Fast and physically interpretable data assimilation to support digital twins of urban water systems

Urban water systems, including Water Distribution Systems (WDSs) and Urban Drainage Systems (UDSs), face growing challenges from climate change, urbanization, and ageing infrastructure. Digital Twins (DTs) offer a pathway for improved management by integrating sensor data with simulation models. Reliable operation of DTs requires continuous model updating, achieved through data assimilation (DA). This study presents a Control Theory-based Data Assimilation (CTDA) method that uses Proportional–Integral–Derivative (PID) controllers to correct unreliable boundary conditions in simulation models. CTDA adjusts nodal demands in EPANET model (WDS) and nodal inflows in SWMM model (UDS) based on sensor observations. Two Serbian case studies - Bečej (WDS) and Belgrade (UDS) - demonstrate improved agreement between simulated and observed data, confirming the method’s capability for continuous model correction. CTDA enhances DT operation by providing computationally efficient, interpretable, and transferable real-time model updating for urban water systems.