Water

Smart water management is mostly about two numbers: how much water leaks out of the network, and how fast the city knows when too much of it arrives at once. Everything else on a vendor slide is downstream of those two.

The atlas holds 30 documented entries across 23 cities in 19 countries, spanning 2002 to projects still in planning for 2026. Water is unusual in this atlas because it has a genuinely comparable metric — non-revenue water — which makes it one of the few fields where cities can be lined up honestly.

Leakage: the field's one comparable number

Non-revenue water is the share of treated water that never reaches a paying customer, mostly through leaks. It is reported consistently enough to compare, and the trend lines below are all published by the utilities themselves.

Two honest caveats before anyone copies a row. Every figure here is utility-reported. And leakage rates depend heavily on network age, pressure, soil and city size — Hong Kong halving a high number is arguably a harder achievement than Singapore holding a low one.

The most useful failure in the atlas

Baghdad's Al Mansour smart-meter pilot deserves to be read by anyone about to buy metering. An independent 2021 evaluation found it delivered no reduction in consumption and none in network losses. The reason was structural: the 149 meters were never networked, so remote monitoring was impossible and staff still read them manually every six months — exactly like the conventional meters they replaced.

The recorded lesson generalises well beyond Iraq: the intelligence lives in the system around the meter — telemetry, billing reform, someone acting on the data — not in the meter itself. A metering programme without those three is a hardware replacement with a better logo.

Seoul's rollout shows the opposite configuration and, notably, an unexpected benefit: across four pilot districts, meter freeze incidents fell from 1,853 to 47 — because digital meters are read remotely, faults surface immediately, and the winter failure mode changed. Self-reported, but concrete, and not the benefit the programme was sold on.

Too much water, too fast

The other half of this topic is arrival rather than loss. Copenhagen's cloudburst plan — a multi-decade, multi-billion-kroner response to the 2011 flood — is the largest, and it is mostly concrete and landscape rather than software. Singapore's PUB network and Shanghai's flood and typhoon command are the sensing versions; Shanghai reports about 8 million camera snapshots screened automatically in 2025, detecting 378 waterlogging incidents.

The distinction matters for procurement. Sensing tells a city where water is; it does not move any. Cities that have measurably reduced flood damage in this atlas did drainage work, and used sensing to target and verify it.

Water as an energy asset

Three entries invert the usual framing and treat water as fuel. Helsinki's Katri Vala plant recovers over 1,000 GWh of heat a year from purified wastewater and cooling return — enough to cover most of central Helsinki's summer district heating. Geneva's GeniLac pumps cold lake water through the city, cutting cooling electricity by up to 80% in connected buildings. Dresden's Kaditz wastewater plant covers around 70% of its own electricity from biogas, solar and an outflow turbine.

The twenty-two-year bookend

Copenhagen has forecast harbour bathing water since 2002, combining rainfall and overflow data into red/green warnings. In 2024, Paris opened the Seine to legal public swimming for the first time in over a century, with sites opening or closing daily on water-quality results. Same instrument, two decades apart — and the clearest illustration in this atlas that the visible payoff of water sensing is usually a decision rule, not a dashboard.

Questions to ask before procurement

• Is the meter networked? If not, you have bought a meter. See Baghdad.
• Who acts on an alarm, within what time? Detection without a repair crew changes nothing measurable.
• What is the baseline non-revenue water? Establish it before the contract, or no later claim can be checked.
• Does billing change? Volumetric billing is often the mechanism that makes metering pay, not the telemetry.
• Sensing or drainage? For flooding, be explicit about which problem the budget is solving.
• Is there waste heat in the system? Wastewater and cooling returns are proven, under-used urban energy sources.

The practical takeaway: water is one of the few smart-city fields with a hard, comparable, money-denominated metric — and the entry with the most rigorous evaluation in it is a failure. Buy the system around the meter, name who responds to an alert, and measure the baseline before anyone installs anything.