evidoria

← Back to browse

Good practice Imported

SIWA Leak Finder — VA SYD's AI-Based Non-Revenue Water Reduction in Malmö, Sweden

Sweden · Malmö · See the Sweden profile

Evidence: Observational / pre–post Top 75% 47/100 · Ask Evidence Copilot about this practice

VA SYD, the public water utility serving Malmö, Lund and neighbouring municipalities, piloted Siemens' SIWA Leak Finder AI to spot leaks as small as 0.5 l/s and cut non-revenue water from 10% to 8% in a pilot zone — a result that led utility NSVA to adopt a similar system.

8 % (down from 10%)
Non-revenue water in pilot zone (2019-2024 pilot)
0.5 litres/second
Minimum detectable leak size (2019-2024 pilot)
SIWA Leak Finder — VA SYD's AI-Based Non-Revenue Water Reduction in Malmö, Sweden

Details

Maturity
Pilot
Promoter
VA SYD
Period
2019-2024
Region (NUTS)
SE224
Keywords
water utility, leak detection, machine learning, infrastructure

Context

VA SYD is the publicly owned water and wastewater utility for Malmö, Lund, Eslöv and Burlöv in southern Sweden, serving 546,000+ customers across roughly 5,000km of pipelines (2,000km carrying drinking water), with a share of treated water historically lost to undetected leaks (non-revenue water, NRW).

Objectives

Detect leaks earlier and at smaller scale than manual acoustic surveys, using AI analytics trained on the utility's own flow- and pressure-sensor data, while keeping operational data on-premises rather than in the cloud.

Activities

From 2019, VA SYD piloted Siemens' SIWA Leak Finder in a defined zone of roughly 5,000 consumers, led by development engineer Simon Granath, deployed on-premises to keep data inside VA SYD's own infrastructure.

Results

Non-revenue water in the pilot zone fell from 10% to 8%, and the system detected leaks as small as 0.5 litres per second — smaller than VA SYD's previous methods could reliably catch.

Conclusions

The results led neighbouring utility NSVA to begin implementing a similar system, and VA SYD cites the pilot as a step toward its 2030 target of zero unplanned service disruptions; the figures cover a single pilot zone and no independent third-party audit of them has been published.

Implementation
Indicative cost
Medium (€50k–€500k) — Cost not publicly disclosed; the pilot layered AI analytics onto VA SYD's existing sensor infrastructure rather than new capital works — treated as medium band pending disclosure.
Time to results
Medium (1–3 years) — Piloted from 2019; results reported through 2024, with neighbouring utility NSVA beginning a similar rollout.
Staffing & skills
VA SYD engineering team (led by development engineer Simon Granath), Siemens (SIWA Leak Finder vendor)

Conditions for success

  • An existing flow- and pressure-sensor network to train the AI model on historical data
  • On-premises deployment to keep operational data under utility control
  • A defined pilot zone to validate results before wider rollout

Where it fits

Governance type
multi-municipal public utility
Scale
pilot zone (~5,000 consumers) within a 546,000+ customer network

Commonly funded by

Own resources / municipal budget LIFE Programme

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful