Modelling and Forecasting the Unit Cost of Electricity Generated by Fossil Fuel Power Plants in Sri Lanka

Abstract

The national grid system which is evolved to deliver electricity must be always kept in balance so that it must have a sufficient production to meet the demand of electricity while minimizing the generation cost. This study presents a statistical time series model for forecasting the Unit Cost (UC) of generation of electricity in fossil fuel power plants by using two approaches namely Auto Regressive Integrated Moving Average (ARIMA) and time series regression. This is conducted as a case study in a Diesel/Heavy Fuel Oil (HFO) power plant in Sri Lanka which consists of two sub stations. ARIMA (1,1,0) and ARIMA (2,1,2) were selected as the best models with the lowest Akaike Information Criterion (AIC) under the ARIMA model approach while two dynamic regression models with coefficient of determination (R²) value 0.55 were selected under time series regression approach for Station 1 and Station 2 respectively. The regression model was identified as the best forecasting method for two stations with the minimum Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The forecasts of the future generation cost of electricity are extensively helpful for the national grid system for financial and capacity planning, fuel management and operational planning.