A water budget is only as useful as the evidence behind each number. A spreadsheet can balance perfectly and still mislead a committee if an unknown quantity has silently been entered as zero, if readings come from different timestamps, or if a calculated quantity is presented as though it were directly measured.
The practical goal is not to make every number equally precise. It is to make the quality and origin of every important number visible.
1. Unknown is not zero
Zero means the quantity is known to be zero for the defined period and boundary. Unknown means the quantity has not been established.
• No tanker arrived during the period, verified from the operating record: zero tanker inflow.
• Tanker records were not available for the period: unknown tanker inflow.
• No meter exists on an incoming source: incoming quantity may be unknown or estimated, but it is not automatically zero.
Replacing unknown values with zero can create a false water-balance residual and lead to the wrong operational conclusion.
2. Use explicit evidence classes
Measured — obtained directly from an appropriate sensor, meter or verified quantity measurement.
Calculated — derived mathematically from measured/configured inputs using a stated method.
Observed — a directly observed operating condition that is not itself a quantity measurement.
Estimated — a quantitative approximation based on incomplete measurement or a stated estimation method.
Supplier-declared — a quantity stated by a supplier or external party but not independently measured by the society.
Assumed — an input deliberately chosen for planning or scenario analysis rather than observed as fact.
Planned — a future quantity, event or operating condition expected or scheduled but not yet observed.
Testing — a method or capability being evaluated and not yet treated as a released or universally validated operating measurement.
Unknown — the quantity or condition has not been established.
A report may simplify these labels for its audience, but the underlying record should preserve the distinction.
3. Match the accounting boundary
The same physical movement can have different accounting treatment at different boundaries.
A UGT-to-OHT transfer is internal at a whole-society boundary. Counting it as new external supply would double-count water already inside the property.
At a tower or OHT boundary, however, that same transfer can legitimately be recorded as inflow to the selected tank or tower.
• boundary type;
• boundary name;
• water service;
• period start and end;
• included storage assets;
• included external sources and withdrawals.
4. Align timestamps before comparing storage
Storage is a stock at a particular time. Supply and withdrawal are flows over a period.
Opening storage and closing storage should use clearly defined timestamps. Source and outflow records should cover the same period. If a tank reading is stale or communications are missing, the record should show that rather than silently using it as though it were current.
TownEve tank readings are reported every 30 seconds by hardware design. During connectivity loss, hardware can retain up to 150 readings and upload retained data after reconnection. At the fixed 30-second cadence, the full buffer represents up to approximately 75 minutes of retained readings when the hardware remains powered and all 150 entries are available. Historical continuity can therefore be preserved across shorter communication interruptions, while remote visibility resumes after connectivity returns.
5. Keep measured depth separate from derived volume
TownEve directly senses water depth. Level and stored volume are derived from the configured tank geometry. Tank-specific 0% and 100% reference levels define the usable range.
This distinction matters because a depth reading can be measured while the resulting usable volume is calculated. The calculation also depends on correct tank geometry and configuration.
6. Do not confuse net storage change with gross flows
When inflow and outflow occur simultaneously and neither is independently measured, tank-level change shows the net effect. It does not uniquely reveal gross inflow and gross outflow.
Under a suitable controlled boundary, one side can sometimes be constrained so that storage change supports a derived estimate of the other side. TownEve has tested manually controlled no-inflow and valve-isolation methods and continues to refine them. Observation duration is deployment-specific.
A controlled-window estimate should remain labelled as calculated or estimated as appropriate. It is not automatically representative daily consumption and is not automatic leak detection.
7. Record configuration and unit quality
• unit of measure;
• tank geometry/configuration version where relevant;
• 0% and 100% usable-water references;
• meter/sensor identity;
• reading timestamp;
• calibration or verification status where applicable;
• source document or operational record;
• person responsible for manual entries.
Unit conversions should be explicit. Litres, kilolitres, cubic metres and percentages should not be mixed without a stated conversion.
8. Treat reconciliation differences as investigation triggers
A water-balance difference can arise from many causes: timing mismatch, unrecorded supply, unrecorded withdrawal, measurement uncertainty, configuration error, internal transfer treatment, overflow, drain-down, maintenance use, or a physical loss.
A non-zero result is a prompt to investigate. It is not, by itself, proof of leakage, theft, excess resident consumption or equipment failure.
9. Use a minimum data-quality check before decisions
□ The boundary and water service are explicit.
□ Opening and closing storage use compatible timestamps.
□ Unknown values have not been replaced by zero.
□ External supply is separated from internal transfers.
□ Measured and calculated quantities are distinguishable.
□ Units are consistent.
□ Tank geometry and usable-water references are current where volume is derived.
□ Planned supply is not presented as received supply.
□ A residual is not presented as a diagnosed cause.
□ Material data gaps are disclosed beside the decision.
10. A simple confidence statement
Instead of inventing a universal numerical confidence score, a society can state whether the record is:
Decision-ready — material quantities and boundaries are sufficiently evidenced for the stated decision.
Usable with limitations — the record supports directional planning but contains material estimates or gaps.
Investigation-only — the data identifies an issue or question but is not sufficient for a quantitative conclusion.
Insufficient — material boundary, timestamp or quantity information is missing.
The classification should be accompanied by the reason. It should not hide the underlying evidence classes.
Should a missing meter reading be entered as zero?
No. Zero is a known quantity. A missing or unavailable quantity should remain unknown, estimated or otherwise classified according to the evidence available.
Is tank volume directly measured by TownEve?
TownEve directly senses water depth. Stored level and volume are derived from measured depth and configured tank geometry.
Can tank-level data measure consumption?
It can support derived withdrawal or consumption estimates under suitable operating boundaries. If unknown inflow and outflow occur simultaneously, level change alone cannot uniquely separate them.
Does a water-balance discrepancy prove a leak?
No. It is an investigation trigger. The cause should be recorded only after supporting evidence establishes it.
Should planned water supply be included in actual supply?
No. Planned or expected supply belongs in forecasting until receipt is verified.