Checking the methods behind reported water totals
A water total can reconcile without its method being reproducible. Before treating withdrawal minus consumption minus discharge as a successful check, ask how each term was obtained and whether the calculation uses independent inputs. The arithmetic can identify a question; it cannot supply the missing measurement or boundary evidence.
This study retains three deliberately selected issuer cases from Nitium’s fixed historical research edition 0.12.4-local649-recovered-nbcc. It uses nine existing numeric observations and three existing disclosure analyses, adding no company-data records. The purpose is to compare review decisions, not water volumes, company performance or the prevalence of reporting practices.
The exact official editions were retrieved on 15 October 2026 and their selected pages reread for this study. They remain historical FY2024–25 cases, not a latest-report survey. No underlying meter logs, treatment worksheets or employee-based estimation inputs were inspected. An original calculation experiment below isolates a further question: can a formula hide an input change?
Three different review decisions
Mazagon Dock: retain the complete method note
Mazagon Dock Shipbuilders’ standalone FY2024–25 BRSR reports withdrawal 381,718, consumption 77,414.80 and discharge 304,303.20, all in kilolitres. Their arithmetic residual is 0.00. Footnotes use 20% consumption and 80% discharge assumptions. The continuation uses actual measured treated water for FY2024–25, assumes full recycling and excludes the corresponding volume from withdrawal for the discharge calculation. The prior year uses an operating-hours assumption. Source: MDL BRSR, printed/PDF 2 and 37–38.
The unadjusted shortcut 381,718 × 20% = 76,343.60 does not reproduce reported consumption. Preserve the reported figures and adjustment note. Request the worksheet, treated-water input and its place in each formula. Neither the zero residual nor the shortcut establishes independently measured consumption or a complete reconstruction of the disclosed method.
RCF: a source category needs its boundary
Rashtriya Chemicals and Fertilizers’ standalone FY2024–25 BRSR reports withdrawal 23,058,094 kiloliters, consumption 20,777,785 kilolitres and discharge 3,605,264 kilolitres. Withdrawal minus consumption minus discharge is −1,324,955 kilolitres. Its Others category contains 7,982,599 kilolitres produced in two in-house sewage-treatment plants, partly shared with Bharat Petroleum Corporation Limited. Source: RCF BRSR, PDF 1/29–30, printed 67/95–96.
The category explains why the withdrawal total needs a transfer-boundary review; it does not explain the residual or quantify the shared portion. Request the treatment-system boundary, transfer records and calculation coverage. Preserve the signed difference without assigning it to sharing, recycling, storage, leakage or error.
City Union Bank: equality does not resolve a dash
City Union Bank’s standalone FY2024–25 BRSR gives withdrawal and consumption as 93,966.48 kilolitres each. The consumption-cell footnote describes usage estimated from prescribed norms, employee count and working days; it supplies neither the norm nor calculation inputs and does not separately establish withdrawal’s measurement basis. Total discharge is a dash. Source: City Union Bank BRSR, unnumbered PDF 1/47–48.
Withdrawal less consumption is 0.00, but the three-term residual is unavailable. Do not replace discharge with zero or infer that all withdrawn water was consumed. Request the estimation inputs and separate withdrawal/discharge evidence, retaining the distinction between the note’s attachment and what remains unestablished.
What the residual is allowed to say
For this study, the residual is a signed diagnostic calculated only when all three reported terms are numeric, cover the selected period and use compatible units. It is kept outside the reported-observation fields. This subtraction is not a substitute definition of consumption, a physical-system balance or a rule requiring every published total to close.
Use three separate review outcomes: arithmetic available or unavailable; method sufficiently documented or unresolved; and boundary relationships supported or unresolved. A zero belongs only in the first outcome. Recording it as an overall “passed” status would hide the other questions. A nonzero value likewise supplies no automatic explanation or correction.
Original experiment: when the check cancels itself
All quantities and records in this experiment are invented. Two worksheets describe the same fictional annual reporting window and unit, kilolitres. Both display withdrawal W = 1,200, consumption C = 450 and discharge D = 750. The displayed residual R = W − C − D is zero in each. Their displayed numbers are identical; their input relationships differ.
Worksheet A imports all three quantities from separate supporting records. For this experiment only, none of those records calculates its value from either of the other two. Their physical accuracy is not assumed. Worksheet B imports withdrawal and discharge but calculates consumption as W − D. This is a formula-dependency comparison, not a claim that one method is suitable or that any issuer uses Worksheet B.
Now perform a controlled copy test. Change only the entered withdrawal from 1,200 to 1,220 kL and leave every other input record unchanged. In A, consumption remains 450 and discharge 750. The residual becomes 1,220 − 450 − 750 = 20 kL. The check exposes the changed relationship, without identifying which underlying record is correct.
In B, consumption automatically becomes 1,220 − 750 = 470 kL. The residual remains 1,220 − 470 − 750 = 0 kL. The added 20 kL has moved into a dependent output. A green zero would therefore fail to reveal the changed withdrawal entry, even though the worksheet recalculated exactly as designed.
The algebra explains the result: W − (W − D) − D = 0. With exact arithmetic and this formula, the residual must cancel for any permitted numeric W and D. Repeating it is an implementation check, not an independent test of the two input quantities. Rounding can introduce a small difference in other implementations; that would still not create independent evidence.
A has greater ability to expose this particular isolated change because the other inputs stay fixed. That does not make its records true or complete. If W and C both change by the same 20 kL in A, its residual is zero again. A common error or compensating change can therefore escape this check too. No residual alone certifies the underlying records.
Restore the baseline and remove discharge. A now lacks one required term, so the residual is unavailable. B lacks an input to consumption as well as the residual. Preserving a previously calculated 450 in B would conceal the lost dependency. Neither worksheet may replace the missing input with zero simply to keep its output populated.
Trace the calculation before requesting better data
Start with a small dependency map for the actual worksheet. Give each displayed quantity its source reference, reporting period, unit and role: supplied input, calculated result or unresolved source state. For a calculated result, name every input and adjustment it uses. “Estimated” is useful context but is not enough to reproduce a formula.
Test a copy, never the controlled source file. Change one input at a time and record which outputs move. A residual that remains fixed may be doing exactly what its formulas require. Ask what separate evidence can test the input instead: an approved record reconciliation, a supporting calculation or a confirmation of the covered system. The reporting team should not invent a substitute measurement.
Check physical scope separately. A shared treatment facility, internal return or transfer to another organisation needs an explicit relationship to the chosen reporting entity and period. A unit match does not establish that relationship. Retain unsupported allocations as unresolved rather than choosing a transfer amount that forces arithmetic closure.
Hand over the original source values, the dependency map, the reproducible diagnostic and the outstanding evidence request together. State what the check can detect and what it cannot. A reviewer should be able to remove a required input and see the dependent conclusions disappear, while the source’s original published representation remains intact.
Scope and related work
The reported water data hub provides broader historical observations. The water-metric guide addresses disclosure definitions, and the water balance workbook checks physical period accounts and storage carryforwards. This study instead tests whether the relationship among reported totals supplies an independent check.
The cases are purposive and cannot support a league table, a representative sector conclusion, a site-level water audit or a performance trend. The experiment does not reconstruct any company’s calculation. No residual is called a leak, saving, compliance result or verified physical flow. Report files, copied layouts, logos and third-party assurance material are excluded.
For help organising water-reporting evidence and targeted clarification requests, contact Nitium Advisory with a non-confidential outline of the reporting purpose and unresolved records. Technical measurement, plant operation and compliance decisions remain with their authorised specialists.
Prepared 15 October 2026 with AI-assisted research and internal source and calculation checks. The named issuers are not presented as clients or endorsers. This study does not independently assure their measurements. Original source editions and their qualifications remain controlling.
Comments