Forecasting models for daily natural gas consumption considering periodic variations and demand segregation

Due to expensive infrastructure and the difficulties in storage, supply conditions of natural gas are different from those of other traditional energy sources like petroleum or coal. To overcome these challenges, supplier countries require take-or-pay agreements for requested natural gas quantities. These contracts have many pre-clauses; even if they are not met due to low/high consumption or other external factors, buyers must completely fulfill them. A similar contract is then imposed on distributors and wholesale consumers. It is, thus, important for all parties to forecast their daily, monthly, and annual natural gas demand to minimize their risk. In this paper, a model consisting of a modulated expansion in Fourier series, supplemented by deviations from comfortable temperatures as a regressor is proposed for the forecast of monthly and weekly consumption over a one-year horizon. This model is supplemented by a day-ahead feedback mechanism for the forecast of daily consumption. The method is applied to the study of natural gas consumption for major residential areas in Turkey, on a yearly, monthly, weekly, and daily basis. It is shown that residential heating dominates winter consumption and masks all other variations. On the other hand, weekend and holiday effects are visible in summer consumption and provide an estimate for residential and industrial use. The advantage of the proposed method is the capability of long term projections, reflecting causality, and providing accurate forecasts even with minimal information. •We predict natural gas demand for residential/industrial areas on a yearly, monthly, weekly, and daily basis.•We show that residential heating dominates in winter and masks all other variations.•Our model consists of a modulated expansion in Fourier series where deviations from comfortable temperatures is a regressor.•On a monthly and weekly basis, we generate forecasts with and without temperature within a reasonable error margin.•On the daily basis, we use a day-ahead feedback to increase the forecasting accuracy.