Smart water metering and leak detection help African utilities and estates cut non-revenue water. IoT meters, LoRaWAN, analytics, prepaid billing and rollout.

Smart Water Metering in Africa: Leak Detection Guide

2026-08-29

Water that is produced, treated and pumped but never paid for is called non-revenue water. It leaks from old pipes, flows through meters that no longer read correctly, and is taken through illegal connections. For many African utilities, estates and industrial parks, it is among the largest hidden costs. Smart water metering and leak detection turn that loss into data you can act on.

This guide is for utility managers, estate and facility operators, and local government water agencies. It explains the technology in plain terms, shows what a rollout looks like and highlights the traps.

The problem: you cannot manage what you cannot measure

Water systems lose money in two ways. Physical losses are leaks and bursts. Commercial losses come from meters that under-read, billing errors, and unauthorised connections. In many networks nobody knows which is which, because bulk and customer meters are manual, old or missing.

A useful local example comes from Kenya. According to reporting on Nairobi's smart meter programme, the utility aims to cut non-revenue water from 54 percent to 39 percent by 2028/2029, using smart meters and an AI-powered leak detection system. Whatever the outcome, the target shows how large the problem is.

Common challenges in our region:

Manual meter reading that is slow, error-prone and open to collusion.
Estimated billing, which frustrates customers and causes disputes.
Leaks that run for weeks before anyone notices a wet patch.
Intermittent supply, which makes flow patterns hard to interpret.
Weak collections, because customers pay late or not at all.

The solution: connected meters and analytics

Smart meters

Modern smart meters, as the Nairobi article describes, can use ultrasonic measurement, which is accurate at low flows and has no moving parts to wear out, and they send readings automatically over GSM, LoRaWAN or NB-IoT. For many African sites the choice of network is practical:

Network	Best for	Consideration
GSM or LTE-M	Urban areas with good coverage	Ongoing SIM cost per meter
NB-IoT	Cities where operators support it	Availability varies by operator
LoRaWAN	Estates, campuses, towns with your own gateways	You operate the gateways, low data cost

District metered areas

Place bulk meters on each zone, called a district metered area (DMA), and compare what enters with what customers' meters record. The difference shows where the losses are. Night-time flow, when legitimate use is minimal, is a strong leak indicator.

Analytics and AI

A leak detection engine can combine several signals:

Minimum night flow per zone.
Pressure anomalies from pressure loggers on the network.
Continuous flow alarms from customer meters that never show zero, which suggests a leak behind the meter.
Machine learning to separate normal pattern changes from real leaks, and to rank alerts by estimated loss.

A simple alert rule might look like this:

alert: probable_leak
when:
  dma_night_flow_m3h: above 1.3 x 30_day_baseline
  for: 3 nights
action:
  - create_work_order: field_team
  - notify: network_engineer

Billing and customer service

Connected meters enable accurate bills, usage alerts and prepaid options, where supply stops when credit runs out, much like prepaid electricity. Payments can integrate with mobile money and bank transfer, and customers can query balances through WhatsApp.

Benefits

Lower non-revenue water, as leaks are located faster and under-reading meters are replaced.
Better revenue, through accurate metering and easier payment.
Less manual work, since meter reading trips are eliminated.
Transparent billing, which reduces disputes.
Planning data, helping you decide where to replace pipes first.

If you manage a utility, estate or industrial park and want a realistic plan, request a quote.

Proof and examples

Smart metering and network analytics are established in the water industry, and African utilities such as Nairobi's are investing in them. We will not promise percentage savings, as results depend on the starting condition of your network, meter quality and how quickly crews act on alerts. Some losses are only fixed by replacing pipes, not by software.

Our experience with sensor networks is reflected in StrikeClimate, and our solutions page outlines what we offer in IoT and software. Related reading: our article on AI and IoT flood early warning for Nigerian communities, which uses a similar water-level sensing approach, and automating insurance claims and fraud detection, where the same anomaly-detection ideas apply to a different industry.

Pitfalls to plan for

Smart metering projects often fail on the basics. Meters are installed in flooded or inaccessible chambers where signals cannot escape. Customer records do not match meter serial numbers, so bills go to the wrong account. Field crews are not equipped or motivated to fix the leaks the system finds. Survey chamber conditions before ordering, clean up your customer database first, and set a service target for how quickly each alert must be repaired.

Implementation roadmap

Baseline the network (weeks 1 to 4). Establish how much water enters and how much is billed. Check existing meters.
Choose a pilot zone (weeks 3 to 6). Pick a zone with a clear boundary, so a DMA can be set up.
Install bulk and customer smart meters (weeks 6 to 12). Add pressure loggers at key points.
Deploy gateways and the platform (weeks 8 to 14). Dashboards, alerts and work-order integration.
Run the pilot for three to six months. Measure losses found, leaks fixed and billing accuracy.
Add billing and payment integration. Prepaid or postpaid with mobile money and WhatsApp updates.
Scale zone by zone, using pilot data to prioritise.

Customer water usage can reveal when people are at home, so it is personal data under the NDPA 2023. Limit access, secure the platform and explain the data use to customers.

FAQ

Do smart meters work with intermittent supply?

Yes, but design for it. Meters should log zero-flow periods and air in the pipe, and analytics must be tuned to your supply schedule.

Which network should we use?

Use cellular where coverage is good and per-meter costs are acceptable, and LoRaWAN where you control a compact area and want low data costs. Many projects mix both.

Will it stop theft and illegal connections?

It makes them easier to find, because zone balances show unexplained losses. Stopping them still needs field enforcement and customer engagement.

Get started

Whether you run a city network, an estate or a factory campus, a well-scoped pilot will show where your water is going. Book a free consultation or chat with us on WhatsApp and we will help you design it.