Simulating Residential Water Demand with a Stochastic End-Use Model

Abstract

A water demand end-use model was developed to predict water demand patterns with a small time scale (1 s) and small spatial scale (residence level). The end-use model is based on statistical information of users and end-uses: census data such as the number of people per household and their ages; the frequency of use; duration and flow per water-use event; occurrence over the day for different end-uses such as flushing the toilet, doing the laundry, washing hands, etc. With this approach, water demand patterns can be simulated. The simulation results were compared to measured water demand patterns on attributes such as peak flow and daily total water use, as well as on the shape of the pattern and the frequency distribution of flows and accelerations in flow. The simulation results show a good correspondence to measured water demands. Because the end-use model is based on statistical information rather than flow measurements, the model is transferable to diverse residential areas in different countries. The model can be applied in the design stage (prebuild), in scenario studies, and in water quality distribution network models.