A stochastic model for representing drinking water demand at residential level

In this study, attention is initially focussed on modelling finely sampled ( 1 min) residential water demand time series. Subsequently, the possibility of simulating the water demand time series relevant to different time intervals and many users is analysed by using an aggregation approach. A cluster Neyman-Scott stochastic process (NSRP) is proposed to represent the residential water demand and a parameterisation procedure is implemented to respect the cyclical behaviour usually observed in any working day. A validation is performed on the basis of the one-minute data collected on the water distribution system of Castelfranco Emilia located in the province of Modena ( I). The elaborations performed show the validity both of the NSRP model and the parameterisation procedure proposed to represent the residential demand with fine time intervals ( up to 5 - 10 min). On the other hand, when a procedure of aggregation is applied to represent the water demand of a high number of users, the results are no longer satisfactory since only the mean is preserved while the other statistics, and in particular the variance, are underestimated.