Building LCA framework for the UK, the EU and Sweden

01The short version

• A building LCA is an LCA of LCAs. You take a bill of quantities with thousands of lines, link each line to a ready-made dataset (EPD or generic value) with LCIA results per module, multiply and sum. The hard parts are quantities, classification, data choices, scenarios and rules.
• Same foundation, different rules. Sweden and the UK both build on EN 15978 and EN 15804. The differences sit in the method profile: modules, building parts, GWP variant, area, generic data and uplifts.
• Sweden today: klimatdeklaration (climate declaration) for A1-A5, klimatskärm (building envelope) + load-bearing parts + internal walls, uptake is not counted, absolute kg CO2e per building part to Boverket's register. Industrial buildings and workshops are exempt.
• Sweden 2028/2030 (proposal): the climate declaration is repealed at the end of March 2027 and replaced by a requirement in PBL (plan- och bygglagen, the Planning and Building Act). EU report in GWP-total from 2028, Swedish limit values in GWP-GHG per m² BTA from 2030, with a separate category for logistics halls.
• UK: no legislation. RICS WLCA v2 (Version 3, Aug 2024) requires A-C with D separate, 60 years, its own building-element categories and uplifts. UKNZCBS V1 gives voluntary limit values per sector and start year.
• EU: (EU) 2026/52 in force since May 2026: GWP-total, 50 years, total per m² useful floor area, seven reporting columns.
• For the data model: store raw data at line level with provenance, calculate per method profile, save reports as immutable snapshots with a scope fingerprint. Store area and volume, and one row per use with its area (mixed-use buildings are weighted by area).
• Validated against real data from a first client: 306 OneClick exports, a verifier's report template and a complete BIM project. The structure held and the line totals reconcile in every file (section 13). The client's own method sits in a client layer on top of the market model.

02What is different about building LCA

Functional unit: a building over a study period

The object is the whole building and the result is normalised to kg CO2e per m². Two things must always be stated: the normalisation unit (Sweden m² BTA, EU m² useful floor area, UK m² GIA and, for industrial and storage buildings, m³ internal volume under RICS) and the reference study period (none for today's Swedish climate declaration, 50 years for the EU and for Sweden from 2028, 60 years for RICS). EN 15978 calls the whole the functional equivalent; results may only be compared between buildings with the same one.

The inventory is a bill of quantities

1. Source documentsDrawings, BIM, cost estimate, delivery notes
2. Bill of quantities"External wall YV-01, 2,400 m²"
3. Building partClassified: RICS 2.5.1, CoClass B10.AD, building envelope
4. Material line150 mm mineral wool → 10,800 kg
5. DatasetEPD or generic, value per module
6. ScenariosA4 transport, A5 waste, B4 replacement, C
7. UpliftsCoverage, contingency, top-up
8. Reportkg CO2e/m² per module and building part

Declared unit and conversion

An EPD states results per declared unit (1 kg, 1 m³, 1 m², 1 piece). The bill of quantities states the quantity in the designer's unit. Conversion factors are needed between them: density, mass per area, thickness, mass per piece. Unit errors are the most common error, so the factor is always stored with its source.

Scenarios and design stages

A1-A3 come from the dataset. Everything else is scenarios from the EPD, national default values or the project. An LCA is carried out several times as the project matures, and early stages are compensated for with uplifts.

The Swedish climate declaration is made as-built, before the slutbesked (completion certificate). RICS requires an early baseline plus as-built.

03The standards pyramid

From the foundation up to the client. Each level builds on the one below.

EN 15978:2026 is published (CEN Nov 2025, BS EN July 2026, SS-EN available) and replaces the 2011 edition after roughly 24 months of coexistence. New: module A0, voluntary B8, D split into D1/D2, indicators aligned with EN 15804+A2. RICS v2 and the climate declaration are still based on the 2011 edition.

04Life-cycle modules A0-D

The modules are the most important dimension in the data model. Choose a method to see what it counts.

The EU report has seven columns: A1-A3 | A4-A5 | B1-B4 | B6 | C1-C4 | D1 | D2. A0 does not appear in the EU text. RICS and the EU split B1 into B1.1 (materials, carbonation) and B1.2 (refrigerants).

A5 is split differently, and the code A5.1 means two things

05Product level: EPDs and EN 15804

An EPD is a third-party verified type III declaration (ISO 14025) under EN 15804, with LCIA results per module and indicator for a declared unit. It is published by a programme operator (International EPD System, BRE, EPD Hub, IBU, EPD Norge) and is valid for 5 years.

+A1 versus +A2: why the GWP figures cannot be compared directly

The GWP variants

RICS accepts +A1 (sequestration is then calculated separately). Boverket's climate database is calculated under EN 15804:A1 and supplies "GWP-GHG AR4". Where an EPD lacks GWP-GHG, Boverket proposes: GWP-GHG(A1-A5) = GWP-total(A1-A5) + 44/12 × biogenic carbon in product and packaging.

Timber often gets a negative GWP-total in A1-A3 that is only balanced out in C. That is why Sweden uses GWP-GHG and RICS reports sequestration separately. The model flags derived values.

Zero is not the same as missing

An EPD can state a module as a value, ND/MND (not declared) or MNR (not relevant). The model distinguishes between them; otherwise a missing C module looks like zero emissions.

Data sources and formats

06Sweden

Climate declaration, 2022 up to and including March 2027

Byggtjänst's interpretation with CoClass codes

What the register accepts

XSD 2026-06-02: all fields are integers except four text fields, no enumerations.

What is coming: Boverket report 2026:16 (remiss, consultation)

The limit values are around 35% below the 2020 median. In the hall data, Boverket added +40 kg/m² where finishes and services were missing (around 20 where they were reported), and measured A5 energy was 10-60% of the default value. The roadmap (draft 2026:29) covers only A1-A5; the 2040 milestone is based on the EU target of a 90% reduction compared with 1990, linear 2030-2040. No figures are given for 2035-2050.

Boverket's climate database 02.07.000

07UK

There is no legal requirement on embodied carbon web. Requirements come from the RICS standard, planning requirements, voluntary standards and clients.

RICS WLCA 2nd edition, Version 3 (Aug 2024)

Uplifts

Contingency applies to all modules except B6-B8. Data quality is scored for the 10 most important products: geography 0-10, technology 0-6, product specificity 0-6, time 0-8, granularity 0-5, verification 0-5 (max 40). The weighted mean gives the uplift: ≤ 10 points 7%, 35-40 points 1%, 40 points 0%. Quantity uncertainty: 0 measured, 1 designed, 3 estimated, 5 benchmark, which gives 0-4%. The confidence score must be reported.

Default values

UKNZCBS V1 Rev 1 (Apr 2026): Storage & Distribution

Industry standard with mandatory third-party verification, only for completed and occupied buildings. The upfront limit covers A1-A3, A4, A5.2 and A5.3 without sequestration, with the RICS uncertainty uplift. In: RICS 0.1.2, 1, 2, 3, 4.1-4.3, 5 except solar PV, first fit-out. Out: 0.1.1, A5.1, external works (8), solar PV (own limit), loose furniture. The limit year is the date of commencement. Internal volume is a mandatory field for storage. Offices within the warehouse are classed as Offices.

kg CO2e/m² GIA. *Shell & core limits are used only in Annex E and F2; the main claim is tested against the whole building including first fit-out. Solar PV: 1,000 kg CO2e/kWp in the standard, 750 in the submission template. Operational energy, new build 2026: unheated 39, heated 88, cold store 169 kWh/m²/yr.

BREEAM UK New Construction V7, Mat 01 web

BREEAM certification is more common than stand-alone LCA in the UK market, and owners often require "Excellent" on every building, so Mat 01 drives LCA volume.

In the model the three stages are linked assessments of one building, the BREEAM profile carries the required stages and the tool rule, and the tool used is recorded so that compliance can be checked.

Other systems and reference values

LETI has bands per type (office, residential, schools, retail; band A around 300-350 kg/m² GIA), RIBA 2030 has whole life-cycle targets (office 750, residential 625), the GLA has benchmarks, SCORS covers only the structure, and the draft London Plan 2026 proposes 500/700 for residential under/over 11 m and 700 for Use Class E web. None of them has a storage category.

08The EU layer

Delegated Regulation (EU) 2026/52

EPBD Art. 7 (2028/2030, roadmaps by 1 Jan 2027), CPR 2024/3110, EN 15978:2026 and Level(s) have not been verified against the primary text web.

Certification on the continent. In Germany the main scheme is DGNB (System 2023, criterion ENV1.1 for new buildings), which refers in many places to the calculation rules of QNG (Qualitätssiegel Nachhaltiges Gebäude). Assessments made for BREEAM, DGNB or a client method are often the same model adapted, so the model keeps one inventory and reports it per profile. DGNB parameters are not yet modelled web.

09Comparison, parameter by parameter

Five differences that affect the figure most

1. Building-part scope: services, finishes and external works in or out. Boverket added +40 kg/m² for missing finishes and services in halls.
2. Normalisation: BTA (outside) versus GIA (inside) versus m³ internal volume (RICS for storage).
3. Generic data: +25% on Boverket's product data; RICS has uncertainty uplifts instead.
4. A5: measured energy in Sweden versus 40 kg/m² GIA in RICS.
5. GWP variant: GWP-GHG versus GWP-total, a large difference for timber.

Limit values by building type

All values are held in benchmarks.json with scope, status and source, and are linked to the neutral building types in building-uses.json. The values cannot be compared directly: UKNZCBS is upfront per m² GIA, SE is A1-A5 GWP-GHG per m² BTA, FR is dynamic LCA over 50 years, NL is a shadow cost in €/m² BVO per year.

The Nordic countries calculate per year over 50 years: Denmark (from 1 July 2025, incl. B6) 6.7 small houses, 7.5 multi-dwelling buildings, offices, retail and storage, 8.0 other, plus 1.5 for A4-A5; Finland (from 9 Jan 2026) 16 to 29 kg/m² heated net area per year depending on type web. UKNZCBS and SE are read from the primary documents; FR, NL, LETI, DK and FI are based on secondary sources.

10Building parts in scope

There is no official crosswalk between the Swedish and British building-part systems, so we build our own with a neutral code as the pivot. The UK codes are RICS v2's building-element categories. The CoClass codes come from Svensk Byggtjänst's mapping for the climate declaration. The table is read from building-part-crosswalk.json v0.4 (the market layer). A client's own report rows are added as an extra column in the client layer.

BREEAM UK and GLA exports use the older RICS code tree, where the same number means different things (8.2 is roads in the old tree but planting in v2). Which code system applies is therefore determined by the tool, according to source-labels.json.

11The data model

Design principles

1. Raw data at line level, always. Quantity, unit, raw classification, raw material name, dataset, scenario. It is the only thing that can be recalculated.
2. Method profile as configuration. Each method is a record: modules, indicator, study period, area, building-part scope, uplifts, data hierarchy.
3. Classifications are many-to-many, with a neutral building-part code as the pivot.
4. Every uplift is its own record, never baked into a factor.
5. Indicator values in a long table: dataset × module × scenario × indicator → value + status.
6. Every result carries a scope fingerprint.
7. Reports are immutable snapshots with a hash.
8. Version everything that changes: databases, methods, default values, mappings.
9. Provenance everywhere: where the quantity, the factor and the scenario came from.

Entity diagram

erDiagram
    ORGANISATION ||--o{ PROJECT : "owns / assesses"
    PROJECT ||--o{ BUILDING : contains
    BUILDING ||--o{ AREA_MEASUREMENT : areas
    BUILDING ||--o{ ASSESSMENT : "assessed in"
    METHOD_PROFILE ||--o{ ASSESSMENT : governs
    ASSESSMENT ||--o{ ELEMENT : BoQ
    ELEMENT ||--o{ MATERIAL_LINE : "consists of"
    MATERIAL_LINE }o--|| DATASET : uses
    MATERIAL_LINE }o--o{ SCENARIO : "A4 A5 B4 C"
    DATASET ||--o{ INDICATOR_VALUE : "module x indicator"
    ELEMENT }o--o{ CLASSIFICATION_CODE : classified
    CLASSIFICATION_CODE ||--o{ CLASSIFICATION_MAPPING : mapped
    ASSESSMENT ||--o{ ADJUSTMENT : uplifts
    ASSESSMENT ||--o{ RESULT : calculated
    ASSESSMENT ||--o{ REPORT_SNAPSHOT : reported
    REPORT_SNAPSHOT }o--o{ BENCHMARK_VALUE : compared
    ASSESSMENT ||--o{ QUALITY_FLAG : flags

Entities

Quality flags (starter list)

Verification, portfolio metrics and restatement

• Verification is a loop. The assessor sends a results template, the BIM model, drawings and a cost plan; the verifier returns queries per part or material; the assessor answers and updates until the verifier signs off or concludes that the assessment cannot be verified. The model stores the deliverables, each round of queries and responses, the outcome and the summary report. Continental verifications are often done without drawings, which is why the deliverables list matters.
• Quantities need a source order. Each part keeps which source gave the quantity (BIM, drawings, specification, cost plan, delivery notes, procurement) and the evidence behind it. Conflicts and double counting across models are flagged, not silently resolved.
• Specifications are checked against datasets. Strength class, cement type, SCM share, reinforcement type and amount are stored per line, so a steel-fibre slab modelled with rebar, or a CEM III mix modelled with a CEM I EPD, shows up.
• Completeness per part. Comparing each part with buildings of the same typology shows where an assessment is likely incomplete.
• Portfolio metrics are aggregates. An owner's annual target is typically total kg CO2e of all included buildings divided by their total area. Buildings with non-standard elements, such as a cold store, can be excluded or have those elements split out; the decision is stored per metric and year.
• Restatement. When emission factors or methods change, earlier years are recalculated as new snapshots while the original figures stay reproducible.

Calculation flow

for each material line:
  1. roll up classification → neutral building part → in the profile's scope?
  2. convert quantity to the dataset's declared unit (stored conversion)
  3. choose indicator per profile (GWP_GHG or GWP_total), flag if missing
  4. A1-A3 = quantity × value   [SE: conservative value if generic]
  5. A4: SE climate database transport legs; RICS table 17 with 43% empty running
  6. A5 waste: SE (waste factor − 1) × (A1-A3 + A4); RICS rate × (A1-A3 + A4 + C)
  7. B4 = (A1-A3 + A4 + A5 + C) × number of replacements(RSP, ESL)   [if the profile requires B]
  8. C1-C4, D1, D2 from dataset or scenario
at assessment level:
  9. A5 energy measured (SE) or 40 kg/m² GIA (RICS); demolition 35 kg/m² demolished GIA
 10. uplifts in the profile's order: coverage, contingency, uncertainty, top-up
 11. aggregate per module group and building part
 12. normalise (m² BTA / m² GIA / m³ / m² useful), both scenarios → ReportSnapshot

Re-reporting means running the steps again with a different profile. Hard incompatibilities are flagged instead of being silently converted: +A1 data in a +A2 profile, EPDs without C and D, GWP-GHG where GWP-total is required, imported totals that cannot be broken down, different area standards.

The current project-lca.schema.json is kept as a flat report view for the first client, so the app is not affected.

12Populating on an ongoing basis

• Always import the raw line first, then normalise. Store the source file and line number.
• A new method version becomes a new profile; old reports must remain reproducible.
• A new database version becomes new datasets; the old ones are kept.
• Mappings have a confidence and an author, so manual decisions can be reused.
• Flags are created on import and follow the line until they are resolved.

13Validation and client layer

The model was tested against real data from a first client, a European logistics portfolio: 306 OneClick exports with content, the verifier's report template and method, and a complete BIM project (12 IFC models, quantity take-offs, three OneClick designs). Client data is kept outside the repo. Method and lessons: docs/LCA_VALIDATION.md.

What held

• The chain project → building → assessment → building part → material line → dataset/result → report carried both the portfolio and a BIM project with 37,675 building parts.
• The line totals match the tool's own module totals in 306 of 306 files. Recalculation per method profile works at line level.
• BIM quantities match the LCA tool for the large items (ground slab and steel within 1%).

What broke and has been fixed

The portfolio by method (kg CO2e/m², median, range for unclear building parts)

The client layer

Property owners and verifiers often have their own methods: a selection of building parts, their own area (often lettable area), their own uplifts and an Excel template. These sit in a client layer on top of the market model, never in it. An assessment has one calculation method and a list of evaluated profiles that can mix market profiles (UKNZCBS, SE 2030, EU) and client profiles. The first client's entire method could be expressed in this way, with no changes to the core schema.

14Regulatory radar

What applies and what is coming in the UK, Sweden and the EU, ordered by your market lens, and whether the model can handle it. Future regulations are held as profiles marked horizon and must be verified against the primary text before they are used for compliance. Details: docs/LCA_REGULATORY_RADAR.md.

Why the model can handle new rules

• Every regulation is a versioned profile, and limit values are data (category × year × value × status).
• All areas are stored in parallel (BTA, ljus BTA (above-ground BTA), Atemp, GIA, NLA, useful floor area, SREF, BVO, NRF, volume) and the profile chooses the denominator.
• Both GWP variants plus MKI, RE2020 Ic and CML; modules according to EN 15978:2026.
• Trigger dates determine the regime; product data from CPR, product passports and CBAM is kept separate from EPDs; licence status per dataset.
• Corporate reporting and change of ownership (Scope 3 category, year, ownership share).

15Open questions

16Glossary and sources