Analisi della vulnerabilità dei sistemi di distribuzione idrica rispetto a potenziali fenomeni di contaminazione

Drinking Water Distribution Systems (DWDSs), as well as other complex systems are characterized by an intrinsic vulnerability which can be both quantitative and qualitative: the first one related to the possible lack of water and the second one related to presence of substances concentrations higher than the legal threshold. Assessing the DWDSs’ vulnerability and resilience in an objective way is necessary to understand the potential impacts of random disturbances, deliberate attacks, and natural disasters and to be aware of the system critical points that need to be monitored. The work that is presented, in collaboration with Cafc S.p.a. (the Udine water utililty), aims to be a preliminary study on the vulnerability assessment and reduction of a real DWDS, i.e. the Udine DWDS. A general assessment of the vulnerability of DWDS in Udine was made using indices and metrics derived from the Complex Networks Theory (CNT). By comparing these topological parameters with those of other real and literature networks, it was possible to understand the level of redundancy and robustness, as indicators of resilience of the network under consideration. This analysis, already in use for different types of network and based solely on its topology, is a fundamental step to have an overview of the network and understand the critical areas of improvement. From the interpretation of the results of this first analysis, it was clear that, in terms of operating conditions, a major improvement in the network would be achieved by reducing the failures research time. In order to reach this goal, it was decided to partitioning the network through a procedure based on the CNT. The Water Distribution Network Partitioning (WDNP), which aims to subdivide the network into hydraulically independent subsystems, has provided for the use of optimization procedures using Genetic Algorithms (GAs) operating on a suitably calibrated hydraulic network simulation model. The size and level of detail of this model has certainly been an added value throughout the study, which among the many examples of literature can be considered a rare case of applied engineering. Always considering the operating conditions, it was decided to study the qualitative vulnerability deriving from the Disinfection by-Products (DBPs). These substances, resulting from the reaction of the disinfectant with the natural organic matter (NOM) in the network, are harmful to human health. Therefore, knowing both the concentration of these substances in the DWDS and their variation with the initial concentration of disinfectant, it becomes possible to simultaneously assess and manage any critical issue. To this end, it was decided to build a chemical-hydraulic model, using the EPANET-MSX software, capable of predicting the decay of disinfectant and the formation of DBPs. Being the Udine network disinfected by chlorine dioxide, the expected DBPs are chlorite and chlorate whose concentrations have a legal limit of 0.7 mg/l. There is few literature on the subject, and the attempts to model these substances have significant approximations that reduce the quality of model predictions. Moreover, there are no cases of chemical modelling of these substances for real networks of the size and level of detail of the WDN of Udine. The results of the simulation, compared with the concentration values of the three chemical species measured in the real network have confirmed an excellent predictive ability of the model that was finally used to simulate a contamination scenario in order to assess the possibility of increased concentration of DBPs above the legal limit.