Abstract of Thesis presented at COPPE/UFRJ as a partial fulfillment of the requirements for the degree of Doctor of Science (D.Sc.)

The Employment of Spatio-Temporal Aproach in Natural Gas Demand Forecasting

Jorge Henrique de Castro

March/2012

Advisor:

Alexandre Pinto Alves da Silva

Department:

Eletrical Engineering

      Following the goal of diversifying the energy generation matrix, natural gas has played an important role in recent years. Therefore, its distribution network planning is a critical factor for success, since natural gas demand exists only if the ducts reach consumers. Considering the particularities involved in the expansion of a natural gas distribution network and aiming at mitigating the impact of uncertainties, scenarios are created to support decision making. These scenarios take into account a wide set of exogenous variables. In addition to that, the spatial analysis of the gas pipeline network, associated with socioeconomic variables and consumption, improve business evaluation, and, consequently, the planning and decision making on sales, engineering, and operation. However, it can be noticed that the natural gas demand forecasting has been primarily based on temporal information, especially by using time series econometric models. Only a few researchers have tried to embody the spatial dimension in the modeling process. Therefore, this thesis, based on the local distribution of natural gas, introduces the application of the space-time perspective in the development of natural gas demand forecasting models. The history of natural gas consumption and socioeconomic spatially distributed data are used as input variables, investigating the effect of the spatial dimension regarding space-time representation and spatial dependency / heterogeneity.