Forecasting Seasonal Gas Demand and Storage Needs over Iran's 5th Development Plan Period Using Neural Network and ARIMA Approaches

High domestic natural gas consumption and major differences between demand in cold and warm seasons pose major challenges in Iran. The seasonal nature of demand for natural gas seasonality has created many problems for honoring export commitments, injection of gas into oil fields and meeting peak demand of domestic users. To respond to high demand for natural gas in the cold season the country has to store excess natural gas during the warm season for use in winter. Optimal storage capacity, in turn is dependent on a solid projection of future peak seasonal demand. This paper is prepared with the objective of projecting future peak demand and associated need for storage facilities. In this paper we use Neural Network and Seasonal ARIMA methods to forecast seasonal natural gas consumption over Iran’s 5th development plan period and calculate the requited storage capacity. The results show that neural network methodology has superior predictive power compared to the ARIMA method. The seasonal nature of demand for natural gas is expected to continue. The model results indicated that in addition to the already planned 4BCM/Y of storage capacity, the country needs to build an additional 10BCM/Y of gas storage facilities in order to effectively respond to peak demand for domestic consumption and to ensure optimal management of gas supply and demand.