Translate | ISO 22057:2022 EPD Data Templates, BIM Properties, Life-Cycle Modules and Environmental Claims

How do you translate an Environmental Product Declaration (EPD) for BIM, construction-product life-cycle assessment or sustainable building procurement without changing the figures or their meaning? International architects, manufacturers, engineers and sustainability teams need to localize product descriptions, environmental indicators, life-cycle module labels and data-template fields. A seemingly simple mistake—translating ‘per cubic metre’ as ‘per tonne’, treating an undeclared value as zero, or changing a property ID—can invalidate both the human conclusion and software exchange.

ISO 22057:2022 addresses data templates for using environmental product declarations for construction products in building information modelling. According to the official ISO catalogue, the published 2022 standard provides principles and requirements for structuring EPD-related environmental and technical data, using data-template concepts associated with ISO 23386 and ISO 23387. The catalogue indicates it is marked for future revision, but the 2022 edition remains the published reference as of 5 November 2026. The public scope explicitly excludes assessing the environmental performance of a whole construction works.

This focused guide explains how to translate EPD data templates, BIM property names, global warming potential (GWP) indicators, declared units, life-cycle modules, version records and environmental claims. The aim is to preserve evidence and interoperability while helping real readers understand the numbers. You will find clear explanations, a terminology crosswalk, twenty-four original worked translation cases, bilingual release checks, FAQs and a four-week learning plan. These examples are teaching scenarios, not copied standard clauses or an assertion that any product has a particular certification.

Quick answer

• Preserve machine-readable property IDs and environmental numeric data exactly.
• Translate human display labels and explanatory notes through controlled concept mapping.
• Keep declared unit, product scope, life-cycle modules and missing-data status explicit.
• Separate product-specific EPDs from generic life-cycle datasets or marketing claims.
• Do not infer building-level impact or product superiority from a single EPD cell.
• Validate both software interoperability and the reader’s environmental interpretation.

Contents

1. Understand why the EPD data must remain machine-interpretable
2. Keep the declaration’s product identity intact
3. Separate declared unit, functional unit and physical quantity
4. Preserve lifecycle modules and boundaries
5. Distinguish GWP indicators and numerical evidence
6. Do not invent values for unreported modules
7. Separate verified EPD evidence from marketing language
8. Explain data-template keys and human-facing labels
9. Maintain EPD validity and version dates
10. Connect BIM objects to the correct EPD record
11. Differentiate generic LCA data from product declarations
12. Avoid unsound environmental comparisons
13. Handle circularity and end-of-life scenario descriptions
14. Check the receiving BIM and procurement workflow
15. Glossary
16. Worked cases
17. Bilingual QA
18. FAQ
19. Four-week plan

1. Understand why the EPD data must remain machine-interpretable

A designer may receive an Environmental Product Declaration as a PDF, a table, a machine-readable dataset or an object attached to a BIM product type. Translating a PDF caption is different from localizing a structured property name. The latter may break a software link if an identifier is changed. The central question is not merely whether the words are fluent; it is whether the receiving software and human reader still refer to exactly the same environmental quantity.

The official scope of ISO 22057:2022 concerns structuring environmental and technical data from EPDs for use in building information modelling. It references data-template principles associated with ISO 23386 and ISO 23387 and includes generic life-cycle data where suitable EPD information is absent. It does not itself constitute a whole-building environmental assessment. Translate within that narrow data-interoperability purpose.

Practical decision check: Can both the human reviewer and software system still identify the same property, product and value? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

2. Keep the declaration’s product identity intact

An EPD may apply to one manufacturer, product family, plant, geography, formulation or declared scope. A translated product label that becomes too general may lead a reader to apply data to a material not actually covered. A multilingual procurement team needs exact product description and applicability rather than an approximate marketing name.

Preserve EPD registration number if present, product identifier, publisher, manufacturer, issue date, declared scope, version and verification status as shown in the source. Translate descriptions for readability, but retain official names and codes in linked fields. Where one family has several variants, do not merge values across them.

Practical decision check: Does the translation let the buyer distinguish the declared product from a similar but unassessed alternative? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

3. Separate declared unit, functional unit and physical quantity

Numbers in a construction EPD may be reported per kilogram, cubic metre, square metre, item or functional service. A result per one tonne is not interchangeable with a result per one cubic metre unless a documented conversion and assumptions support it. Translating ‘per declared unit’ as ‘for the whole building’ is a profound meaning error.

Copy the exact reference unit, amount, density assumptions if provided, functional description and scenario limitations. Verify that unit symbols survive localization and that charts do not silently relabel the denominator. If the target locale commonly uses other units, display approved equivalents alongside original values with transparent rounding.

Practical decision check: Can the reader state what exact quantity the reported environmental result belongs to? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

4. Preserve lifecycle modules and boundaries

Building-product EPDs often organize information by life-cycle stages or modules. A number for product manufacturing cannot be casually treated as a whole-life impact. Construction transport, installation, use, end-of-life and benefits or loads beyond the assessed system can require separate scenarios and accounting conventions.

Keep each published module identifier, source scope and declared or nondeclared status unchanged. Translate plain-language descriptions without renumbering machine keys. Never sum modules just because they appear in adjacent columns; scope, indicator and compatibility conditions determine valid interpretation.

Practical decision check: Does a translated headline tell which modules are included and which are missing? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

5. Distinguish GWP indicators and numerical evidence

A table may list global warming potential values and multiple environmental indicators with their own units and data characteristics. A translator must not change the sign of a value, combine different indicators or turn a numerical result into a claim of environmental superiority. A low number is meaningful only relative to the indicator, declared unit and compared scope.

Preserve indicator abbreviations when they are controlled, while explaining their meaning in a separate locale field. Check negative numbers, exponent notation, CO₂e symbols, order of magnitude and decimal separators. Have a qualified environmental-data reviewer validate any calculations or comparisons not explicitly present in the source.

Practical decision check: Can the reader trace a climate-impact figure to its exact indicator, unit and module? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

6. Do not invent values for unreported modules

A blank cell, ‘not declared’, not applicable, not assessed and numerical zero are different conditions. Filling blank EPD cells with zero may artificially reduce total impact. A missing value may prevent a particular comparison, while a verified zero would be a specific quantitative statement.

Create separate labels for missing, inapplicable, below-reporting-threshold, calculated zero and unavailable data according to the actual source. Keep machine-coded missing values stable so analytics do not treat them as numeric observations. Flag ambiguous cases for the declaration owner.

Practical decision check: Does an absent value remain absent rather than becoming a zero or an implied positive conclusion? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

7. Separate verified EPD evidence from marketing language

An EPD communicates quantified information under an applicable program and methodology, but its existence does not itself certify that a product is the ‘greenest’, ‘carbon neutral’ or best choice for a project. Comparative claims require appropriate conditions, common scope and documented analysis.

Keep the EPD’s reported data distinct from brand slogans and buyer interpretations. Translate verification statements using their actual scope, version and role; do not invent independent certification of performance beyond what the declaration says. If a vendor adds a sustainability claim, label it as the vendor’s claim and review separately.

Practical decision check: Is the sentence an EPD datum, an independent verification statement or a sales interpretation? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

8. Explain data-template keys and human-facing labels

A property template may include stable identifiers, property names, definitions, data types, units and allowable values. Software may expect a key or dictionary reference to remain exact. A translator who localizes a key rather than its display label can break the exchange, even when the resulting prose looks correct.

Build an explicit two-layer model: machine identifier unchanged, target-language display label approved. Translate definitions carefully, preserve permitted values and check field types. Maintain a glossary by property ID so future languages map to the same concept rather than to each other’s approximations.

Practical decision check: If a localized field is re-exported, will the software still recognize the original property? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

9. Maintain EPD validity and version dates

A declaration may have an issue date, validity period, amendment history and replacement edition. A product specification may refer to a particular revision rather than the most recent file. Translation should not automatically treat an expired or superseded declaration as current.

Preserve registration, version, validity and retrieval date separately. If a source page says an EPD has been replaced, note that fact without silently rewriting the historical report. If a procurement decision relies on a valid declaration, the procurement reviewer must check the actual program record.

Practical decision check: Does the recipient know which version of the data was used and whether it was valid at the decision date? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

10. Connect BIM objects to the correct EPD record

A BIM material or product instance may reference an EPD, a generic dataset, a supplier-submitted file or a manually entered proxy. Those sources carry different degrees of product specificity. A translated description must not make a generic placeholder look like a product-specific verified declaration.

Keep each object’s classification code, type, quantity takeoff basis, EPD reference, fallback-data flag and revision aligned. If several product variants share a BIM family, check that each instance points to the intended source. A software-friendly export is not necessarily an environmentally correct mapping.

Practical decision check: Is the referenced declaration matched to the actual product type and the intended measured quantity? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

11. Differentiate generic LCA data from product declarations

Generic life-cycle inventory or impact data can support early-stage analysis when product-specific EPDs are unavailable. It is not an identity statement about any particular supplier. Translating generic dataset labels as if they were verified product records makes procurement conclusions misleading.

State the data’s generic nature, geography, time basis, modeled assumptions and intended use where provided. Avoid inserting a manufacturer’s name into a generic row. When a project later replaces a proxy with a specific declaration, document the change and reassess comparisons.

Practical decision check: Would the target-language reader clearly see that the estimate is generic rather than product-specific? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

12. Avoid unsound environmental comparisons

Two EPDs may use different product categories, declared units, functional assumptions, geographic conditions, modules or program rules. It may be invalid to conclude one product is lower-impact merely because the numerical GWP cell is smaller. Translation can accidentally create a comparative claim by deleting the qualifier that makes one figure noncomparable.

When preparing a comparison, state the exact common basis and unresolved differences. Keep original units and modules visible. If the source provides numbers without comparability analysis, translate the numbers, not a new ranking. Direct comparisons require appropriate specialist methodology.

Practical decision check: Are the two reported figures actually comparable, or did the translation conceal a scope mismatch? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

13. Handle circularity and end-of-life scenario descriptions

Recycling potential, reuse scenarios, waste-processing assumptions and disposal routes may appear in environmental reporting. A stated scenario is not a guarantee that the product will be recycled at a real construction site. The translation must separate assumed treatment from a documented operational outcome.

Translate assumed scenario verbs as modeled, assumed or forecast where appropriate. Preserve boundaries, allocation assumptions and whether effects are separately declared. Marketing teams should not turn a modeled possibility into a promise of recovery without separate evidence.

Practical decision check: Is the described material flow hypothetical, planned or recorded as actually completed? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

14. Check the receiving BIM and procurement workflow

A translated template may pass linguistic review but fail in the BIM authoring tool because of text expansion, unsupported characters or changed enumerated values. Conversely, a technically valid data import may show a misleading translated tooltip to the person making a product selection.

Test import and export, property IDs, field types, units, product references, null values, hyperlinks and accessibility. Ask an environmental-data specialist to inspect samples with negative numbers and missing modules. Confirm the consumer can reconstruct every product claim from a controlling declaration.

Practical decision check: Can both software validation and human meaning checks pass without altering the source environmental evidence? Identify the value, its unit, its declared scope and the source document or machine field that controls it. Then have a second person compare that information with the target text. A correct translation never introduces an unsupported environmental advantage or a missing numerical observation.

Bilingual EPD and BIM terminology crosswalk

These explanations are concept-level English anchors for multilingual term selection. A construction product’s official declaration, data schema and local environmental rules control the final technical wording; a general dictionary cannot establish the correct property type or declared boundary.

1. Environmental Product Declaration (EPD)

A structured declaration of environmental information based on an applicable program and methodological basis, not a blanket endorsement of a product’s sustainability.

2. BIM

Building information modelling, including structured information associated with building elements and products; not merely a three-dimensional picture.

3. Data template

A structured set of properties, definitions and data types intended to support interoperable product information.

4. Property identifier

A stable machine-recognizable reference to a property; translating it casually may break mappings.

5. Display label

The human-facing description of a property, which can often be localized without changing its identifier.

6. Declared unit

The quantity to which EPD results refer, whose type and magnitude must remain unchanged.

7. Functional unit

A defined function and reference performance basis used for analysis where applicable, not automatically equal to a product count.

8. Life-cycle module

An identified stage or part of a life-cycle assessment result; a label must retain its source boundary.

9. GWP

Global warming potential, an environmental indicator commonly expressed in CO₂-equivalent units under a specified method and scope.

10. Impact indicator

A quantified environmental result category with a defined unit, method and interpretation.

11. Reference service life

A documented assumption or parameter for use-stage considerations, not an unconditional promise about actual product life.

12. Generic data

A non-product-specific dataset selected to represent a class or process under specified assumptions.

13. Product-specific data

Information attributed to a particular declared product or defined product family and scope.

14. Not declared

A reporting status showing no result was provided for a field or module, not a numerical zero.

15. End-of-life scenario

An assumed or evaluated treatment pathway that does not by itself establish what actually happens to a product.

16. Comparability

The extent to which two results can validly be compared under compatible scope, unit, method and assumptions.

Twenty-four original worked EPD translation cases

Each invented example uses a source phrase, a plausible but incorrect interpretation and the faithful English meaning to preserve in other languages. None is reproduced from ISO’s copyrighted requirements. Quantities, codes and claims should be checked in the controlling data source.

Case 1: Declared unit mismatch

Example source: “GWP result reported per 1 m³ of product.”

Misleading translation: “The value is per one tonne.”

Faithful English sense: “The figure is referenced to one cubic metre of the declared product.”

A mass-based denominator and volume-based denominator are different. A conversion may require verified density and other assumptions.

Evidence and QA action: Confirm original unit and data type. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 2: Product family boundary

Example source: “Declaration covers panel grades P1 and P2 only.”

Misleading translation: “All panel grades are covered.”

Faithful English sense: “Only grades P1 and P2 fall within the declaration’s stated product coverage.”

A product-family claim must not expand to grades outside the declared scope.

Evidence and QA action: Compare manufacturer product list. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 3: Missing module result

Example source: “Module C4: not declared.”

Misleading translation: “Module C4 has zero impact.”

Faithful English sense: “No result is declared for module C4.”

Not declared is a data-status statement, not a quantified environmental result.

Evidence and QA action: Check EPD table and null encoding. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 4: Generic early-design dataset

Example source: “Generic insulation dataset used until supplier EPD available.”

Misleading translation: “Supplier insulation verified by an EPD.”

Faithful English sense: “A generic insulation estimate is used pending suitable supplier-specific data.”

A proxy data source cannot be presented as manufacturer-specific verification.

Evidence and QA action: Review source metadata. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 5: Declared product vs project

Example source: “Result applies to one declared product unit.”

Misleading translation: “The whole building emits this amount.”

Faithful English sense: “The value is per declared product unit, not a whole-building total.”

A product result requires quantity takeoff and proper methodology before aggregation.

Evidence and QA action: Check BIM quantity basis. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 6: Manufacturing module

Example source: “Reported GWP includes product stage modules A1–A3.”

Misleading translation: “GWP covers full service life.”

Faithful English sense: “Only the identified product-stage modules are included.”

Life-cycle module labels determine the boundary and must not be silently broadened.

Evidence and QA action: Compare EPD module matrix. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 7: Comparative uncertainty

Example source: “Data cannot be compared directly with declarations using different assumptions.”

Misleading translation: “Product A is definitively greener.”

Faithful English sense: “The differing assumptions prevent a direct comparison.”

Comparative confidence requires methodological compatibility, not a quick ranking.

Evidence and QA action: List units and module differences. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 8: EPD validity

Example source: “Declaration valid through 31 March subject to program rules.”

Misleading translation: “Declaration permanently valid.”

Faithful English sense: “The stated validity runs through 31 March under applicable program rules.”

A date-limited declaration is not a perpetual certificate. Source verification may still be needed.

Evidence and QA action: Check version, year and program record. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 9: BIM property key

Example source: “Property ID p-7ac corresponds to thermal conductivity.”

Misleading translation: “Rename key as ‘heat performance’ in the database.”

Faithful English sense: “Keep p-7ac as the property identifier and localize its displayed description only.”

A machine key cannot be freely translated without breaking lookup.

Evidence and QA action: Run import/export property lookup. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 10: Negative value

Example source: “Reported module value is −1.25 kg CO₂e.”

Misleading translation: “Reported value is +1.25 kg CO₂e.”

Faithful English sense: “The reported value has a negative sign and the stated CO₂e unit.”

Loss of sign reverses the numeric meaning, regardless of what the underlying model represents.

Evidence and QA action: Compare raw numeric cell. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 11: Scientific notation

Example source: “Impact result shown as 2.4E−3 per declared unit.”

Misleading translation: “Impact is 2,400 per declared unit.”

Faithful English sense: “The numeric result uses exponent minus three.”

Exponential notation must survive locale conversion intact. Decimal or exponent errors can be very large.

Evidence and QA action: Perform numeric parse verification. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 12: Recycled-content claim

Example source: “Supplier states recycled content is approximately 25%.”

Misleading translation: “EPD verifies exact 25% recycled content.”

Faithful English sense: “The supplier reports an approximate recycled-content figure.”

Supplier assertion, numerical precision and EPD verification are separate questions.

Evidence and QA action: Trace claim to its source. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 13: Modeled recovery

Example source: “End-of-life scenario assumes material recovery.”

Misleading translation: “Every installed product will be recycled.”

Faithful English sense: “The modeled scenario includes assumed material recovery.”

A modeled pathway does not guarantee real-world collection or processing.

Evidence and QA action: Review scenario assumptions. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 14: Planned replacement

Example source: “Product substitution under consideration; new EPD not yet linked.”

Misleading translation: “Approved product already substituted.”

Faithful English sense: “The substitution is under review, with no linked replacement EPD yet.”

A planned design choice is not an implemented change. Product mapping is still incomplete.

Evidence and QA action: Check BIM revision record. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 15: Programme operator

Example source: “Declaration published under Program X; verification conditions stated separately.”

Misleading translation: “Program X endorses the product as best.”

Faithful English sense: “The declaration follows Program X, without an added superiority claim.”

Publication or verification under a program is not automatically a ranking.

Evidence and QA action: Read declaration’s verification note. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 16: Different geography

Example source: “Dataset represents production in region R, not region S.”

Misleading translation: “Data accurately describe production in all regions.”

Faithful English sense: “The dataset’s geographic scope is region R.”

Geography can affect the underlying model and comparability. A translation must not universalize local data.

Evidence and QA action: Check geography metadata. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 17: Reference service life

Example source: “Reference service life assumed to be 30 years under stated conditions.”

Misleading translation: “Product guaranteed for 30 years.”

Faithful English sense: “The model assumes a 30-year reference service life under specified conditions.”

An assumption for modeling is not a product warranty.

Evidence and QA action: Read conditions and service-life field. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 18: Undeclared parameter

Example source: “Water-use indicator omitted from supplied EPD.”

Misleading translation: “Product uses no water.”

Faithful English sense: “The supplied declaration contains no value for that indicator.”

Absence of reported data does not establish zero consumption.

Evidence and QA action: Check complete declaration. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 19: Manufacturer naming

Example source: “EPD applies to manufacturer M at Plant 2.”

Misleading translation: “EPD applies to every M factory.”

Faithful English sense: “The declaration refers to the specified manufacturing site.”

Plant scope matters for sourcing and impact data. Entity names alone may be too broad.

Evidence and QA action: Check facility identifier. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 20: Two products comparison

Example source: “A and B have GWP values with different declared units.”

Misleading translation: “B is conclusively lower impact.”

Faithful English sense: “The figures cannot support a direct comparison without harmonizing the basis.”

Numerical size alone is not a cross-product conclusion.

Evidence and QA action: Compare units and declared function. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 21: BIM tooltip

Example source: “The object contains a reference to EPD-100; document not attached.”

Misleading translation: “The model contains a verified full EPD.”

Faithful English sense: “The object has an EPD reference but no attached declaration file.”

A pointer is not proof that the document was present, valid or checked.

Evidence and QA action: Open referenced file link. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 22: Dataset update

Example source: “Life-cycle dataset revised on 1 September; previous result retained in archived model.”

Misleading translation: “All models now use the new results.”

Faithful English sense: “The dataset changed on 1 September, but archived models still retain older results.”

Data updates do not automatically rewrite controlled project versions.

Evidence and QA action: Compare file and dataset version IDs. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 23: Carbon neutral wording

Example source: “Product impact estimated; no carbon-neutral claim made.”

Misleading translation: “This product is carbon neutral.”

Faithful English sense: “The source only provides an impact estimate and makes no neutrality claim.”

A translator may not add a high-stakes environmental marketing assertion because a GWP value exists.

Evidence and QA action: Look for an explicit substantiated claim. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Case 24: Module D qualifier

Example source: “Potential effects beyond the system boundary reported separately.”

Misleading translation: “Subtract module D from every result to get net zero.”

Faithful English sense: “The separately reported beyond-boundary effects require their own interpretation.”

Separate module treatment and methodological assumptions cannot be erased by a simplistic arithmetic rewrite.

Evidence and QA action: Check declaration and comparison method. Test that a reader receives the same evidence strength, product identity, value, module boundary, measurement unit and date as in the source. Any missing declaration field remains missing; do not fill it with a convenient value or introduce an unreviewed calculation.

Eleven-stage data translation and release workflow

1. Locate the controlling declaration

Collect the full EPD, registration reference, product coverage, publisher, relevant program metadata and any approved machine dataset. Avoid translating marketing snippets without confirming the original environmental evidence.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

2. Map structured property fields

For each BIM property, list stable identifier, data type, unit, display label, source definition and permitted values. Separate machine keys that cannot change from human descriptions that can be localized.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

3. Protect numerical data

Lock numbers, units, indicator abbreviations, negative signs, exponents, reference quantity and module code values. Use automated comparisons where possible before any file is re-imported.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

4. Confirm product scope

Check plant, product type, grade, family, geography and declaration validity. A target-language name must not make one product’s EPD apply to a different item.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

5. Preserve evidence status

Distinguish verified data, supplier-provided statements, generic assumptions, proxy datasets, missing values and modeled scenarios. Translate the status explicitly rather than letting readers assume all data are equally established.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

6. Review the declaration’s boundaries

Check life-cycle modules, data coverage and declared or functional units. Do not add, sum or compare environmental results without a competent LCA reviewer and suitable methodology.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

7. Test translated term choices

Create a bilingual terminology table with approved target language, disallowed false friends, property ID and example sentence. Focus on scientific terms, qualifiers and environmental claims.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

8. Run software validation

Import the translated data into a test BIM environment and inspect property mappings, category filters, null values, data types and hyperlinks. A localization that breaks software joins is not complete.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

9. Review human decisions

Ask a product specifier, procurement reader and sustainability specialist whether they infer the same product coverage, value and caveats from the source and target text.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

10. Check claims and summaries

Audit executive summaries, tooltips and marketing copy for unsupported statements of certification, neutrality, superior performance or guaranteed recycling. Keep distinctions between observations, models and promises.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

11. Release versioned outputs

Record data version, locale, translation reviewer, technical sign-off, correction route and links to the original EPD. A future product or declaration revision needs a controlled update, not silent overwriting.

For this stage, record one real check on an invented practice dataset: identify the exact source column or source paragraph, preserve its unique reference and explain what kind of false conclusion a poor translation would create. Keep the human-language output and raw machine data available for an independent reviewer.

Frequently asked questions

Does ISO 22057:2022 assess the sustainability of an entire building?

No. Its official public scope concerns data templates that let EPD information be used in BIM. Whole-building environmental assessment is outside the scope described by its ISO catalogue abstract.

Can EPD values be translated into another language without changing units?

Yes, but translate descriptive labels separately from numeric quantities and machine keys. Preserve units, reference quantity, indicator meaning, life-cycle modules and applicable assumptions.

What is the difference between an EPD and a marketing claim?

An EPD reports environmental information under a specified program and method. A claim of being best, carbon neutral or universally greener needs further substantiation; it is not a default conclusion from the existence of an EPD.

Should missing module values be entered as zero?

No. A missing or not-declared field is not a zero observation. Retain source status and encode it correctly in machine-readable datasets.

Can I rename a BIM property ID in the target language?

Keep the stable property ID intact unless the authorized data schema explicitly changes it. Translate the display label and descriptive metadata through a controlled locale layer.

Are two products comparable when their GWP values differ?

Only if the units, modules, product functions, methodology and other assumptions permit a valid comparison. Translators should not generate a product ranking from incomparable table cells.

What if an EPD has expired?

Translate the historic record accurately but indicate version and validity where documented. The responsible procurement or sustainability team must confirm whether it is suitable for a current decision.

Can an AI system translate a dataset?

AI may help with display labels in an authorized environment, but structured field mappings, units, numbers and evidence claims require systematic verification. Never use translation to modify source measurements.

What should a translated BIM object display to a buyer?

Show intelligible product description together with the controlling identifier, EPD reference, data source, declared unit, version and any limitations needed for a correct procurement decision.

What quality assurance is most important?

Validate semantic equivalence and machine interoperability together. A translation can read well while linking the wrong product, changing units or converting missing data into zero.

Four-week learning and QA plan

Week 1: Data and vocabulary

Study a fictional EPD table with sixteen indicators, module labels and one declared unit. Build a glossary and mark machine identifiers that cannot be translated. Render human-facing labels in a second language while preserving numerical meaning.

Keep an audit trail with a source excerpt, chosen target-language term, evidence boundary, reviewer comment and correction. At the end of the week, ask a different reader to state which values were reported, which were absent and which comparisons are not authorized. That is a stronger test than surface fluency alone.

Week 2: Scope and module discipline

Work through examples of product-family coverage, geographic limitations, undeclared modules and generic proxy data. Translate each claim, then ask an independent reader to name its precise boundaries. Correct overgeneralized or unsupported comparisons.

Keep an audit trail with a source excerpt, chosen target-language term, evidence boundary, reviewer comment and correction. At the end of the week, ask a different reader to state which values were reported, which were absent and which comparisons are not authorized. That is a stronger test than surface fluency alone.

Week 3: BIM interoperability

Create a mock product template with IDs, localized display fields, declared units, dates and data source links. Import and export it through a test spreadsheet or BIM-compatible schema. Find changed keys, missing values and incorrect type coercion.

Keep an audit trail with a source excerpt, chosen target-language term, evidence boundary, reviewer comment and correction. At the end of the week, ask a different reader to state which values were reported, which were absent and which comparisons are not authorized. That is a stronger test than surface fluency alone.

Week 4: Full evidence and decision review

Produce a bilingual two-page EPD interpretation sheet plus a small model object mapping. Have a competent environmental-data reviewer challenge units, assumptions, claims and version status. Record defects and approved corrections before release.

Keep an audit trail with a source excerpt, chosen target-language term, evidence boundary, reviewer comment and correction. At the end of the week, ask a different reader to state which values were reported, which were absent and which comparisons are not authorized. That is a stronger test than surface fluency alone.

Scope, sources and connections

This is independent translation education concerning the public scope of ISO 22057:2022, not a reproduction of standard clauses, official conformity guidance or a whole-building life-cycle assessment. Environmental declarations and their use are subject to relevant program rules, current product data, applicable jurisdictions and qualified professional review. The publisher’s official catalogue was checked on 5 November 2026.

Source: ISO 22057:2022 — official ISO catalogue. The catalogued successor work item is under development; use the currently applicable controlled edition rather than treating a draft as a published replacement.

This guide addresses a narrow EPD-to-BIM translation problem within the existing ecosystem. Read The Master Art of Translation for the general meaning-transfer architecture, the protected Vocabulary Learning Hub for word learning, and How English Works for sentence structure and semantic precision. None of these established owners is replaced here.