Key takeaway
A residual can guide investigation, but its classification depends on metering, billing and authorized-use evidence.
Define the boundary before calling volume lost
A utility archive can contain system input, customer readings, billing adjustments and estimated operational use. Subtracting billed volume from input produces a difference, but it does not identify one physical failure. Start with the system boundary, common period and compatible volume units. Explain imported or exported supplies, storage effects and any estimates. A number with an undefined boundary cannot be compared reliably across periods or systems.
EPA's public-water-system guidance distinguishes billed and unbilled authorized consumption, apparent losses from issues such as metering or data handling, and real physical losses. Non-revenue water includes unbilled authorized use as well as apparent and real losses. Its top-down approach can provide a starting estimate, with further work needed to investigate. The July 2013 framework was read 10 October 2026; it is not an observed audit of a particular utility.
Complete a simple annual balance
This hypothetical annual example assumes consistent boundaries, cubic meters, no separately changing storage term and documented volume categories. Input is 100,000 m³. Billed authorized consumption is 80,000 m³ and unbilled authorized consumption is 5,000 m³, so total authorized consumption is 85,000 m³. The residual water loss is 15,000 m³. If an apparent-loss estimate is 4,000 m³, the remaining real-loss estimate is 11,000 m³.
Non-revenue volume is 20,000 m³ because only 80,000 m³ was billed. Calling that full 20% a physical leakage rate would wrongly include unbilled authorized use and estimated apparent loss. The 11,000 m³ residual is also an estimate, not a measured leak or proof of its location. Retain the evidence and uncertainty behind each component before using the balance to prioritize an investigation.
| Annual component | Illustrative volume | Calculation or basis |
|---|---|---|
| System input | 100,000 m³ | Boundary-aligned input |
| Billed authorized | 80,000 m³ | Billing and meter evidence |
| Unbilled authorized | 5,000 m³ | Documented operational-use estimate |
| Total water loss | 15,000 m³ | 100,000 − 80,000 − 5,000 |
| Apparent loss estimate | 4,000 m³ | Separate estimation evidence |
| Residual real-loss estimate | 11,000 m³ | 15,000 − 4,000 |
| Non-revenue water | 20,000 m³ | 100,000 − 80,000 |
Preserve estimate lineage rather than filling gaps silently
For each component, record the measurement source, adjustment method, effective period, unit, reviewer and whether it is metered or estimated. Link corrections to original records. If a billing adjustment concerns last year's consumption, show its allocation rather than counting it again as this year's flow. Keep customer-meter inaccuracies distinct from system-input uncertainty; both can affect a balance but do not represent the same event.
For a useful internal replay, trace one operational-use estimate to its method and one apparent-loss component to the evidence supporting it. When supporting information is absent, label the component unverified and retain the resulting uncertainty. A balanced spreadsheet obtained by allocating every unexplained difference to leakage hides exactly what the audit should investigate. The review should show where the residual comes from and where its interpretation stops.
Show how one uncertain input changes the conclusion
In this hypothetical sensitivity, suppose only the input estimate varies by 2,000 m³ while all other stated components remain fixed. At 98,000 m³ input, residual real loss becomes 9,000 m³. At 102,000 m³, it becomes 13,000 m³. This sensitivity calculation illustrates dependence on one assumption; it is not a full uncertainty interval, because the other components may also be uncertain and related.
The completed review therefore describes an 11,000 m³ point estimate with a disclosed input sensitivity. It does not promise that repairing pipes will recover that amount or produce a particular cash saving. Actual intervention choices need engineering, metering and operational evidence. A reader evaluating the archive should ask whether historical balances retain their estimate versions, later revisions and boundaries so a change in methodology is distinguishable from a change in the water system.
AWWA’s method resource explains that the annual audit’s real-loss residual provides limited insight for leakage control and describes a separate component-analysis step. That distinction helps choose the next task: improving an uncertain audit input and investigating leakage mechanisms are related but different work. The example’s sensitivity calculation can identify a measurement question; it cannot replace the additional engineering evidence needed to locate losses or select an intervention.
Use the balance as an evidence map
The deliverable is a category-level water-balance sheet linked to source methods and unresolved components. It can reveal that a history supports non-revenue-volume reconstruction while physical-loss classification remains incomplete. That is different from the billing-interval reconciliation guide, which checks a meter-to-bill relationship rather than assigning system-wide authorized use and loss categories.
Inventory can describe the categories; Due diligence can clarify which analysis a recipient seeks. Begin with synthetic values and keep sensitive network, customer and facility details in authorized systems. VOID has no upfront seller referral fee and may receive conditional receiving-program compensation. Obtain permission for exact metadata and a named recipient before an introduction. Samples, evaluation and licensing remain separate decisions, and this balance supplies no market-value or buyer-demand claim.