Author(s): Stefania Piazza; Mariacrocetta Sambito; Gabriele Freni

Linked Author(s): Stefania Piazza, Mariacrocetta Sambito, Gabriele Freni

Keywords: Water distribution network; Threshold index; EPANET model; Advective-dispersive-diffusive model

Abstract: The present study aims to provide a quick decision-making tool, in order to identify the most effective analysis model to be applied to water distribution networks (standard advective model or advective - diffusive models), in the event that one wants to simulate the fate of dissolved contaminants. State-of-the-art applications in water distribution modelling are based on advective approaches in which contaminants are supposed to follow water flows and diffusion can be neglected. One of the most used advective models is EPANET (Rossman et al., 2000). It is known that EPANET adopts a simplified advective approach in modeling the water quality, considering simplified reaction kinetics: (〖∂C〗_i (x,t))/∂t=-u_m (〖∂C〗_i (x,t))/∂x-KC_i (x,t). This model is based on a mass balance of the plug flow substance, where the mass of the substance is assigned to discrete volume elements. Subsequently, the reactions within the single element are triggered. At this point, the substance mass is advanced from one element to another and the volumes of mass and flow mix with each other at the nodes of the network. These simplifications have an enormous advantage, i.e. to reduce the computational load, not producing particularly significant errors in purely turbulent flow regimes. On the other hand, it is not always possible to adopt these simplifications since dispersion-diffusion phenomena are inherent in the transport processes and are due to concentration gradients, which cause the migration of contaminants from areas of high concentration to areas of low concentration. It has been observed that these processes become relevant in the presence of low speeds and a Reynolds number lower than 50,000, as often happens in urban water distribution networks at night. The use of diffusive-dispersive models allows obtaining enormous advantages as regards the modeling of the behavior of solutes, as it improves the prediction of the pulse shape. On the other hand, the computational load increases considerably. Previous studies, conducted on the real network of Zandvoort (The Netherlands), have shown that, despite mainly turbulent flows, due to the topological complexity of the network, the flows tapped at the nodes, the pipelines size, the diffusive – dispersive processes were found to be relevant due to the presence of laminar and transition flow in some of the pipes for some hours during night. Therefore, the purpose of this study is to determine an indicator that is able to establish a priori the behavior of the water network, in such a way as to define a threshold, beyond which the simple advective model is not efficient for modeling of water quality and a more complete diffusive – dispersive approach is needed.

DOI: https://doi.org/10.3850/IAHR-39WC252171192022422

Year: 2022