One question, clear ownership
Can a connected model help operators assess whether tomorrow’s demand can be met within the reservoir operating range? The scenario below is a pilot design, not a report of a live utility implementation.
Connect the evidence
- Ingest timestamped tank levels, flows and relevant device states.
- Flag suspect observations and retain the previous validated state.
- Provide a documented forecast with a baseline fallback.
- Run the hydraulic model using traceable boundary conditions.
Evaluate in shadow mode
Compare predictions and recommendations with actual operation while existing controls remain in charge. Track data availability, forecast error, storage trajectory error and operator usefulness. Record interventions and exceptional events before judging model performance.
Define the progression gate
Move beyond the pilot only after agreed performance and operational criteria are met. A recommendation interface and an automatic control system have different authority and assurance requirements. Read the digital twin guide for the architecture.
Sources & further reading
- EPANET: water distribution modelling ↗US Environmental Protection Agency · 2020 release; living resource
External source · Checked 24 September 2026 - AI Risk Management Framework ↗National Institute of Standards and Technology · 2023 framework; 2024 generative AI profile
External source · Checked 24 September 2026 - Best Practice Modelling Guidelines ↗eWater · 2011
External source · Checked 24 September 2026
Source findings are distinguished from editorial interpretation. Apply current local criteria and project evidence when making engineering decisions.