Abstract
Distribution network modeling is often used to investigate and manage water quality variations in a distribution network. It relies on pipe network simplification through skeletonization and uses water demand patterns that are often generalized or derived historical monthly water usage records. As automatic water meter reading and supervisory control and data acquisition (SCADA) technologies are widely used, it is possible now to explore the hydraulic complexity in the network. Processes such as stochastic and pulse water demand on solute transport characteristics can be investigated. Fidelity and appropriateness of network modeling by network simplification can be quantified. In this paper, these two questions are assessed by using real-time water demand measurements and comparative network simulations for an independent segment of a large water utility in the U.S. An all-pipe all-demand (APAD) model and an hourly demand variation curves (HDVC) demand model are simulated for the same network operations. The results show the prevalence of intermittent and pulse water demand particularly in network perimeters and dead-end branches. The results also highlight different node hydraulic properties such as Re, water age, and flow oscillation when water demand in APAD model is replaced by HDVC-based time-continuous generalized demand patterns. The degrees of such difference varies specific to the distribution network configurations such as H-loop, branches and dead-ends. These additional insights provide further understanding of the varying flow properties and their impacts on the movement of water parcels in pipe configurations. It is suggested that APAD network simulation be used for accuracy-demanding water quality simulation.