Skip to content

Every figure on this page is read from the reference tables the product runs on — nothing here is typed in by hand.

What it does

Carbon as a design constraint,
with the evidence attached.

CarbonSpec ranks construction materials on four axes at once and shows the data point, standard and date behind every score. It is decision support for professional judgment, never a certification — and the figures below are the reference tables it actually reads, counted as this page loaded.

No account needed: get a project in · compare two EPDs · compare design options · match reclaimed steel · substitutions and value engineering · what it does and does not do

What it does

Gates before scores

On a higher-risk building, a material that cannot meet Regulation 7(2) is removed before anything is compared — named, with the rule and the citation. No weighting reaches across a regulatory gate, so it is not ranked low, it is out.

Four axes, and no invented order between them

Health, fire and building safety, embodied carbon and durability, weighted per project type. Where one option is at least as good on all four it is preferred outright; where each beats the others on something, the conflict is reported rather than settled by an arithmetic nobody agreed to.

Every score names its evidence

Each axis returns the data points behind it, the standard they came from and the date. A number you cannot trace is a number you cannot defend to a client.

Says when it does not know

A material nobody has assessed is capped and labelled, not given full marks. Absence of evidence is reported as absence, never as a clean result.

Prices carbon in pounds, properly

Carbon-adjusted future cost separates a CBAM liability from the societal value of emissions, because a border charge and an appraisal value are not the same money.

Checks the paperwork is still valid

An EPD lapses after five years. The scorer says when a figure is no longer backed by a current declaration, and how common that is in its category.

Quantifies reuse against real production

Reclaimed steel, brick, aggregate and surplus stock are listed with their condition, location and lister — demolition contractor, processor or stockholder — and what each avoids is banded from published declarations, so a circular listing is compared with what a new one actually costs.

Says how much of the building it covers

A concept-stage figure leads with the share of the schedule it is built from, and every line it could not count says what would bring it in. A 60%-covered estimate that says 60% is worth more than a 100% estimate that is quietly 60%.

What you get

3 documents you can hand to a client and 3 screens you work in. Two of them refuse to issue under conditions they state — which is the promise being made about every figure they do produce.

Concept-stage carbon statement

PDF

The concept assessment as a document somebody else can read without the interface around it. Coverage comes before the range, because a reader who takes one figure away should take away the qualified one.

For: The Stage 2 → Stage 3 handover, and the project director who will read one paragraph of it

  • Coverage — what share of the schedule carries a figure, stated first
  • The range, in kgCO₂e/m² and in tonnes, against the target and its declared module scope
  • Biogenic carbon reported separately, never netted into the total
  • The breakdown by NRM element, largest first
  • Constraints already closed by regulation, where the project is a higher-risk building
  • The asks, ordered by how much of the schedule each one would bring in
  • Basis and limits — including that the ranges cover A1–A3 only

Refuses: Will not issue without a schedule, or without a floor area. Every figure in it is per square metre, and issuing against a stand-in denominator would put a wrong number into the one artefact whose purpose is to be handed to somebody else.

Material selection record

PDF

Why these materials and not the others. Exclusions come first, because the sentence that changes what gets specified is “this is not permitted here”, not a score.

For: The client, the design team record, and whoever asks in two years why this was specified

  • Excluded before scoring — each with the rule, the requirement, the evidence and the citation
  • In contention — the materials nothing beats outright, deliberately not numbered
  • Per axis: the score, the confidence, the rationale, the data points and the sources behind it
  • Supplier intelligence per material, with CBAM exposure and carbon-adjusted cost
  • Beaten outright — the one ordering in the document that needs no judgement

Refuses: Prints no position numbers against the materials in contention. Where each option beats the others on something, any order encodes a preference, and a “1.” is read as a verdict however carefully the surrounding text is worded.

Single-material report

PDF

The same record narrowed to one material — including one the gates removed, because “why can’t I use this?” is a fair question whose answer is worth exporting.

For: The specifier defending one choice, or querying one refusal

  • For a permitted material: every axis with its evidence, and its supplier options
  • For an excluded one: the rule, the requirement, what this material's evidence showed, and the instrument it rests on

Concept assessment

On screen

The live version of the statement, with the asks answerable in place — supply a density and the coverage rises and the range narrows as you watch.

For: The architect, while the build-ups can still change

  • The range and the coverage, with the verdict against the target
  • The element breakdown, by share
  • The density asks, grouped so six lines blocked on one figure are one question
  • Lines naming a material the library does not recognise, with the classes it does

Refuses: Holds the whole assessment back until there is a floor area, rather than computing it against a stand-in.

Evidence standing

On screen

Which of this project's materials loses its evidence before anyone audits the building — the check nobody runs, because nothing in the industry sends a reminder.

For: Whoever will be asked, at handover, to show the figures still hold

  • Every EPD judged against the project's completion date, not today
  • Expired, expiring within three months, and never issued — counted separately, because they call for different actions
  • Per-axis coverage: how many materials carry what each axis claims to assess

Refuses: Runs no check at all without a completion date. Defaulting to today would pass every declaration that is current now and expires before handover — precisely the set this exists to find.

Procurement leakage

On screen

Design intent against what was actually bought, with the drift quantified — the part of a project's carbon that no design-stage report can see.

For: The team that has to explain the gap between the design and the building

  • Carbon, cost and compliance drift per substitution
  • Completeness stated on every total, computed by the impact engine shared with TrueGauge

How to work through it

A project holds 5sections, tabbed in the order the work happens rather than the order they were built. Nothing is gated — you can open any of them at any time — but each says what it needs, and withholds its answer rather than guessing when it does not have it. This list is generated from the product’s own navigation, so it cannot describe a tab that no longer exists.

  1. 1

    Concept

    RIBA Stage 2

    Paste the area schedule. The range comes from the published EPD distribution of each material class, because at concept stage you have chosen a class and not a product.

    You give it
    An area schedule, a floor area, and a density for any line measured by area or volume
    You get back
    A range in kgCO₂e/m², the share of the building it covers, and the questions that would close the rest
  2. 2

    Shortlist

    Stage 3–4

    Materials are gated on regulation first, then what survives is compared. Where one beats another on every axis that ordering is real; where each wins somewhere, the trade is the finding rather than a score.

    You give it
    The project type and whether it is a higher-risk building
    You get back
    What is prohibited and why, what is genuinely in contention, and what is beaten outright
  3. 3

    Carbon budget

    Stage 3 onwards

    Spend against a target while the design is still moving, rather than measuring after it is finished. Undecided routes are carried as ranges.

    You give it
    Quantities per building element, and a target
    You get back
    Spend against budget by element, with the decisions that still determine the outcome
  4. 4

    Evidence

    Any stage

    An EPD is valid five years from issue and nothing in the industry sends a reminder. Declarations are judged against your completion date, not today.

    You give it
    A practical completion date
    You get back
    Which materials lose their evidence before handover, and what each axis can actually see
  5. 5

    Procurement

    Stage 5

    What was specified against what was bought. The drift introduced after design is the part no report usually catches.

    You give it
    What was actually purchased
    You get back
    Carbon, cost and compliance leakage, quantified per substitution

Start at Concept even if the project is past Stage 2 — it is the only section that takes a whole building at once, and the coverage figure it produces tells you how much of the rest is worth doing. See it run on a real schedule before you sign in.

The 4 axes it scores on

Weights are rubric data in the database, not code, so the balance can be tuned per project type without a deploy. Each axis returns a score, a confidence and the data points that produced them.

  • Health / Chemical Hazard
  • Fire & Building-Safety
  • Embodied Carbon
  • Durability

What it knows

Reference tables the scorer reads, counted from the modules themselves.

Carbon, benchmarked twice
80 bands across 9 categories from German EPDs, and 64 more from Swiss and European LCA work on the same method (EF 3.1).
5 of 8 categories corroborate each other; the 3 that diverge each say why.
Hazard, from the authority lists
The UK REACH Candidate List (334 entries with their Article 57 grounds) and Annex XIV, the Authorisation List (97 entries with sunset dates).
89 of them appear in the inventories held here, matched by CAS number.
EPD validity
7,025 published declarations, banded across 11 categories, so a lapsed EPD can be put in context.
32.6% of them are already past validity — 71.4% for coatings, the worst category.
Official carbon values
31 years of UK Government appraisal values, 2020–2050, with the GDP deflator that restates them into any price year.
£277.22/tCO₂e central for 2026. These price the societal term, never a border charge.
Freight factors
3 modes from the DESNZ conversion factors, each naming the exact published row it came from.
Vehicle and vessel choice moves road 9.0-fold, so the variants are carried rather than averaged away.
Reuse, against real production
What new production emits, banded per category across 10 categories, so a circular listing has something real to be compared with.
Module D is carried beside it and never added — it is the material's own end-of-life credit, not a second benefit of buying second-hand.
Class typicals
15 construction classes carrying what the class is normally like on fire, service life and maintenance.
Shown only where a product's own figure is missing, so a gap reads as a gap rather than as a verdict.

What it refuses to do

Each of these is a number the tool could produce and does not, because producing it would be wrong in a way nobody would notice.

No ordering is invented where none exists

Weighting fire safety against embodied carbon is a judgement, not an arithmetic, and ISO 14044 keeps that judgement with the reader. What survives the gates is split into what is beaten outright — a real ordering, needing no preference — and what is genuinely in contention, which is listed without a winner.

A partial total can prove an exceedance and never compliance

Lines that could not be quantified only ever add, so a concept figure is a floor. Exceeding a target on a floor is a finding; sitting under one is not, and no verdict is issued until the schedule is complete.

Biogenic carbon is reported, never deducted

Timber's A1–A3 total is negative at every percentile because the uptake is booked there and the release falls in C3/C4. Netting it turns a CLT scheme into a near-zero building on paper, so storage is stated as a claim and kept out of the total.

Module D is never netted into a total

The end-of-life credit is reported apart from the production burden. Summing them counts a future benefit against a present one and flatters reuse twice — EN 15804 forbids it and the code enforces it.

Two carbon prices are kept apart

A CBAM liability is money handed over at a border; the Green Book value is what a tonne is worth to the country. Nobody pays the appraisal value to customs, so one figure cannot do both jobs.

Two benchmark sources are never averaged

German EPDs and Swiss LCA data are held side by side on the same method. Where they disagree the reason is named. Averaging them would produce a third number nobody published.

A substance in an inventory is not a substance in a product

REACH Article 33 bites at 0.1% by weight in an article. An elementary flow is usually an emission, and treating one as the other would manufacture a legal obligation.

What it cannot answer yet

Rendered from the same list the roadmap is planned against, so this section cannot quietly fall out of date while the product moves on.

What is missingStatusWhat fills it today
carbongwp_per_functional_unitsuppliedÖkobaudat EPD bands across ten categories, plus a second independent distribution from BAFU:2026 on the same method, so a figure can be positioned and the positioning corroborated.
hazardsvhc_reach_flagreference heldBoth authority lists, wired into scoring: the UK REACH Candidate List (334 entries, with Article 57 grounds and listing dates) and Annex XIV, the Authorisation List (97 entries, with sunset dates). 89 of them appear in the inventories held here. A substance a manufacturer declares is checked against both and named in the rationale — and a material with no hazard evidence at all is now capped rather than scoring full marks, so never assessed no longer looks like clean.
fireeuroclass_ratingclass typicalThe class typical from materialClasses.ts, shown only where a product's own rating is absent, so a gap reads as a gap rather than a verdict.
durabilityexpected_service_life_yearsclass typicalA range per class from BS 7543 and ISO 15686 practice, and — since `durability.ts` — a statement of which classes that range actually settles. Over a 60-year study, concrete, brick and structural steel are never replaced at either end of their range, so the class figure answers the question and the product's own life adds nothing. Paint spans 3 to 11 replacements and sealant 2 to 5, so for those it answers nothing. The gap below is therefore narrower than it looks: it is real for the twelve straddling classes and closed for the three that do not.
healthvoc_emission_classmissingNull. The axis scores on a fallback wherever it is unset.

Carbon is supplied twice over and independently corroborated. Hazard now holds the authority list — the UK REACH Candidate List, bridged by CAS to the inventories here — which is half the axis: a declared substance can be checked, but nothing yet says what a given product contains. Fire and durability lean on class typicals that are right for a class and are not evidence for a product. VOC has nothing. What remains is no longer an extraction problem and mostly not an acquisition one either: fire ratings, VOC classes and product composition are per-product declarations held by manufacturers, and the route to all three is the same supplier-declaration intake rather than another database.

4 of the 5 fields behind those axes are still open. The one ranked next is svhc_reach_flag, and what it needs is not another dataset: the remaining half is not a data-acquisition problem at all. Nothing published says which SVHCs a given construction product contains — that is the manufacturer's declaration, and the route is the supplier-declaration intake TrueGauge already has.

Commercial figures — supplier prices, lead times, the CBAM liability rate — remain assumed demonstration values, and are labelled as such wherever they appear. No public source publishes real UK supplier quotes, and inventing them would make the one number a specifier could check the one number they could not trust.

Where to start

The worked example runs the real modules on a messy schedule and needs no account — it is the fastest way to see whether the output is the shape you need. The product itself keeps your schedule against a project and lets you close the asks one at a time.