Per-unit sustainability starts with boundaries and meters

The GHG Protocol requires all emission sources and activities within the chosen inventory boundary to be accounted for and reported, with exclusions disclosed and justified.[S30] The practical sequence follows directly. Choose and document the boundary first, then populate it. A factory cannot demonstrate completeness by adding up whichever records happen to be available. Availability is a data-quality issue, not a reason to redraw the physical or organisational scope without saying so.

Three questions should remain separate. Which sites and activities belong to the operating review? Which sources belong to the organisation's emissions inventory? Which life-cycle stages belong to a product assessment? A cut, make, trim and finishing site can answer the first with a gate-to-gate operating account. That account may feed the other two, but it does not answer them by itself. Upstream yarn and fabric production, outbound transport, garment use and end-of-life treatment do not become factory observations merely because the factory makes the garment.

Product accounting makes the additional definition explicit. The GHG Protocol Product Standard requires the studied product, unit of analysis and reference flow to be defined.[S32] A reference flow might describe the quantity of product needed to fulfil the chosen unit of analysis. That is different from a plant KPI denominator, which might be good garments accepted during a shift. Confusing them can produce a numerically plausible but materially false claim.

GFT would maintain a boundary register for every resource account. It names the sites, buildings, production processes, ancillary services, energy carriers, water uses and waste streams included. It records outsourced operations, shared utilities, on-site generation, storage, transfer points and any interface where another organisation holds the source evidence. It also identifies exclusions and their rationale. An unknown item remains an unresolved item, not a silent zero.

The period belongs to the same register. A meter interval, invoice period, production shift and reporting month rarely align by accident. The accounting rule should state how readings are assigned to the operating calendar, how opening and closing readings are obtained, and how late invoices or waste tickets are treated. A monthly intensity figure should not combine a billing-period numerator with a calendar-month denominator without disclosing and testing the effect of that choice.

Boundary changes need version control. Installing a compressor, outsourcing garment washing, adding a canteen or bringing a previously shared boiler inside the account can change resource use without any change in process efficiency. The register should retain the effective date, reason, responsible approver and policy for historical comparison. Earlier results should not be silently recalculated under a new scope. Where a like-for-like restatement is possible, both the original and restated bases should remain visible.

The production boundary must match the resource boundary as closely as the decision requires. If a meter serves both sewing and an unrelated tenant, total site output is not an adequate denominator until the shared load is resolved. If a boiler serves a process that is outside the garment account, its fuel cannot simply be divided by garment output. Allocation may be necessary, but it needs a documented physical or operational basis and an uncertainty statement.

This is not bureaucracy around an otherwise simple ratio. It is the control that makes the ratio interpretable. A narrow, well-declared account is useful. A broad claim with an unknown boundary is not.

The IEA links energy-indicator design with the data required to produce those indicators and with surveying, metering and modelling practice.[S39] For a new factory, this means the metering plan should follow the energy-use map rather than a procurement catalogue. The main incomer establishes the site account. Additional meters are justified where they separate a significant use, test a suspected loss, allocate a shared load or verify an improvement action.

The energy-use map begins with physical inspection. It traces electricity, fuel and any purchased thermal energy from receipt to the systems that consume them. Production equipment, compressed air, boilers, HVAC, lighting, pressing and other loads are included where they actually exist. The map distinguishes a measured branch from an inferred branch and records loads that serve more than one process. It should be possible to walk from a line on the map to a meter, invoice, calculation or explicit estimate.

Each meter then needs an operational register. The register names the meter identifier, physical location, boundary served, energy carrier, reading unit, multiplier, reading convention and responsible owner. It states whether the display is cumulative or interval-based, how often the value is captured, and which review uses it. Calibration or verification status, access constraints and the treatment of an unavailable reading belong there as well. A device number alone does not establish what the reading means.

Read frequency should follow the decision. A monthly main-meter close can support absolute reconciliation and a plant KPI. A daily compressor reading may support investigation of an abnormal load. Interval data may be warranted where the timing of demand matters. Capturing frequent values without a review question creates storage and checking work but little control. Conversely, a monthly total cannot prove when a short-lived abnormal load occurred.

Cumulative meters require particular care at replacement and rollover. The record should retain the final reading of the removed device, the initial reading of its successor, the change timestamp and any multiplier change. Consumption for the interval is calculated only after those events are reconciled. Treating the new display as a continuation of the old one can create an apparent negative or an implausible surge. Such corrections are normal accounting controls, not analytical embellishment.

Submeters should reconcile to the main boundary without being forced to match it exactly. The close identifies metered uses, explicitly estimated unmetered uses and the residual. The residual is investigated when it exceeds a defined review threshold or changes unexpectedly. It should not be distributed across departments merely to make the schedule balance. A visible residual tells the factory where its energy map remains incomplete.

Utility invoices provide an independent commercial reconciliation, but they do not replace physical readings. Invoice quantity, tariff period, estimated-billing status and any correction should be compared with the meter account. Differences can arise from period cut-offs, multipliers, estimated bills or meter identity. Finance and engineering should resolve the same exceptions rather than maintain separate totals for external payment and internal management.

Missing data need a status, not just a substitute number. An estimated interval should identify the reason, method, author and evidence used, such as adjacent validated readings or a documented operating relationship. The replacement value remains distinguishable from a measurement in every later KPI. When the actual reading arrives, the correction is versioned and the affected results are restated under a controlled rule.

Meter ownership should end in action. Engineering maintains the energy-use map and meter register. Operations explains changes in production conditions. Finance reconciles purchased quantities. The sustainability or performance owner checks that the published KPI uses the approved boundary and denominator. A small team can hold several of these roles, but the review should still show who can accept an estimate, approve a correction and authorise an operating response.