Challenge
The council knew roughly how much water its network was losing. It did not know where.
The existing performance picture came from a top-down annual water balance: an Infrastructure Leakage Index of 2.93, placing the network in Band B of the IWA and Water New Zealand matrix — a well-managed system — against annual real losses of around 714,000 m³ and non-revenue water just under 20%. On those numbers, the network looked like a reasonable candidate for incremental improvement rather than urgent intervention.
But a top-down balance is built on assumptions. It divides a system-wide input volume against estimated consumption, and in this network the estimation problem was acute: there are no customer water meters. Billed consumption was derived entirely from flat-rate rating assumptions. A single conservative default — a 2% meter under-registration allowance applied to volumes that were never metered in the first place — was on its own misallocating roughly 66,000 m³ a year out of physical leakage and into apparent losses.
There was also no zone-level visibility. The network had zones on paper, but their hydraulic boundaries had never been proven. Some mapped boundary valves turned out to be in the wrong place, on the wrong pipe, or not closed at all. Without proven watertight zones, minimum night flow measurement is meaningless — unmetered flow crossing an unproven boundary makes a leaking zone look healthy.
The council needed the losses attributed to specific zones, measured rather than modelled, so that leak detection crews could be sent where the water was actually going.
Solution
Aqua Analytics was engaged by a South Island district council to deliver a full DMA validation across eight district metered areas — a network of approximately 200 km of reticulation mains serving around 8,000 connections. Field measurement found network real losses of 38.5 L/s, about 1.18 million m³ a year, and an Infrastructure Leakage Index of 6.2; more than double the desktop estimate the council had been planning against.
The project was delivered in three phases, with a formal hold point between desktop and field phases so that no field costs were committed before the overall delivery programme was agreed. Field delivery ran across an eleven-day window, with valve operations carried out by the network operator’s field crews under the direction of the Aqua Analytics field team.
- Phase 1 — desktop review and boundary design: We rebuilt and audited the top-down water balance, reviewed GIS, hydraulic model outputs, pressure zones, reservoir levels and historical night-flow records, and redesigned the zone configuration from the nine assumed at tender to eight. Three candidate sub-zones had too few connections for statistically meaningful night-flow analysis and were consolidated; one oversized zone was split at a hydraulic break identified in a 1999 pressure-zone study, which also relieved a long-standing summer peak pressure complaint at the top of the hill. Every proposed boundary was cross-checked against known hydraulic breaks rather than drawn on convenience.
- Boundary integrity proven before anything was measured: Each zone was isolated and tested by Partial Pressure Testing — valves closed to isolate the zone, with pressure at a nominated critical point allowed to decay to a controlled minimum of roughly 10 m rather than to zero. PPT gives the same boundary proof as full zero-pressure testing while keeping a residual head in the mains, which protects supply to vulnerable customers and reduces water-quality and air-ingress risk. Stable reduced pressure with no recovery proves isolation. Pressure recovery proves a breach.
- Sectorisation to find every breach: Where pressure recovered, the zone was progressively subdivided and additional valves operated until the leaking boundary was located. This is the slow, unglamorous part of the work and it is where the value is: each breach found is a boundary error corrected permanently in the client's asset records.
- Live pressure monitoring at every boundary: Loggers were installed in hydrants on both sides of each boundary valve and at each critical point, transmitting live so the field team could read the pressure response in real time and make isolation decisions on the spot rather than the following day.
- Minimum night flow in proven zone: Once boundaries were confirmed, zone flow was measured over the 02:00–04:00 window when legitimate demand is lowest. A metering function set was defined for each zone — the exact combination of meters added and subtracted to give net flow in — and its completeness formally recorded. Where a permanent meter had failed, a reservoir level-derived flow was substituted and the workaround documented rather than quietly relied upon. Temporary insertion flowmeters closed the remaining measurement gaps.
- Leakage split to New Zealand guidelines: Measured night flow was separated into legitimate customer night use, customer private supply-pipe leakage, and exceptional night use, per the New Zealand Water Loss Guidelines minimum night flow method. The residual is network real losses, converted to a daily and annual volume, normalised per connection and per kilometre of main, and benchmarked as an Infrastructure Leakage Index for each zone.
- Handed over as a working system, not a report: Two standard operating procedures were issued — one for ongoing MNF monitoring with exception thresholds and an escalation path, one for boundary and valve management so that the isolation the programme established is not undone by the next emergency repair — along with a corrected GIS layer and a ranked capital list for the permanent metering needed to keep the zones measurable.
Results
- Losses attributed zone by zone. Measured night flow across the monitored zones totalled 51.6 L/s, of which 38.5 L/s was network real losses after deducting customer components. Zone performance ranged from an ILI of 1.2 to 13.7. The single worst zone alone accounted for roughly a third of all measured network losses, at 1,049 L/connection/day.
- Seven of eight zones proven watertight. Six passed PPT directly. The seventh required three attempts. The eighth, the largest and most hydraulically central, feeding most of the rest of the network, could not be isolated for a standalone test without taking the main reservoir offline and cutting supply to a large share of the town. It was instead proven discrete by elimination: every adjacent zone was independently isolated and confirmed, establishing that no unmeasured flow crosses its boundary during the night-flow window.
- Prior leak detection measurably paid off. Two of the eight zones had received an acoustic leak detection inspection and repairs three months earlier, under a separate engagement with Aqua Analytics. They returned the two best results in the network, ILI 1.2 and 3.3. Zones with no recent active leakage control returned 5.4 to 13.7. The same measurement programme that quantified the problem also independently verified that active leak detection had worked.
- Asset records corrected on the ground. The programme found and resolved a series of defects no desk review would have caught:
- A mapped boundary valve that did not isolate the zone. Its recorded diameter was a third of the main it supposedly sat on — it was on an unmapped branch line. An adjacent valve provided true isolation and the asset record was corrected.
- The smallest zone resisted isolation across two full test attempts despite the pump station being taken offline and every known isolation point closed. Systematic sequential testing traced the cause to an undocumented sluice valve, originating from a water source that differed from the client’s own schematic.
- Valving around a disused booster station was configured differently from the record, requiring additional closures to separate two adjacent zones.
- A passing valve located by sectorisation and referred for maintenance.
- A reservoir outlet meter was found non-functioning, and a second zone’s meter set was producing a physically impossible negative net flow — both were flagged with the reasons documented rather than absorbed into the result.
Project Benefits and Conclusion
The council came away with a ranked leakage-control programme ordered by measured loss, with a survey method set for each zone — and the first zone on that list holds roughly a third of the network’s losses, which is the difference between a targeted campaign and a district-wide active leak detection sweep. Every zone now carries a measured Net Leakage Rate and ILI, so future improvement is provable rather than asserted, and the permanent metering recommendations are tiered by priority against the specific measurement gaps the field programme exposed. Two standard operating procedures leave the council’s own team running the weekly monitoring and protecting the zone boundaries, so the programme doesn’t decay the moment we demobilise.
A top-down water balance tells a utility roughly how much water it is losing. It cannot tell them where, and when the underlying consumption data is thin, it can be wrong by a factor of two. This programme replaced assumption with measurement: eight zones with proven boundaries, minimum night flow measured inside them, and losses attributed zone by zone — turning a district-wide estimate into a ranked, costed work programme. It also independently confirmed that the zones which had received active leak detection were the zones performing best.
Aqua Analytics offers Non-Revenue Water Consulting to help with design, validation, and baselining district-metered areas (DMAs) across Australia and New Zealand, and builds bottom-up water balances that make leakage reduction measurable.
If your zones exist on paper but have never been hydraulically proven, talk to our team.
