Where NRW programmes stall — and what each one needs.
Non-revenue water programmes rarely fail on the technology. They stall on zone integrity, on meters that lie quietly, and on repairs nobody went back to measure. These are the ones we meet most often.
| What goes wrong | How you notice | The fix |
|---|---|---|
| Zone boundaries not actually closed | Night flow never settles; the balance refuses to close | Do the valve survey before the metering. One unrecorded open valve makes the whole DMA arithmetic meaningless |
| Bulk meter sized for peak flow | Night flow reads zero, or steps between values | Size for the smallest flow you need to see. An oversized meter goes blind at night — precisely when you are looking |
| Meter under-registration read as leakage | Losses concentrate in the zones with the oldest meters | Test a sample of consumer meters before you chase pipes; an under-registering fleet looks identical to a leak on a balance sheet |
| Pressure reduced without a baseline | Complaints rise and the loss figure barely moves | Establish the minimum service level and a before-reading first. Reduce blind and you have traded water for complaints |
| Apparent losses budgeted as real ones | Zone loss is high but the survey finds nothing | A survey that finds nothing is information. Separate apparent from real losses before allocating a repair budget |
| Repairs never verified | The same zone keeps reappearing on the priority list | Re-measure night flow after every repair. An unverified repair is an assumption with a work order attached |
| Logger gaps on the nights that matter | The balance fails exactly on the days you wanted to examine | Buffer locally and backfill — a gap during an outage removes the night the leak was loudest |
| A study rather than a programme | Losses are back within a year of the report | The programme is the reading taken every night; the report is only what you did with the first few |
Two of these account for most wasted repair budget. Apparent losses — under-registering meters, billing gaps, unmetered connections — sit inside the same number as real leakage but need a completely different response, and a crew sent to find a burst that was never there erodes confidence in the whole programme. Pressure management is the other: it works, and it works badly when applied without a measured baseline and an agreed minimum service level.
South African networks add one more. Pump stations that stop and restart with the grid put a transient through the network on every return, which both causes bursts and distorts the night-flow record. Zones that look erratic are often not leaking more — they are being measured through a restart, which is why the loggers either side of an outage need to keep reporting.