System boundary and cut-offs: what is in an avoided-emissions number, what is out, and how do you say so?

Short answer. The system boundary says which life cycle stages are counted, for the solution and for the incumbent. The cut-off says which small flows were left out, and on what threshold. In avoided emissions the boundary has to be equivalent on both sides. Excluding something is acceptable. Not saying so is not. Carbbin states the boundary, every exclusion and the reason on each case.

Why it matters

An avoided-emissions number is a difference. Leave a stage out on one side only, and the difference changes for reasons that have nothing to do with the technology. Leave it out on both sides without saying so, and nobody can tell whether it mattered. Most boundary problems are not wrong arithmetic. They are silent.

What the standards say

ISO 14067 treats the boundary as the basis for deciding which processes are in (6.3.4.1). The criteria, such as the cut-off, are identified and explained. A stage, process or flow may be excluded only if that does not significantly change the overall conclusions, and every exclusion is stated with its reasons and implications (6.3.4.1). Insignificant flows may be left out for practical reasons, but they are reported as data exclusions, and the effect of the cut-off on the result is assessed and described (6.3.4.3). Where no product category rule applies, decisions to include or exclude rest on a sensitivity analysis (6.4.5). Each emission is assigned to the life cycle stage in which it occurs (6.1). For comparisons, Annex B (normative) asks for an equivalent boundary and equivalent criteria for including inputs and outputs. Carbon offsetting is never part of the result (3.1.1.7, 6.3.4.1).

The ILCD Handbook asks for a system boundary diagram, and in comparative studies the stages included for each option (Provision 6.6 IV, p. 106). It asks for a list of exclusions, kept up to date until the end of the study (Provision 6.6 V, p. 106). Offsets and similar measures outside the analysed system are excluded from the boundary, the inventory and the interpretation (Provision 6.6 VII, p. 106). The cut-off is a percentage of the approximated overall impact; for comparative studies it also applies to mass and energy (Provision 6.6 VIII, p. 107), and in a climate-only study it relates to climate change (p. 107, note 85). Parts that are truly identical in the compared systems may be left out of all of them; parts that are only similar stay in (Provision 6.10 V, p. 146).

Project Frame (Guidelines 2024) puts it in two steps. First think across the whole life cycle: "a cradle-to-grave system boundary is essential in qualitative analysis" (Section 1, p. 7). Then quantify what is material. The materiality screen rates each effect; if effects cluster at the very low and very high ends, those under 10% may not be material, but if many sit near 10%, they are treated as material when their total may be significant, and an effect the analyst remains unsure about moves to full quantification (Section 3, p. 20). Sources that are equal for both sides drop out of the difference (Section 3, p. 17), and "the reader should understand which, if any, Solution Effects were excluded from quantification and why" (Section 3, p. 17).

How Carbbin does it

• Carbbin screens every life cycle stage of both the solution and the incumbent, cradle to grave, before quantifying.
• Carbbin quantifies the stages that differ materially and states the rest as excluded, with the reason.
• Carbbin uses the same boundary on both sides. If a stage is in for the solution, it is in for the incumbent.
• Carbbin states the boundary, the exclusions and the cut-off on every case, so a reviewer can see what was left out before asking.
• Carbbin does not drop a flow because it is hard to quantify. Where the flow has no emission factor of its own, Carbbin uses the closest documented proxy and flags it, so a missing data point does not quietly become a zero.
• Carbbin places end of life in the year it happens, on the most probable route for those materials and markets, and discloses the route.
• Carbbin keeps offsets and purchased credits out of the number. They are not part of either system.

Common mistakes

• A boundary declared cradle to grave that the inventory does not fill, with no transport or end-of-life lines on either side. The label says complete; the numbers say otherwise.
• Different boundaries on each side, such as a cradle-to-grave solution against a use-phase-only incumbent.
• Fuel factors that cover combustion only under a boundary that claims the full life cycle. The upstream share of a fuel is often material.
• "Not estimable" used as a reason to exclude a flow that could exceed the cut-off. ISO 14067 asks for the effect of the cut-off to be assessed (6.3.4.3).
• An exclusion written in one segment that silently applies to all of them, or the reverse.
• Leaving out parts that look identical but are not, such as the same material used in different components (ILCD Provision 6.10 V).
• Credits for recovered energy at end of life buried in the headline, with no line showing what the result is without them.
• Offsets inside the number. Neither ISO 14067 nor ILCD allows it.

Sources

• ISO 14067:2018. Greenhouse gases: Carbon footprint of products. Requirements and guidelines for quantification. Clauses 3.1.1.7, 6.1, 6.3.4.1, 6.3.4.3, 6.4.5 and Annex B.
• European Commission, JRC. ILCD Handbook: General guide for Life Cycle Assessment, Detailed guidance. First edition, March 2010. Provisions 6.6 and 6.10.
• Project Frame. Project Frame Methodology: Evaluating Greenhouse Gas Impact for Early-Stage Investments. December 2024. https://projectframe.how/s/2024-Project-Frame-Methodology.pdf