Where did the water go? Non-revenue water for municipalities.

Leakage becomes a reportable number

The recast EU Drinking Water Directive (2020/2184) did something quietly significant for water utilities: it made water loss a number that has to be assessed and reported. Member states had to evaluate leakage levels – using the Infrastructure Leakage Index or an equivalent method – and communicate the results to the Commission by 12 January 2026. By 12 January 2028 the Commission sets an EU-wide threshold; utilities above it will need an action plan.

The reporting obligation formally targets larger suppliers (over 10,000 m³ a day or 50,000 people served), but the direction applies to everyone: “how much do we lose” is stopping being a rough annual estimate and becoming a figure someone will eventually ask for.

What non-revenue water is made of

Non-revenue water is the difference between the water that enters the network and the water that ends up on bills. It splits into three very different buckets:

• Real losses – leaks on mains and service connections, bursts, reservoir overflows. This water was treated and pumped, so every cubic metre carries energy and chemical costs before it disappears into the ground.
• Apparent losses – water that was consumed but never measured correctly: ageing meters that under-register, data errors, unauthorised use. This is lost revenue rather than lost water.
• Unbilled authorised consumption – flushing, firefighting, network maintenance. Legitimate, but it belongs in the balance, not in the mystery column.

The split matters because the remedies are different: real losses are found in the field, apparent losses are found in the meter fleet and the data.

Why the balance fails without synchronised readings

The arithmetic of a water balance is trivial – input minus billed consumption. What breaks it in practice is timing. If the district meter is read automatically at midnight but customer meters are read manually across three weeks (or estimated), the balance mixes different periods, and the resulting “loss” figure is mostly noise.

This is the same mechanism as the main-meter gap in an apartment building, at network scale. Continuous remote reading fixes it structurally: when every meter in a zone is read at the same moment and every reading passes through the same validation, the difference between input and consumption becomes a signal instead of an artefact.

Zones and night flow: where the signal lives

A network-wide loss percentage tells you that water is disappearing; it does not tell you where. That is what zone metering is for – breaking the network into measurable districts so the balance can point at a place. We have written about the mechanics separately in zone metering in practice.

Within a zone, the most useful single indicator is minimum night flow: consumption between roughly 02:00 and 04:00, when legitimate use is at its lowest. A zone whose night flow creeps upward has, with high probability, sprung a leak – days or weeks before anyone sees water on the street.

What the platform side looks like

In Divako, the water-loss module keeps the balance continuously: NRW percentage and water balance per zone, minimum night flow trends per zone and per meter, and alarms that go to email, webhooks or an existing incident system when a threshold is crossed. The readings themselves can arrive over LoRaWAN, wM-Bus or NB-IoT – and by walk-by or drive-by collection for the meters no network reaches – all landing in the same data model.

Where to start

A useful water balance does not require 100% meter coverage on day one. The practical order is: the system input meters first, then boundary meters for the largest zones, then customer metering zone by zone. Even a monthly balance per zone beats an annual network-wide estimate; a daily one turns loss hunting into routine operations rather than crisis response.

By the time the EU threshold exists in 2028, the comfortable position is simple: a loss number you trust, and a trend that points down.