Minimum night flow (MNF) analysis, step by step

The IWA water balance estimates real losses once a year, as whatever is left after authorised consumption and apparent losses are subtracted. Minimum night flow analysis reaches the same quantity from the other direction. Between 02:00 and 04:00 most customers are asleep, so the water still entering a zone is mostly leakage.

The two methods use different inputs. When they agree, both numbers become more credible. When they do not, one of them contains an error worth finding.

This walks through the method on one example zone.

Before you start: a zone you can trust

Night flow analysis only works on a zone with a known boundary: a district metered area (DMA). Check three things first.

Every inlet and outlet is metered. If water can leave the zone through an unmetered connection, it disappears from the calculation. If it can enter that way, it appears as extra leakage.

Boundary valves are actually closed. One open valve connects the zone to its neighbour, and the night flow then describes neither zone. A pressure-drop test across the boundary confirms it.

The data is fine-grained enough. Inflow logged every 15 minutes or more often is best; hourly is the minimum. Log pressure at a point representative of the zone’s average pressure, not only at the inlet.

The example zone: 1,150 houses, a care home and a bakery, on 1,152 service connections. Inflow is logged every 15 minutes and pressure hourly.

Step 1: Find the minimum night flow

For each night, take the lowest rolling one-hour average of the inflow between 02:00 and 04:00 local time. A single 15-minute reading is too noisy, and one toilet cistern refilling at the wrong moment can move it.

One night is not a result. Collect at least a few weeks and use the median of the nightly values, not the mean. A burst that ran for five nights pulls the mean up but barely moves the median.

Drop nights with missing data, and nights when you know the zone was disturbed: mains flushing, a valve operation, a reservoir refilling inside the boundary.

For the example zone, the median over 30 nights is 6.20 m³/h.

Step 2: Estimate legitimate night use

Not all night flow is leakage. Legitimate night use has three parts:

Household night use: toilets, dishwashers and washing machines on timers, people who work night shifts.
Non-household night use: hospitals, care homes, bakeries, industry, anything that runs at night.
Exceptional night use: garden irrigation, swimming pools being filled, a farm’s livestock troughs.

Published per-property figures vary between countries: 1.7 litres per household per hour is a common UK value, 3 litres per property in Canada and 5 in Malaysia, while German and Austrian figures are given per person. Use a figure that fits your area, and measure wherever you can.

Measurement matters more than the choice of figure. In a study of a 294-connection zone in Gorino Ferrarese, Italy, where every customer meter logged hourly, literature values for night use estimated leakage at three times the measured value. The cause was a handful of customers: two houses irrigating at night and a football pitch using about 20 litres per minute. With hourly readings from customer meters you see those customers directly. Without them, ask who in the zone uses water at night before you trust a default.

For the example zone:

Component	How it was estimated	m³/h
Household night use	1,150 houses × 1.7 L/h	1.96
Care home and bakery	Their own hourly meters, 02:00–04:00	0.35
Exceptional use	None known; no irrigation in October	0.00
Legitimate night use		2.31

Step 3: Subtract to get net night flow

Net night flow = Minimum night flow − Legitimate night use

For the example zone: 6.20 − 2.31 = 3.89 m³/h. That is the leakage rate at night, about 63% of the night inflow.

Net night flow contains both background leakage (many small leaks too small to find individually) and any bursts that are running. The method cannot separate the two. A sudden step up from one night to the next usually means a new burst; a slow rise over months usually means background leakage is growing.

Step 4: Convert to a daily volume with the night-day factor

The tempting step is to multiply by 24 hours. That overstates leakage in most zones, because leakage depends on pressure and pressure changes through the day.

In a gravity-fed zone, pressure is highest at night: demand is low, so little pressure is lost to friction in the pipes. During the day, pressure falls and leaks run slower. The night-day factor (NDF) accounts for this. It is the number of hours at night-time leakage that give the same volume as one real day:

NDF = Σ (P_hour / P_night)^N1, summed over the 24 hours of the day

P_hour is the average zone pressure in each hour and P_night is the pressure during the minimum night flow hour. N1 describes how strongly leakage reacts to pressure. Values around 0.5 fit rigid pipes with fixed-size holes, and values of 1.5 and above fit flexible pipes where leaks open up under pressure; field studies have found values above 2. Use 1.0 for a mixed network when you have not measured it.

The example zone’s pressure profile:

Hours	Average zone pressure
00–06	52 m (the minimum night flow hour falls here)
06–09	45 m
09–17	48 m
17–21	44 m
21–24	49 m

With N1 = 1.0:

NDF = 6 × 52/52 + 3 × 45/52 + 8 × 48/52 + 4 × 44/52 + 3 × 49/52

NDF = 6 + 2.60 + 7.38 + 3.38 + 2.83 = 22.2 hours

Daily leakage = 3.89 m³/h × 22.2 h = 86.4 m³/day

Multiplying by 24 would have given 93.4 m³/day, about 8% more. N1 matters less than skipping the factor: with N1 = 0.5 the result is 89.9 m³/day, with N1 = 1.5 it is 83.2.

Zones with pressure management behave the other way. If a pressure-reducing valve or pump lowers the pressure at night, daytime pressure is higher than night pressure and the NDF is above 24.

The example zone’s leakage is 86.4 m³/day, or about 31,500 m³ a year, which is 75 litres per connection per day.

Step 5: Check it against the water balance

Now compare it with real losses from the water balance for the same zone and the same 30 days. For the example zone, the balance gives 98 m³/day. The night-flow estimate is 12% lower.

Both routes depend on estimates. The balance relies on customer meter accuracy and on unbilled consumption, and the night-flow route relies on night use and N1. We treat a difference of up to about a quarter as agreement. A larger gap usually has one of these causes:

The night-flow estimate is higher: unrecorded night users, irrigation, a reservoir or tank refilling inside the zone, or an open boundary valve.
The balance is higher: customer meters under-recording, unbilled consumption estimated too low, or readings from different periods subtracted from each other. The ILI step-by-step covers these rows of the balance.

The night-flow route does not replace the balance. It checks the balance, and it does so every morning instead of once a year.

Regulators use the same pairing. In England and Wales, reported leakage comes from a water balance that is cross-checked against night flows in district metered areas, and the two are reconciled statistically. In August 2026 Ofwat opened a consultation on how that method should change as smart meter data becomes available. It has funded 10.4 million smart water meters for 2025–2030.

Step 6: Watch the trend, not one night

The analysis is most useful as a daily series. Compare each night with the zone’s own recent history rather than with a target:

A step change from one night to the next points to a new burst, often before it shows at the surface or in a customer call.
A step down after a repair confirms that the leak has been fixed and that no second leak was hiding behind it.
A slow rise over weeks means background leakage is growing, or night use is changing. Irrigation in summer is the usual suspect.

A baseline built from the median of recent nights, with a band around it, separates real changes from normal variation.

Common mistakes

The inlet meter is not accurate at night. A meter sized for peak demand may run near its minimum flow rate at night. Under the EU Measuring Instruments Directive, the permitted error between the minimum and transitional flow rates is ±5%, against ±2% above them. Check where the night flow sits on the meter’s range.
The night window is in the wrong time zone. If loggers store UTC, 02:00–04:00 shifts by one or two hours depending on the season. The nights when clocks change are easy to drop by mistake, or to count twice.
A storage tank inside the zone. A tank or reservoir that refills overnight looks exactly like leakage.
One default for every zone. A zone with a hospital and a zone of holiday cottages do not have the same night use.
Multiplying by 24. See step 4.

Where the platform fits

Water loss management on Divako runs this analysis for each zone every night. It takes the lowest rolling one-hour flow between 02:00 and 04:00 in the project’s local time, and compares it with a baseline built from the median of the last 90 nights. A zone needs at least 30 nights of hourly data before it raises alerts. Zones whose night flow moves away from their own baseline appear first on the dashboard.

Legitimate night use defaults to 1.7 litres per connection per hour, and you can replace it with a declared figure for zones with known night users. The platform converts net night flow into a daily volume and compares it with the zone’s water balance, so a disagreement between the two is visible without a separate spreadsheet. The conversion uses 24 hours. In zones with large pressure variation through the day, treat it as an upper estimate and apply your own night-day factor.

Hourly readings from customer meters make step 2 a measurement instead of an assumption. Divako reads them from remotely read water meters over wM-Bus, LoRaWAN and NB-IoT. For the regulatory context of leakage assessment in the EU, see the Drinking Water Directive page.