A simulation study of stochastic water demands on distribution system transport and water quality over a large "all-Pipe" network system

A non-homogeneous Poisson Rectangular Pulse model (PRPsym) was used to generate stochastic water demands for a large "all-pipes" distribution system and then linked with EPANETto simulate the inherent impacts of water demand variability on the transport and water quality at three different levels of temporal demand aggregation(e.g., 1hour, 10-minute, and 1-minute). The simulation was modeled by "a short duration conservative chemical injection" event, which was injected into the network at two source nodes for three hours. Water quality simulations were performed by using the EPANET toolkit for a 5-day simulation. Simulated chemical concentration were collected at each minute for all nodes at the three different levels of temporal aggregations and the cumulative mass loading transported to each node was calculated. Results suggested meaningful differences in the arrival time and the time to reach half of the maximum mass loadings of the chemical across spatially diverse regions of the system with decreasing temporal aggregation scale. The results from this study provide invaluable information for understanding current modeling limitations and for improving existing modeling techniques to enhance the industry's ability to investigate multiple water quality applications with increased confidence.