Every utility produces more water than it sells. The gap — the water that is pumped, treated and paid for but never billed — is Non-Revenue Water, and in many networks it is a quarter to a half of everything that enters the pipes. It is also the most misunderstood number in the business: managers quote a single percentage, boards set targets against it, and almost all of them are measuring the wrong thing. This is the discipline that fixes that — a standard way to account for every drop, an honest index of how leaky the network really is, and the economics that tell you when to stop chasing leaks.

1 · What Non-Revenue Water actually is

Non-Revenue Water (NRW) is simply the difference between the water a utility puts into its network — the system input volume — and the water it bills its customers for. It is tempting to read that as "leakage," but NRW is broader and more interesting than that. It is made of three very different things[1]:

The reason the distinction matters is that the three have completely different cures and completely different costs. A litre of real loss costs only what it cost to produce and pump. A litre of apparent loss is far more expensive — it was treated, pumped and delivered, and it should have earned the full retail tariff, so you lose the margin too. Lumping them into one percentage hides exactly the information you need to act[2].

Why "% NRW" is a poor KPI A percentage moves when either number moves. Sell more water in a wet year and your NRW percentage falls while you lose exactly the same volume. Two utilities can both report "30% NRW" while one leaks twice as much per kilometre of main as the other. Percentage is fine for a headline; it is useless for engineering. To manage loss you need volumes, split by type, and a physical index that accounts for the size and pressure of the network — the subjects of this article.

2 · The IWA water balance

The international standard for accounting water loss is the IWA / AWWA water balance[1][3]. It is a single bookkeeping table that starts from the system input volume and divides it, step by step, until every drop is in exactly one box. Read it left to right and two columns fall out the right-hand side: Revenue Water (what you billed) and Non-Revenue Water (everything else).

The IWA water balance — every drop of system input in exactly one box System Input Volume Authorised Consumption Water Losses Billed Authorised billed metered + billed unmetered consumption Unbilled Authorised flushing, firefighting, own use Apparent Losses meter under-registration, billing/data error, theft Real Losses background leakage, reported + unreported bursts, tank overflows Revenue Water Non-Revenue Water (NRW) unbilled + apparent + real
Original schematic of the IWA/AWWA M36 water balance. Revenue Water is only the billed-authorised box; everything else — unbilled authorised consumption, apparent losses and real losses — is Non-Revenue Water.

The balance is built top-down: you meter the system input, you total what you billed, and you estimate the smaller boxes (unbilled use, meter error, unauthorised use) so that the components add up. The single most valuable output is not the percentage — it is the split between real and apparent losses, because that tells you whether your money belongs in the field (finding leaks) or in the back office (fixing meters and billing).

3 · Interactive: build your water balance

Set the system input and the three loss components and watch the balance assemble. The green block is the revenue you actually collect; the grey, amber and red blocks together are Non-Revenue Water. Notice how a small slice of apparent loss can cost as much revenue as a much larger slice of real loss — because apparent loss is billed at the full tariff, not just production cost.

The water balance, as volumes
System input volume split into billed authorised (revenue water), unbilled authorised, apparent losses and real losses. NRW is everything that is not billed. The value of loss applies the full retail tariff to apparent losses (lost revenue) and the marginal production cost to real losses and unbilled authorised use.
All water entering the network (own sources + imports).
Physical leakage — the field problem (pressure, ALC, repairs).
Meter error, billing error, theft — the back-office problem.
Flushing, firefighting, utility own use.
Non-Revenue Water
28 %
NRW volume
5,600 m³/d
Value of loss
k$/yr
Verdict

A 20,000 m³/d system at 18% real + 7% apparent + 3% unbilled is 28% NRW. Hold the volume and drag apparent loss up by a few points: the revenue value climbs far faster than the same change in real loss, because every apparent litre is priced at the retail tariff. That is why a leak-only programme can miss half the money.

4 · Real vs apparent — two problems, two cures

Because they cost differently and are fixed differently, the two loss families need separate programmes:

AspectReal (physical) lossesApparent (commercial) losses
Where the water goesInto the ground — leaks & burstsTo a customer, unbilled
Unit costVariable production + pumping costFull retail tariff (lost revenue)
Driven byPressure, pipe age/material, repair speedMeter age/sizing, data systems, theft
Found byDMA night flow, acoustic survey, step testsMeter testing, data audits, field inspection
Fixed byPressure management, ALC, pipe renewalMeter replacement, billing reform, enforcement

The real-loss programme is the engineering one, and it leans directly on the two previous articles: pressure management and DMAs to measure and suppress leakage, and the network topology that decides how easily you can zone the system in the first place. The rest of this article is about measuring the real-loss problem honestly and knowing how far to push it.

5 · Measuring real losses properly — UARL and the ILI

How leaky is "leaky"? Litres per day means nothing without context: a 2,000-km network will always lose more than a 50-km one. The IWA solution is to compare your Current Annual Real Losses (CARL) against a physics-based floor — the Unavoidable Annual Real Losses (UARL): the lowest leakage a well-run network of that exact size and pressure could achieve, with today's technology and unlimited budget[4].

\[ \text{UARL (L/day)} = (18\,L_m + 0.8\,N_c + 25\,L_p)\times P \]

where \(L_m\) is mains length (km), \(N_c\) the number of service connections, \(L_p\) the total length of private underground pipe (km, from the street edge to the customer meter), and \(P\) the average operating pressure (m). The three coefficients capture the three places water unavoidably escapes: the mains, the connections, and the customer's own pipe. Divide your real losses by this floor and you get the dimensionless Infrastructure Leakage Index[4][5]:

\[ \text{ILI} = \frac{\text{CARL}}{\text{UARL}} \]

An ILI of 1.0 is the theoretical best; an ILI of 4 means you are losing four times the unavoidable minimum. Because it is normalised for length, connections and pressure, the ILI lets you compare a Riyadh network with a London one, or track one network honestly across the years even as it grows. The World Bank maps it onto management bands[5]:

ILI bandILI (developed)What it means
A1 – 2Excellent — further loss reduction may be uneconomic
B2 – 4Good — scope to improve in specific areas
C4 – 8Poor — intensify pressure management and ALC
D> 8Very inefficient — resource-poor management of loss

6 · Interactive: UARL & the Infrastructure Leakage Index

Enter your network's size and pressure to compute the unavoidable floor, then set your current real losses to read the ILI. (For clarity this calculator assumes meters at the property boundary, so the private-pipe term \(L_p\) is taken as zero — add it where customer meters sit well inside the plot.) Watch how the floor itself rises with pressure: drop the average pressure and even the unavoidable minimum falls, which is exactly why pressure management is the first move.

Current real losses vs the unavoidable floor (ILI)
UARL = (18·Lₘ + 0.8·N_c)·P litres/day. The green bar is the unavoidable floor for your network; the red bar is your current real losses; their ratio is the ILI. The recoverable leakage is the gap between them.
Total length of distribution mains.
Number of service connections on the network.
Average operating pressure across the network.
Measured physical loss (e.g. from DMA night-flow analysis).
UARL (floor)
923 m³/d
Recoverable
3,077 m³/d
ILI
4.3
Band

A 250-km, 20,000-connection network at 45 m has an unavoidable floor near 920 m³/d. Losing 4,000 m³/d puts the ILI at about 4.3 — band C, "intensify control." Now drop the pressure to 35 m: the floor falls and, in reality, so would your actual losses (via the FAVAD law) — the two moves compound.

7 · The four pillars of leakage management

Real losses are held down by four levers acting together — the IWA's "four pillars." Take away any one and leakage creeps back up toward the unmanaged level[2][6]:

The life of a leak Every litre of unreported leakage is lost over three phases: awareness (how long before you know a leak exists), location (how long to find it), and repair (how long to fix it). DMAs and night-flow monitoring crush the first two; a responsive repair crew crushes the third. The volume lost is leak rate × total run time — so cutting run time is as powerful as cutting the number of leaks.

8 · The economic level of leakage

You can almost always find and fix one more leak — but should you? Beyond a point, the cost of looking harder exceeds the value of the water you would save. That crossover is the Economic Level of Leakage (ELL): the leakage at which the total cost to the utility is lowest[6][7].

Two costs pull in opposite directions. The value of water lost rises with leakage — more leakage, more wasted production. The cost of active leakage control rises as leakage falls — to hold a lower level you must survey more often and chase smaller leaks. Add them and the total is a U-shaped curve; its lowest point is the ELL. Pushing leakage below the ELL is not "better" — it is spending two pounds to save one.

9 · Interactive: the economic level of leakage

The red curve is the rising value of water lost; the amber curve is the cost of the leakage control needed to hold each level; the navy curve is their sum. The marker sits at the economic level of leakage — the minimum of the total. Move your current leakage to see how much you could save by reaching the ELL — or whether you have already over-invested past it.

Total cost of leakage vs. leakage level
Value of water lost = unit cost × leakage (rising). Cost of active leakage control ≈ budget × (reference ÷ leakage) (rising as leakage falls). Total = the sum; its minimum is the economic level of leakage (ELL).
Marginal cost of producing & pumping a lost m³ (use tariff for apparent loss).
Annual ALC spend to hold leakage at the 1,000 m³/d reference level.
Where you are now — compare against the economic optimum.
Economic level (ELL)
1,350 m³/d
Total cost at ELL
0.59 M$/yr
Your total cost
0.79 M$/yr
Potential saving

With water at $0.60/m³ and a $400k/yr control budget, the economic level sits near 1,350 m³/d. Sitting at 3,000 m³/d costs about $0.2M/yr more than necessary — that gap is your business case for more ALC. Push the unit cost of water up (scarcity, energy, desalinated supply) and the optimum moves left: the more your water is worth, the harder it pays to chase the leaks.

10 · The NRW management checklist

The one-line summary Stop quoting a single percentage. Account for every drop with the IWA water balance, separate the field problem (real losses) from the back-office problem (apparent losses), measure how leaky you truly are with the ILI, and drive leakage down to the economic level — no further — with pressure management, active leakage control, fast repairs and renewal.

References & standards

  1. American Water Works Association. Manual M36 — Water Audits and Loss Control Programs (the IWA/AWWA water balance, NRW components, validity scoring).
  2. Farley, M. & Trow, S. Losses in Water Distribution Networks: A Practitioner's Guide to Assessment, Monitoring and Control. IWA Publishing.
  3. IWA Water Loss Specialist Group. Best Practice Standard Water Balance and terminology (Alegre et al., performance indicators).
  4. Lambert, A.O. & Hirner, W. (2000). Losses from Water Supply Systems: Standard Terminology and Recommended Performance Measures — the UARL formula and the Infrastructure Leakage Index.
  5. Liemberger, R. & McKenzie, R. Accuracy Limitations of the ILI — Is it an Appropriate Indicator for Developing Countries? (World Bank Institute ILI management bands).
  6. Thornton, J., Sturm, R. & Kunkel, G. Water Loss Control. McGraw-Hill — four pillars, ALC, and economic intervention.
  7. UKWIR / Tripartite Group. Best Practice Principles in the Economic Level of Leakage Calculation.
  8. World Bank. The Challenge of Reducing Non-Revenue Water in Developing Countries (Kingdom, Liemberger & Marin) — PPP and programme economics.
#NonRevenueWater #NRW #WaterLoss #WaterBalance #LeakageControl #RealLosses #ApparentLosses #ILI #InfrastructureLeakageIndex #UARL #EconomicLevelOfLeakage #ELL #PressureManagement #DMA #ActiveLeakageControl #AWWA #M36 #IWA #PipelineDesign #WaterInfrastructure