Almost every water software vendor now sells a digital twin, and the products underneath are not the same kind of thing. Some are a hydraulic model of the network kept up to date with live sensor data. Others analyse the sensor data directly, with no model of the pipes at all. Both are useful. They answer different questions, and knowing which one you are buying is most of the decision.
What is a digital twin of a water network?
A digital twin is a model of a significant part of the network that is kept current with live data, such as SCADA readings, and that can optionally be used to control the real system. That is close to the definition used by KWR, the Dutch water research institute. A model that is run once for a study is a simulation, not a twin; we set out that line in digital twin vs simulation model, and the half step between them in digital twin vs digital shadow.
What digital twin tools support AI-driven decision-making in water utilities?
The tools that support decisions best combine both kinds: a calibrated hydraulic model that can estimate pressure and flow where there are no sensors, and data analytics that flag what the sensors show. Pure analytics can tell you that a district's night flow has risen. Only a model can tell you what happens to pressure across the network if you close a valve to find the leak.
Model-based twins
- Bentley OpenFlows WaterSight is a cloud product that brings SCADA, GIS, smart meter and sensor data together and uses continuously updated hydraulic models to estimate conditions where there are no sensors, with leak and anomaly detection. It is aimed at operators and engineers as well as modellers.
- Autodesk IWLive Pro connects an InfoWorks WS Pro model to SCADA, with demand forecasts and alerts when the model and the live readings disagree. It runs on the utility's own servers and needs a WS Pro licence. See InfoWorks WS Pro vs ICM.
- DHI MIKE OPERATIONS runs DHI models online for forecasting and operational control of rivers, sewers and supply networks, on DHI's cloud or the utility's own servers. It suits utilities already working in MIKE+.
- Xylem Vue, powered by GoAigua, is a data integration platform that can carry a calibrated hydraulic twin. GoAigua comes from Idrica, the company behind the Valencia twin, and Xylem took a majority stake in Idrica in December 2024.
Analytics platforms often sold as twins
- Autodesk Info360 Insight is cloud analytics and incident management across networks and treatment plants. It sits beside the hydraulic model rather than being one.
- Siemens SIWA is a family of apps for leak detection, pump scheduling and treatment, mostly data and AI driven, sold through Siemens' Xcelerator marketplace. Its newest treatment app is a physics based twin of a wastewater plant, not of a network.
- Aquasuite, from Haskoning, is a data driven "virtual operator" that learns from SCADA history to predict demand and steer production, with modules for leaks and sewers.
- SUEZ Aquadvanced monitors networks in real time for leaks and water quality events.
- TaKaDu is cloud event management: it finds leaks, faults and data problems in the sensor data. It does not describe itself as a twin, and it is better thought of as the analytics layer beside one.
For district metered areas specifically, we compare the leak tools in DMA leak detection software compared.
Can you build a water digital twin with open source tools?
Yes, and the best known water twin shows it can be done. The Valencia twin, which serves about 1.6 million people, is built on EPANET. The free building blocks are EPANET for hydraulics, WNTR from the US EPA and Sandia for leaks, pressure dependent demand and resilience, and EPyT from the University of Cyprus for running EPANET from Python. What you do not get free is the hard part: the live SCADA connection, the calibration that keeps the model honest, and an interface operators will open every day.
How to choose
Start from the question, not the product. If you need to know what happens across the network when you act, you need a model-based twin, which means a calibrated hydraulic model first. If you need to spot leaks and faults sooner, analytics on the sensor data may be enough and is faster to deploy. Then check what your team already runs: a twin built on the modelling software your engineers maintain will be kept up to date, and one built on something new often is not.
Product names and ownership in this market change often, so confirm the current position with each vendor before committing.
What actually determines the answer
A twin is only as good as its calibration and its data. A model nobody has checked against field pressures, meters that drift, and SCADA tags nobody can map to a pipe will produce a confident, live, wrong picture. Most of the cost and most of the value of a twin is in that unglamorous work, not in the platform.
Smart Bhujal builds digital twins around the models and data a utility already has, set out on our water digital twins page and our water distribution modeling page.
