CCWI2017: F10 'Online Demand Estimation of Geographical and Non-Geographical Distributed Demand Pattern in Water Distribution Networks'

The issue of demand calibration and estimation under uncertainty is known to be an exceptionally difficult problem in water distribution system modelling. In the context of real-time event modelling, the stochastic behaviour of the water demands and non-geographical distribution of the demand patterns makes it even more complicated. This paper considers a predictor – corrector approach, implemented by a particle filter model, for solving the problem of demand multiplier factor estimation. A demand forecasting model is used to predict the water demand multiplier factors. The EPANET hydraulic solver is applied to simulate the hydraulic behaviour of a water network. Real time observations are integrated via a formulation of the particle filter model to correct the demand predictions. A water distribution network of realistic size with two configurations of demand patterns (geographically distributed demand patterns and non-geographically distributed demand patterns) are used to evaluate the particle filter model. Results show that the model is able to provide good estimation of the demand multiplier factors in a near real-time context if the measurement errors are small. Large measurement errors may result in inaccurate estimates of the demand values.