Methodology

How WaterBalanceCalc turns your data into a water balance, what it is asking for at each step, and how to read what comes out. The formulas are at the end for anyone who wants to audit a number.

The idea in one paragraph

Over a period — usually a year — a known volume of water went into your network. Some of it was used legitimately and billed. Some was used legitimately but never billed. Some reached customers but was never recorded because meters under-read, connections were illegal or the billing data was wrong. Whatever is left leaked out of the pipes. The water balance is that accounting, done in the IWA standard format, with an honest statement of how sure you are about each number. From it come the indicators — the Infrastructure Leakage Index above all — that say how well the system is managed, what the losses cost, and what they emit.

Before you start: what to gather

A quick balance needs about fifteen numbers. Most sit in three places: the operations team, the billing system and the asset register. Have these to hand:

Don't have something? Enter your best guess with a wide ± and use “Help me estimate” — the assistant proposes a localised value and a defensible uncertainty. You can tighten it later.

What each step asks, and why

1. 1. Set up

   Asks for: Name, period length, benchmarking category, units, depth, and where the system is.

   Why: The period turns volumes into per-day and per-year figures. The category (standard vs low & middle income) picks which band thresholds apply. Location lets the assistant localise its suggestions.

2. 2. Water in

   Asks for: Each source, its volume for the period, the accuracy of its meter, and how the water was treated.

   Why: This is the top of the balance and the best-measured number you have; everything else is compared to it. Treatment category sets the carbon per cubic metre.

3. 3. Authorised consumption

   Asks for: Billed and unbilled use, metered and unmetered, by category. Bulk exports separately.

   Why: Billed authorised consumption is revenue water; the rest of the input is NRW. Splitting by category makes the meter-error step more accurate and the report more useful.

4. 4. Apparent losses

   Asks for: Estimates of illegal use, meter under-registration and data errors, each with a wide ±.

   Why: These are subtracted from water losses to leave real losses. Under-estimating them overstates leakage — so estimate honestly and carry the uncertainty.

5. 5. Network

   Asks for: Mains length, connections (including inactive and illegal), and the service pipe length from boundary to meter.

   Why: The UARL — the leakage floor for your network — is built from these. Service pipe length is the input people most often get wrong.

6. 6. Pressure & supply

   Asks for: Average pressure by zone, weighted by connections; hours of supply if intermittent.

   Why: Leakage scales with pressure and only happens while pressurised. Both the UARL and the per-connection indicators are corrected for them.

7. 7. Money

   Asks for: Tariff, marginal cost, operating cost, and any real losses that could be sold.

   Why: Turns volumes into an annual cost of NRW and its share of opex — the number that makes the case for action.

Uncertainty: choosing honest ± values

Every input carries a ± at 95% confidence: “the true value is within this percentage of my estimate, nineteen times out of twenty.” The calculator converts each ± to a standard deviation (÷1.96), adds the variances of everything that is summed or subtracted, and re-expresses the result as a ± on each component. Two consequences matter:

• Real losses are the residual of large numbers, so their ± is always the widest in the balance. A ±2% on system input and ±3% on billed consumption produces roughly ±10% on real losses when losses are a third of input, and worse when losses are small. That is physics, not a flaw.
• The ± you enter on a big number matters more than the ± on a small one. Sample-based estimates of illegal use can carry ±50% without moving the real-loss band much; a ±5% on system input moves everything.

Reading the results

The ledger

Four columns whose block heights are proportional to volume; the hatched band at the foot of each block is its ± range. Check it against your intuition: if apparent losses look bigger than real losses in a system with visible bursts, revisit the meter-error inputs.

ILI and its band

The ILI compares your real losses to the lowest technically achievable for a network of your length, connections, service-pipe length and pressure. 1.0 is the floor. The band (A1 to D) turns the number into a management message. Read it with its ±: an ILI of 3.0 ±30% could be 2.1 or 3.9 — band A2 or band B — and the honest statement is that you are “around B, most likely”.

NRW per connection and its band

The NRW matrix compares litres of non-revenue water per connection per day (when pressurised) with thresholds that rise with average pressure. This indicator includes apparent losses and unbilled use, so a utility can be A2 on leakage and C on NRW — a signal to look at metering and billing rather than pipes.

Value of NRW

Apparent losses and unbilled consumption are valued at the tariff by default (they would have earned revenue); real losses at marginal cost unless you enter a volume that could be sold. The total is shown as a share of operating cost — the number that most reliably gets a board's attention.

Carbon

Each source's treatment category sets its gCO₂ per m³; the volume-weighted average is applied to every component. The tonnes embodied in real losses are the number to put next to a leakage business case in a net-zero plan.

Using the what-if levers

Change one thing at a time. Pressure and supply hours move real losses through the UARL relationship (leakage is proportional to pressure in this model — a conservative assumption; real networks respond more strongly). Target ILI lets you ask “what would band A2 be worth?”. Target apparent losses per connection models a metering and billing programme, and the recovered volume becomes billed consumption. Coverage models network growth. Every lever feeds revenue, production cost, operating cost, NRW % and carbon.

Every formula

Limits and credits

A top-down balance is an estimate. Confirm real losses with night-flow measurement in district metered areas before they drive capital decisions. The UARL formula is unreliable below about 3,000 connections, 20 connections per km of mains, or 25 m average pressure — use litres per connection per day per metre of pressure there instead. Intermittent supply distorts everything unless expressed w.s.p.

The methodology is that of the IWA Water Loss Specialist Group. This tool follows the WB-EasyCalc (Roland Liemberger, 2006) and WCB-EasyCalc (David Pearson, Stuart Hamilton and Bambos Charalambous, 2023) workbooks, and is offered free in the same spirit. Please cite WaterBalanceCalc (www.waterbalancecalc.com) and the IWA WLSG when you use results.

Open the calculator