DEMAND FORECASTING FOR WATER DISTRIBUTION SYSTEMS

Abstract

Abstract Real-time demand forecasts form the basis for the on-line computerised control of water distribution systems and affect the system security and reliability. This Paper investigates the methodologies for consumer demand modelling and prediction in a real-time environment. An approach based on time series analysis techniques is presented. A model using a combination of exponentially weighted mean and autoregressive structures is developed to predict the daily demand. A set of adaptive templates is used to allocate the daily demand to produce the diurnal demand profile. The models are established based on actual data from an actual distribution system. The performance of the practical implementation of the models is given and assessed. For the case study presented, it is found that the relatively simple time series models can be satisfactorily used to predict consumer demand for the real-time control purposes.