PROFILING NODAL WATER DEMANDS IN STATISTICAL PERSPECTIVE

摘要

Nodal water demands are one of the most important factors to a network hydraulic model. The factors that influence water consumption are complex so that measurement and stochastic characterization of nodal water demands are needed to describe forecasting uncertainty. This paper develops a probability density distribution (PDF) model of nodal water demands not only as a descriptive model for profiling characteristics of nodal demand, but also as a predictive model to automatically and randomly generate nodal demands. Demand multipliers were characterized by PDF curves and obtained based on analysis of sample data of selected users, base demands of all nodes are obtained by averaging monthly demand retrieved from GIS-based billing system, so nodal demand, which is product of base demand and demand multiplier, in any assigned time interval can be presented as PDF curves. In the methodology, some statistical technique, such as distribution curve fitting, hypothesis test, random number generation etc, are used. Comparison of traditional methods, the method profiles nodal water demand in statistics perspective and at the same time requires lesser data measurement. Application to a real case shows promising results.