Forecasting Natural Gas Consumption using ARIMA Models and Artificial Neural Networks

Abstract

The production and distribution function in the natural gas systems are performed by the producer and distributer companies, respectively. The producers companies are responsible by attend the demands of the local distributor companies and the distributor companies are responsible by to lead the gas to final consumers. Daily the producer adjusts its production capacity considering the availability of transportation pipelines, gas pipelines and demands from consumers. In this context the distributor companies are forced to perform, in the early hours, the programming the total gas volume to be consumed in whole day. Considering this problem, the present study examines three different approaches for forecasting gas consumption: using time series models Autoregressive Integrated Moving Average - (ARIMA), Artificial Neural Networks (ANNs) and a Hybrid Methodology of these two techniques. As a study case were used the informations of a distributor company of Brazilian northeast.