Environmental Impact Calculation Methodology

A white paper for users, domain experts, and verifiers

---

1. Introduction

This document explains how Dawn calculates the environmental impact of products, parts, and assemblies. It is written for readers who need to understand, review, or verify the results the platform produces — sustainability professionals, LCA practitioners, auditors, and third-party verifiers — and assumes no prior familiarity with the product.

Dawn is a platform for modelling the environmental footprint of physical products across their life cycle. The footprint of a finished product is built up from many individual contributions: the raw materials it contains, the components purchased from suppliers, the energy used in manufacturing, the transport of goods between sites, the packaging, the use phase, and what happens at end of life. Each of these contributions is calculated separately using established life cycle assessment (LCA) principles, and the results are then aggregated into a complete picture.

This paper describes:

• where the underlying environmental data (emission factors) comes from,
• the fundamental calculation that converts a physical quantity into an environmental impact,
• how individual results are rolled up through a product's bill of materials,
• how each life cycle module (transport, energy, waste, end of life, etc.) is computed,
• how datasets are selected and verified, including AI-assisted matching,
• the assumptions, limitations, and traceability mechanisms a verifier should be aware of.

2. Key Concepts and Terminology

Term	Meaning
Part	Any item in a product structure: a purchased component, a manufactured sub-part, a raw material, or an auxiliary item such as a transport service or a unit of electricity.
Assembly / Bill of Materials (BOM)	The hierarchical structure describing which parts a product is made of, in which quantities. Assemblies can be nested (a product contains sub-assemblies, which contain parts, and so on).
Assembly variant	An alternate sellable configuration of the same assembly. Only BOM composition varies per variant; utilities, packaging, transport, use phase, end of life, and other assembly-level data are shared across all configurations.
Core BOM	The base bill of materials stored on the assembly itself — the configuration used when no variant is selected.
Variant delta	A per-variant change to the core BOM: an added part or child assembly, a quantity or production-loss override of a core line, or an exclusion of a core line. Deltas are stored separately; the core BOM is never duplicated.
Activity / dataset	A record from an LCA database describing the environmental impact of producing or providing one unit of something — for example, "1 kg of aluminium ingot, produced in Europe" or "1 tonne-kilometre of road freight".
Emission factor / impact factor	The amount of environmental impact (for one impact category) caused per unit of an activity. For example, kg CO₂-equivalents per kg of material.
Impact category / indicator	A specific type of environmental pressure, such as climate change, water use, or acidification. Each dataset carries values for a full set of indicators, not just carbon.
Functional unit	The reference quantity of a dataset (e.g. 1 kg, 1 kWh, 1 tonne-kilometre). All emission factors in a dataset are expressed per functional unit.
Life cycle module	A grouping of impacts by where they occur in the product's life: raw materials, inbound transport, production energy, distribution, use phase, end of life, and so on.
Allocation	A percentage split used when a quantity must be divided across several options — for example, when a part is sourced from two suppliers, or when a material is split across several end-of-life treatment routes.

3. Data Foundations: Where Emission Factors Come From

Every impact calculation in Dawn is grounded in an emission factor from one of the following sources:

1. Ecoinvent — the leading international LCA background database. Dawn always make the latest availalable version of Ecoinvent available, and support automatically updating background datasets when Ecoinvent versions update:

• Cut-off (allocation, cut-off by classification): recycled materials enter burden-free; the producer of waste bears no credit for downstream recycling. This is the most widely used model for product footprinting and EPDs.
• Consequential: models the consequences of a marginal change in demand. Available for organisations that require it, and currently only for expert level users.

2. Environmental Footprint (EF) 3.1 datasets — the European Commission's reference datasets for Product Environmental Footprint (PEF) studies.
3. EPD datasets (EN 15804) — verified Environmental Product Declarations, used when a supplier-specific or product-specific declaration is preferable to a generic background dataset.
4. Custom datasets — organisation-specific activities, for example primary data from a supplier, imported into the platform and used exactly like database datasets.

Each dataset carries impact values for the full set of impact indicators defined by its impact assessment method. The platform's default impact assessment method is EF 3.1 (Environmental Footprint 3.1), the method underlying the EU PEF framework. EPD data follows the EN 15804+A2 indicator set, which is itself based on EF 3.1, allowing results from both sources to be reported consistently.

Database access is controlled per organisation, so a calculation only ever draws from the databases the organisation is licensed to use.

4. The Fundamental Calculation

Every impact figure in the platform — no matter how complex the product — is ultimately built from one elementary calculation:

Impact = Quantity × Emission factor

More precisely, for each impact indicator i:

Impact(i) = Q × EF(i)

where:

• Q is the physical quantity of the activity consumed, expressed in the dataset's functional unit (kg of material, kWh of electricity, tonne-kilometres of transport, …),
• EF(i) is the dataset's emission factor for indicator i, per functional unit.

This calculation is performed across the entire indicator set simultaneously — climate change, water use, acidification, eutrophication, resource use, toxicity, and the remaining EF 3.1 categories — not only for carbon.

4.1 Unit conversion

The quantity entered by a user is converted to the dataset's functional unit before the multiplication. Examples of conversions applied:

• mass: grams ÷ 1 000 → kg; tonnes × 1 000 → kg
• energy: MJ × 0.27778 → kWh
• transport work: kg·km ÷ 1 000 → tonne-km

So if a dataset is expressed per metric tonne and a part weighs 2 kg, the quantity used in the calculation is 0.002.

4.2 The two impact figures reported

From the full indicator set, two headline figures are derived for every part, module, and product:

(a) Carbon footprint (GWP100). The climate change indicator, expressed in kg CO₂-equivalents using the 100-year Global Warming Potential. This is the total climate change value (covering fossil, biogenic, and land-use related greenhouse gas emissions as defined by EF 3.1 / EN 15804). It is reported without any normalisation or weighting — it is a direct physical quantity.

(b) EF single score. A weighted aggregate across all 16 EF 3.1 impact categories, calculated following the official EF method:

EF single score = Σ over categories i of:  Impact(i) / NF(i) × WF(i)

where:

• NF(i) is the EF 3.1 normalisation factor for category i — the average annual impact of one person globally (e.g. 7 550 kg CO₂-eq/person/year for climate change). Dividing by it expresses each impact in "person-equivalents".
• WF(i) is the EF 3.1 weighting factor — the relative importance assigned to the category by the EF method (e.g. 21.06 % for climate change).

The normalisation and weighting factors used are the official EF 3.1 values published by the European Commission. The resulting score is expressed in points (Pt); for readability the platform displays it multiplied by 1 000, i.e. in millipoints (mPt).

Two safeguards apply when computing the single score:

• Sub-category indicators are excluded so the same impact is never counted twice (e.g. "climate change – fossil" is not added on top of "climate change – total").
• Inventory indicators are excluded. EN 15804 EPDs also report inventory-type figures (resource use, waste flows, biogenic carbon content) that are not impact categories. These are shown for transparency but never enter the single score.

4.3 Worked example

A purchased steel bracket weighs 0.5 kg and is matched to the dataset "steel, low-alloyed, hot rolled — Europe", expressed per kg, with a climate change factor of 2.1 kg CO₂-eq/kg.

Carbon footprint = 0.5 kg × 2.1 kg CO₂-eq/kg = 1.05 kg CO₂-eq

The same multiplication is applied to all other indicators in the dataset (water use, acidification, …), and the EF single score is then computed from the full normalised, weighted set as described above.

5. Rolling Up the Product Structure

A product's total impact is not a single multiplication — it is the structured aggregation of many distinct contributions, each calculated with the elementary formula from Section 4 and each carrying its own life cycle stage. Three kinds of contributions enter the roll-up:

1. Cradle-to-gate impacts of the parts themselves — every part in the bill of materials contributes its per-unit material/component impact (Section 6.1), scaled by the quantity of that part actually consumed.
2. Module impacts attached to individual parts — a part is more than its material content: it may also carry inbound transport from its suppliers, production energy, and production waste treatment (Sections 6.2–6.5). Each of these is calculated at the part level and scaled by the same effective quantity as the part itself, so a screw used 8× also contributes 8× its transport and 8× its production energy.
3. Module impacts attached to the assembly or product — contributions that only exist at the level of the finished item: transport to customer, packaging, installation, use-phase energy, spare parts, disassembly, and end-of-life treatment of the complete material inventory (Sections 6.6–6.7).

Throughout the roll-up, every contribution retains two attributes: its full indicator set (all EF 3.1 categories are aggregated in parallel, not just the headline figures) and its life cycle stage assignment, so the final result can always be decomposed both by impact category and by stage. The aggregation also records which part and which dataset each portion of the total came from, which is what enables the drill-down described in Section 9.

5.1 Quantity propagation

Each parent–child link in the BOM carries a quantity ("4 screws per bracket"). Quantities multiply down the tree:

Effective quantity of a part = quantity per parent × effective quantity of the parent

A screw used 4× in a bracket that appears 2× in the product has an effective quantity of 8. (When a manufacturing step scraps part of its input, the surviving fraction also enters this multiplication — that refinement is described in Section 5.2.)

The total impact of an assembly is then:

Impact(assembly) = Σ over all parts p of:        Impact per unit(p) × effective quantity(p)
                 + Σ over part-level modules m:  Impact(m) × effective quantity of the owning part
                 + Σ over assembly-level modules: Impact(module)

Nested sub-assemblies are handled recursively with the same rules: a child assembly's entire contribution — parts, part-level modules, and its own assembly-level modules — is scaled by the quantity of that sub-assembly used in the parent. Where a part appears in several places in the structure, each occurrence is counted with its own effective quantity; circular references are detected and excluded so no contribution can be counted twice.

5.2 Production loss (yield)

Manufacturing steps rarely convert 100 % of input into output. Each step in the product structure can carry a production loss percentage — the share of input that is scrapped rather than ending up in the finished product. For a step with loss rate L:

Input quantity required  = declared quantity
Quantity lost as scrap   = declared quantity × L
Quantity reaching output = declared quantity × (1 − L)

Conceptually, the platform asks two questions at every step of the product structure:

1. How much had to be produced and brought in? — this determines the environmental burden, because everything that was made carries an impact, whether or not it survived the process.
2. How much actually continues onward into the product? — this determines the quantities used for everything further down the chain.

These two amounts differ exactly by the scrapped share. The platform therefore splits every quantity at the point where the loss occurs:

• The scrapped share is accounted for right there. Its production impact is kept (it was still manufactured) but reported under the Production loss stage, so the reader can see how much of the footprint was spent on material that never made it into the product. The scrap mass additionally feeds the Production waste module (Section 6.5), where its treatment is modelled.
• The surviving share is what carries on. All quantities deeper in the structure are based on this smaller amount, because only surviving units actually contain those deeper parts and materials.

Example. An enclosure is machined from a casting, and the structure declares 1 casting going into the machining step, with a 10 % rejection rate. Each casting contains 2 kg of aluminium:

• The full casting (1 unit) was manufactured, so its full production impact is counted.
• 10 % of it — 0.1 casting's worth of impact — is reported under the Production loss stage.
• 0.9 casting survives into the product, so the aluminium counted inside the finished product (and later available at end of life) is 0.9 × 2 kg = 1.8 kg, not 2 kg.

This refines the effective quantity from Section 5.1: the quantity of any part that ends up inside the finished product is found by walking down from the product to that part, multiplying at each step by the declared quantity and by the surviving fraction of that step:

Quantity in finished product = (qty₁ × survival₁) × (qty₂ × survival₂) × …   for each step on the path

where survival = 1 − loss rate of that step

The reason for this care is easy to state: scrapped units never reach later life cycle stages. A rejected casting is never transported to the customer, never used, and never disposed of by the end consumer — so its burden must stay in the production stage and must not inflate the quantities of downstream stages. Splitting at the point of loss achieves exactly that.

5.3 Weight roll-up

Several modules (transport, end of life) depend on mass. The platform derives the shipped weight of an assembly from the bottom up:

Weight(assembly) = Σ over children c of:  Weight(c) × quantity(c) × (1 − loss rate of the line)

with leaf-part weights taken from the declared part weight (in kg). Scrap lost in production does not travel with the finished product, which is why the output factor (1 − L) is applied.

5.4 Assembly variants

Some assemblies are sold or stocked in multiple configurations — for example, the same cupboard with legs, with wheels, or as a wall-mounted unit. Dawn models these as variants of one assembly rather than as separate assemblies.

What varies. Only the bill of materials: which parts and child assemblies appear, in what quantities, and with what production-loss rates. Each variant has its own SKU and may carry its own suggested price.

What is shared. All assembly-level life cycle data applies identically to every variant: production utilities, packaging, transport to customer, use-phase energy and spare parts, disassembly, end-of-life pathway allocations, and declared production volume. Changing these on the assembly affects every variant's impact report equally.

Base configuration. The default configuration does not require a separate variant record. It is the core BOM — the part and child-assembly links stored directly on the assembly — and uses the assembly's own SKU, price, and version counter. An assembly with no variant records behaves exactly as it did before variants existed.

BOM resolution. The effective BOM used in every calculation for a selected variant is built by merging the core BOM with that variant's deltas:

Resolved BOM = merge(Core BOM, Variant deltas)

The merge follows a fixed, deterministic order (the same algorithm runs on the server and in the client impact view):

1. Start from all core part lines and child-assembly links.
2. Exclude any core line listed in the variant's exclusion set.
3. Override quantity and production-loss percent on core lines where the variant declares an override keyed to that core line.
4. Add variant-only part lines and child-assembly links that do not exist in the core BOM.
5. Recurse into nested child assemblies using the childVariantId on the link when set; otherwise the child is resolved at its base configuration.

Each resolved line is tagged as inherited from the core (core), added only in this variant (variant), or a core line with changed quantity or loss rate (override). These tags are for traceability in the user interface; they do not change the arithmetic.

Impact scope. All BOM-dependent calculations use the resolved BOM for the active variant:

• quantity propagation and production-loss roll-up (Sections 5.1–5.2),
• shipped weight (Section 5.3),
• cradle-to-gate and part-level module scaling (Section 6.1–6.2),
• production waste from scrap (Section 6.5),
• material inventory for end-of-life treatment (Section 6.7).

Assembly-level modules attached to the product as a whole — utilities, packaging, distribution, use phase, disassembly, and end-of-life transport — are not re-derived per variant; they are shared inputs multiplied into each variant's total through the same assembly-level formulas in Section 6.

Nested assemblies. A variant may specify which configuration of a child sub-assembly to use (for example, resolving a child at "with legs" inside a parent variant). When no child variant is specified, the child's base BOM is used. Quantity multiplication and loss handling through nested structures follow the same rules as Section 5.1, applied to the resolved links at each level.

6. Life Cycle Modules

Impacts are organised into life cycle modules. Organisations name and order their own life cycle stages (e.g. "Raw materials", "Manufacturing", "Distribution", "Use", "End of life") and map each module to a stage, which keeps reporting consistent with frameworks such as EN 15804 (modules A1–A3, A4, B, C, D). The calculations behind each module are as follows.

6.1 Materials and purchased components (cradle-to-gate, A1–A3 analogue)

Each part obtains its cradle-to-gate impact in exactly one of two ways (the platform enforces that they are mutually exclusive):

1. A matched background dataset. The part is linked to a database activity (Section 3) and an amount in that dataset's functional unit (typically the part's mass). Impact = amount × emission factors, per Section 4.
2. A linked component LCA. The part references a full, published LCA model built in the platform — for example a detailed model of an in-house manufacturing process, or a supplier's component modelled from primary data. The published result of that model is used as the part's per-unit impact.

For parts described by a material composition (e.g. "60 % aluminium, 40 % ABS by mass"), the mass assigned to each material is:

Material mass = part weight × material share (%)

and each material is matched to its own dataset.

6.2 Inbound transport (supplier → factory)

Transport impact is based on transport work — mass moved times distance — which is how freight datasets are expressed (per tonne-kilometre).

For each supplier of a part, a transport route is defined as one or more segments, each with a distance and a transport mode (e.g. ocean container ship, heavy road truck). For each segment:

Transport work  = distance (km) × part weight (kg)        [converted to t·km]
Segment impact  = transport work × EF of the transport mode dataset

When a part is sourced from multiple suppliers, each supplier's transport chain is weighted by a sourcing allocation:

Part transport impact = Σ over suppliers s of:  (allocation(s) / 100) × Σ over segments of s

If no allocations are declared, sourcing is split equally across the suppliers (100 % / N each). If allocations are declared, they are used exactly as entered (suppliers without a declared share contribute 0 %).

The result is scaled by the part's effective BOM quantity, like every other contribution. A part with no declared weight contributes no transport impact (and this is visible to the user as a data gap).

6.3 Production energy and utilities

Energy and utility consumption in manufacturing (electricity, natural gas, compressed air, water, …) is recorded per part or per assembly as a consumed amount:

Utility impact = consumed amount × EF of the utility dataset

For example, 0.8 kWh of electricity per unit, multiplied by the emission factors of the selected national grid-mix dataset.

6.4 Packaging

Packaging materials are modelled as parts with their own datasets and quantities, attached to the product under a dedicated Packaging module. Their disposal is modelled separately under Packaging waste using the end-of-life mechanism in Section 6.7.

6.5 Production waste

Scrap mass generated by production losses (Section 5.2) is traced back to its material composition, and each material's lost mass is routed to one or more treatment pathways (recycling, incineration, landfill, …):

Waste treatment impact = lost material mass (kg) × pathway share (%) × EF of the treatment dataset

The transport of production waste to the treatment site is also modelled, using the lost mass and the same transport-work formula as Section 6.2.

Example. Continuing the casting example from Section 5.2: 0.1 casting is scrapped during machining, and each casting contains 2 kg of aluminium, so 0.2 kg of aluminium scrap is generated per product. The organisation declares that 80 % of aluminium scrap is recycled and 20 % is landfilled, and the treatment datasets carry climate change factors of 0.04 kg CO₂-eq/kg for scrap recycling and 0.01 kg CO₂-eq/kg for landfill:

Recycling:  0.2 kg × 80 % × 0.04 kg CO₂-eq/kg = 0.0064 kg CO₂-eq
Landfill:   0.2 kg × 20 % × 0.01 kg CO₂-eq/kg = 0.0004 kg CO₂-eq
Production waste treatment total              = 0.0068 kg CO₂-eq

If the scrap is transported 50 km by truck to the treatment site, the additional transport impact is calculated from the transport work: 0.2 kg × 50 km = 0.01 t·km, multiplied by the truck dataset's emission factors. As always, the same arithmetic runs across the full indicator set, not only climate change.

6.6 Distribution and use phase

Transport to customer uses the transport-work formula with the full assembly weight:

Distribution impact = distance (km) × assembly weight (kg) × EF per t·km

Use phase. Products that consume energy or materials during use are modelled against a declared lifetime reference (e.g. operating hours, wash cycles, years of service):

Lifetime consumption = consumption per lifetime unit × number of lifetime units
Use phase impact     = lifetime consumption × EF of the utility dataset

Spare parts consumed over the product's life are added as additional part quantities under the use phase, and installation and maintenance activities can be modelled as separate modules using the same quantity × emission factor principle.

6.7 End of life

End-of-life modelling starts from the product's material inventory: the platform walks the entire BOM down to leaf parts and computes the mass of each material that leaves with the product:

Material mass at EOL = part weight × output quantity × material share (%)

summed over every occurrence of the material in the product.

Each material is assigned end-of-life pathways with percentage allocations (e.g. 70 % recycling, 20 % incineration, 10 % landfill). Each pathway is matched to a treatment dataset:

EOL treatment impact = material mass × pathway share (%) × EF of the treatment dataset

Transport from the point of use to the waste treatment site (EOL transport to waste) is modelled with the transport-work formula, and disassembly effort can be modelled as its own module.

End-of-life benefits (credits). Where treatment produces secondary material or recovered energy that substitutes primary production, a benefit can be modelled (analogous to EN 15804 module D). A benefit is always tied to a source pathway and enters the total as a negative contribution:

EOL benefit = − material mass × source pathway share (%) × benefit share (%) × EF of the substituted activity

For example: 100 kg of aluminium, 70 % recycled, with 90 % of the recycled output substituting primary aluminium, yields a credit of −(100 × 0.70 × 0.90) × EF(primary aluminium) — reported separately so that gross impacts and credits remain transparent.

7. Dataset Selection and AI Assistance

Matching a part to the right background dataset is the most judgement-intensive step in any LCA. Dawn supports both manual selection (searching the licensed databases directly) and an AI-assisted matching process. The AI assistance automates the search, but the calculation itself is always the deterministic arithmetic described above — the AI never invents emission factors.

The AI-assisted matching follows these steps:

1. Classification. The part is classified by input type (raw material, purchased component, transport, energy, waste treatment, …) based on its name, description, supplier, and position in the BOM.
2. Search. Several search queries are generated from the part's context (materials, geography, supplier information) and run against a semantic search index of all datasets in the organisation's licensed databases.
3. Candidate validation. The candidate datasets are evaluated for representativeness — technological fit, geographic fit, and unit suitability — and the best match is selected.
4. Quantification. The amount in the dataset's functional unit is derived from the part's declared data (typically its weight, with unit conversion applied, e.g. kg → tonnes).
5. Recording. The selected dataset, the amount, and the resulting impact report are stored on the part, and a version snapshot is taken (Section 9).

Every AI-assigned match remains visible and editable: the dataset name, source database, system model, and amount are shown on the part, and users can replace the match manually at any time. Material-type parts are deliberately excluded from direct dataset assignment — their impacts flow exclusively through the material composition and end-of-life mechanisms, preventing double counting.

8. Methodological Choices, Assumptions, and Limitations

Verifiers should be aware of the following platform-level choices:

• Deterministic point values. All results are point estimates. The platform does not currently propagate uncertainty ranges (e.g. Monte Carlo simulation or pedigree-matrix uncertainty factors). Data quality indicators attached to ecoinvent exchanges (reliability, completeness, temporal and geographical correlation) are preserved and exportable for review, but they do not modify the numbers.
• System model consistency. Results computed under the cut-off system model follow cut-off conventions throughout (recycled content enters burden-free; end-of-life credits are modelled explicitly and reported separately as benefits).
• Allocation is user-declared. Supplier sourcing splits, material composition shares, and end-of-life pathway shares are declared by the user (or proposed by the AI and confirmable by the user). The platform applies them arithmetically as entered; only the equal-split default for undeclared supplier allocations is applied automatically.
• Mass is the backbone. Transport, material, and end-of-life calculations depend on declared part weights. Missing weights result in zero contributions for the dependent modules rather than estimates, making data gaps visible rather than silently filled.
• No double counting by construction. A part's cradle-to-gate impact comes from exactly one source (dataset or linked LCA, never both); single-score aggregation excludes sub-categories and inventory indicators; production scrap is separated from product output; and dataset amounts are applied exactly once when a part's impact report is built.
• Database versions. Calculations reference a specific database version (e.g. ecoinvent 3.12, EF 3.1). When databases are updated, existing results remain tied to the version they were calculated with until recalculated.

9. Traceability and Versioning

To support verification, the platform maintains a full audit trail:

• Part-level impact reports store, alongside the results, the source dataset's identity, database, system model, declared amount, and life cycle stage.
• Version snapshots are taken whenever a dataset is assigned or changed on a part, preserving the previous state.
• Published LCA reports are immutable records of a product's results at publication time, including the complete indicator set, the per-stage breakdown, and the model structure that produced them.
• Contribution tracking records which part, module, and dataset each portion of the total comes from, enabling drill-down from a product-level figure to the individual quantity × emission factor calculations beneath it.
• Assembly version snapshots preserve the resolved BOM and calculated impact for each assembly configuration. The base configuration and each variant maintain separate version streams — a historical report is always identified by assembly, variant slot, and version number. Editing the core BOM creates a new main version for the base slot and auto-update snapshots for every variant slot that reflect the new merged BOM without advancing each variant's main version. Editing a variant's deltas alone advances only that variant's version. Shared assembly-level changes (utilities, packaging, transport, end-of-life allocations, and similar) write auto-update snapshots across the base slot and all variant slots at each slot's current main version. Snapshots store the resolved part list, child links (including any specified child variant), and the full impact breakdown at capture time.

10. Summary of Calculation Formulas

Contribution	Formula
Any single activity	Impact(i) = quantity × EF(i) per indicator i
Carbon footprint	GWP100 total, in kg CO₂-eq (unweighted)
EF single score	Σ Impact(i) / NF(i) × WF(i) over the 16 EF 3.1 categories, in points (displayed as mPt = Pt × 1000)
BOM roll-up	Σ part impact per unit × effective quantity (quantities multiply down the tree)
Assembly variant BOM	Resolved BOM = merge(Core BOM, exclusions, overrides, additions) — all BOM roll-ups use the resolved structure for the active variant
Production loss	input = planned qty; scrap = qty × L; output = qty × (1 − L) — scrap burden reported separately
Inbound transport	distance × weight → t·km × EF, × supplier allocation %
Production energy	consumed amount × EF of utility dataset
Production waste	lost material mass × pathway share % × EF of treatment
Transport to customer	distance × assembly weight → t·km × EF
Use phase	consumption per lifetime unit × lifetime units × EF
End-of-life treatment	material mass × pathway share % × EF of treatment
End-of-life benefit	− material mass × source pathway share % × benefit share % × EF of substituted activity

---

This document describes the calculation methodology implemented in the Dawn platform as of June 2026. Emission factor values themselves originate from the referenced databases (ecoinvent, EF 3.1, EN 15804 EPDs) and are subject to those databases' own documentation and review processes.