Improved Near Real-Time Hydraulic Model Recalibration in a Water Distribution System with Advanced Metering Infrastructure (AMI) Data

<p dir="ltr">Real-time or near real-time (NRT) anomaly detection and localization in water distribution networks (WDNs) relies on live digital twins with accurate hydraulic models. The increasing deployment of advanced metering infrastructure (AMI) significantly enhances demand estimation and digital twin readiness. This study presents the integration of AMI data into the Anomaly Leak Finder (ALF), a software solution for anomaly detection and leak localization in WDNs, by dynamically updating the aggregated AMI consumptions within the hydraulic model in NRT. Field validation was carried out through two hydrant tests on a real WDN. The integration of AMI data led to a 50% reduction in mean absolute percentage error (MAPE) for pressure and flow compared to models relying on conventional monthly billing data, demonstrating a significant improvement in model accuracy. With AMI integration, ALF successfully detected and localized the test hydrants, highlighting the value and potential of using live AMI data to support real-time hydraulic model updates.</p><p dir="ltr">This paper was presented at the 21st Computing and Control in the Water Industry Conference (CCWI 2025) at the University of Sheffield (1st - 3rd September 2025).</p>