A novel econometric model for peak demand forecasting

The future of generation, transmission and distribution of electrical energy depends primarily on accurate demand forecasting. Demand forecasting can be carried out on the basis of load in MW or energy in MWh [1]. Electricity forecasts are needed for generation and network expansion planning; evaluation of tariffs; operations and despatch management. The electricity supply industry in Zimbabwe is facing many technical problems (such as generation deficiencies) and other economic challenges resulting in wide spread load shedding programmes [7, 10]. An insight into a solution of these problems will be an in-depth study of the load forecasting techniques being currently used. A simplified economic model, based on a multiple linear regression approach, for the prediction of the electricity demand in Zimbabwe is proposed in the paper: the peak demand is defined as a function of GDP, CPI, population and temperature. A 30-year historical data set (from 1980 to 2010) was gathered from several sources. Statistical techniques using the MATLAB, SPSS and EXCEL software environments were then used in building the prediction model and their performance compared. The main contributions of the proposed paper are: An enhanced understanding of the load forecasting issues being faced by electricity supply industries in developing countries such as Zimbabwe, for example. The predicted results for 2011 to 2015 were benchmarked against those obtained by the African Development Bank.