winters method, the Seasonal ARIMA model, the dynamic linear model, and the autoregressive neural network model were used. Through the considered accuracy metrics, the Seasonal ARIMA model showed the best predictive performance for the analyzed period.