Digital twins as virtual replicas that synchronise with physical assets through sensor data and simulations are emerging as transformative tools for water utilities. They offer the ability to monitor infrastructure in real time, predict system behaviour and optimise operations across the entire water cycle. This review critically examines the architectures used in digital twins for water utilities, the workflows employed for model calibration and the benefits measured in empirical studies. A systematic literature search identified journal articles and conference papers on digital twins in water distribution, supply and wastewater systems. Empirical case studies were analysed to extract information on system architectures, data integration methods, calibration techniques and quantitative performance metrics. Results show that typical architectures comprise multi‑layered structures integrating physical sensors, communication networks, data repositories, simulation engines and user interfaces. Calibration workflows use hydraulic models (EPANET), data assimilation techniques and machine‑learning algorithms such as temporal graph convolutional networks to align the digital model with sensor observations. Measured benefits include high prediction accuracy (R² up to 0.972 and MAE values around 0.011 for pump speed estimation improved leak detection and energy savings. However, challenges remain regarding data quality, computational demands and organisational readiness. The review concludes that digital twins hold great promise for resilient and sustainable water utilities but broader adoption will require standardised frameworks, capacity building and robust evaluation of socio‑economic impacts.
Qi Li (Thu,) studied this question.