Weekly peak load forecasting for fast-developing cities

Abstract

The fast pace of the development activities in the Kingdom of Saudi Arabia (KSA) has resulted in the pressing need for more dependable forecasts for expanding industrial and residential electrical demands. Rapid-developing countries like the KSA experience a large, varying load growth rate, which has dynamic rather than static characteristics. The authors present a method for developing adequate time series models for the dynamic characteristics of the electrical power demands of a typical rapid-developing electric utility from actual five-year load demand time series. Two models have been developed. The first uses a long-trend component modelled by a polynomial function and a seasonal-cyclic component modelled by Fourier expansion. The second uses an autoregressive time series model for the seasonal-cyclic component. These models have been used to generate weekly peak-load forecasts for one year ahead.