Disinfection by products estimation in a water distribution network

Even though disinfection is necessary to ensure water safety for human consumption, some disinfectants produce disinfection by-products (DBPs) that may be dangerous for human health. Current European legislation obligates water distributors to limit some DBPs concentration to final consumers. Then, water companies must control these compounds and are obligated to periodically monitor their network. DBPs modeling can be very useful for estimating online DBPs concentration throughout the network, increasing DBPs control and knowledge, but avoiding DBPs analytics time and resources consumption [1]. Trihalomethanes (THM), the first DBPs discovered, have long been the most studied and modeled. Previous studies have mostly used linear relations between variables and THM concentration, but also computational modelling, mechanistic and data driven models [2, 3]. Even though, there are still challenges to beat: most studies use a small database and laboratory-scale for model building, forgetting the impact of network pipelines and season. In addition, significant variables for DBPs’ formation such as retention time are most of the time neglected due to its difficulty to measure. Finally, THMs are not the only DBPs generated from disinfection or even the most toxic: other DBPs must be studied, and their formation pathways along the network investigated. In this study, data from a full-scale distribution network was used: online sensors and sampling campaigns. To include hydraulic conditions as retention time, EPANET software and R programming are used to simulate the network. Different models, mechanistic and data driven, have been used to estimate the chlorine decay and DBP formation within the network. Results of the calibration and validation of these models and the conclusions obtained are presented.Peer ReviewedPostprint (published version