Miglioramento dell'affidabilità di EPANET: risoluzione dei problemi di riflusso dell'emettitore e di segnalazione nodale nella formulazione FAVAD

Durable access to clean water relies on the precise management of Water Distribution Networks (WDNs), yet accurately modeling leakage and nodal behavior remains a significant technical challenge. This thesis addresses two critical limitations within the EPANET hydraulic engine: the unphysical occurrence of backflow through emitters during negative pressure scenarios and the lack of transparency in reporting nodal demands. To resolve these issues, an automated pipeline was created with the aid of Docker. This kind of development environment helped in the process of rapid iterative running and testing of the C-based source code. The core logic of engine modified with track-back analysis of hydraulic solver. With change of many modules in the engine code finally emitter backflow problem solved, and total nodal outflow separated to components: delivered, Leak-1 (Power Law leakage), and Leak-2 (FAVAD leakage). For validation of final version, four tests designed to run in the Fossolo WDN benchmark, a network with 58 pipes and 37 nodes. Numerical results confirm that the modified version successfully eliminates emitter backflow under negative pressure and separated total nodal outflow to their components correctly. Crucially, comparative analysis (Test-4) demonstrated that neglecting leak area variability in FAVAD approach modeling of leakage can lead to a significant underestimation of leakage rates. In the normal scenario, the model reported a leakage flow of 48.728 L/s (59% of total outflow) using FAVAD parameters, compared to only 18.299 L/s when area variability was ignored. These findings highlight the necessity of the FAVAD approach modeling leakage, especially for WDNs with high pipe elasticity. The resulting version provides more reliable, transparent, and physically accurate tool for the advanced modeling and management of leakage in hydraulic infrastructures for engineers and researchers.