HydroTwin-MPC: A Physics-Informed AI Digital Twin for Climate-Resilient Urban Water Resources Management

This paper presented HydroTwin-MPC, an integrated, physics-informed digital twin for climate-resilient urban water-resources management. The framework combines process-based supply and drainage models, real-time state estimation, an accelerated physics-informed surrogate, probabilistic forecasts, anomaly detection, risk-aware model predictive control, detailed safety replay, and human governance. The architecture is designed to be modular, auditable, and deployable in stages. In a synthetic semi-arid city benchmark, the complete framework outperformed fixed rule control across flood volume, unmet demand, energy, overflow duration, and leakage response while satisfying hard constraints. Robustness tests showed declining but positive benefit under joint sensor outage and rainfall bias. Ablation results identified uncertainty-aware optimization and state assimilation as critical components. These findings support the research hypotheses within the simulation domain. The strongest conclusion is procedural rather than numerical: a trustworthy water digital twin is a maintained engineering decision system, not a dashboard or an unconstrained AI predictor. Its value depends on verified asset data, balance closure, uncertainty communication, safe fallback, human authority, cybersecurity, equity constraints, and lifecycle ownership. The next step is a transparent shadow pilot with independent validation; automatic control should follow only after evidence demonstrates safety and public value. For engineering practice, five recommendations follow. Begin with a decision and baseline rather than a technology catalogue. Verify topology and water balance before training advanced models. Treat uncertainty and fallback as primary design requirements. Preserve operator approval until evidence supports bounded automation. Maintain the twin as a governed asset with versioned data, models, constraints, and tests. For research, integrated evaluation should replace isolated accuracy claims. Forecast skill, physical consistency, decision latency, tail risk, cybersecurity, equity, and realized operational benefit belong in the same evidence framework. Common open benchmark scenarios would make comparisons more meaningful and accelerate independent replication. For policy, digitalization can strengthen climate adaptation only when institutions can operate and maintain it. Capacity building, interoperable data, independent audit, and protection of low-resource communities are therefore part of the innovation itself. Funding models should support measurement and model stewardship beyond the initial demonstration period. HydroTwin-MPC is offered as a design and verification blueprint. Its synthetic results justify a carefully controlled pilot, not direct operational deployment. A successful next study would publish the baseline, acceptance gates, failure log, override record, and paired outcomes so that the claimed value can be examined and reproduced. The broader opportunity is to move urban water management from reactive operation toward anticipatory, transparent, and adaptive stewardship while retaining the conservative safety culture of civil engineering. That transition is credible only when physical law, empirical evidence, institutional judgment, and public accountability remain connected.

[1] M. M. H. van der Laan et al., “From digital twin paradigm to digital water services,” Journal of Hydroinformatics, vol. 25, no. 6, pp. 2444–2461, 2023. [2] P. Ghorbani Bam et al., “Digital Twin Applications in the Water Sector: A Review,” Water, vol. 17, no. 20, Art. 2957, 2025, doi: 10.3390/w17202957. [3] Y. Yang et al., “Digital twinning of river basins towards full-scale, sustainable and equitable water management and disaster mitigation,” npj Natural Hazards, vol. 1, Art. 43, 2024, doi: 10.1038/s44304-024-00047-2. [4] J. E. Pesantez, A. Alghamdi, S. Sabu, C. Mahinthakumar, and A. Berglund, “Using a digital twin to explore water infrastructure impacts during the COVID-19 pandemic,” Sustainable Cities and Society, vol. 77, Art. 103520, 2022, doi: 10.1016/j.scs.2021.103520. [5] World Bank, “Introducing Digital Water,” Water Global Practice, Washington, DC, 2024. [6] World Bank, Utility of the Future 4.0: Taking Water and Sanitation Utilities Beyond the Next Level. Washington, DC, 2024. [7] NITI Aayog, Composite Water Management Index: A Tool for Water Management. New Delhi, India, 2018. [8] NITI Aayog, Reuse of Treated Wastewater in Urban/Peri-Urban Agriculture in India. New Delhi, India, 2023. [9] L. A. Rossman, EPANET 2.2 User Manual, EPA/600/R-20/133. U.S. Environmental Protection Agency, Cincinnati, OH, 2020. [10] L. A. Rossman and W. C. Huber, Storm Water Management Model User’s Manual Version 5.2, EPA/600/R-22/030. U.S. Environmental Protection Agency, 2022.