EPDs & Carbon Data

An Environmental Product Declaration is a standardised report of a product's environmental performance across its life cycle, based on an underlying LCA and presented to fixed rules so it can be compared. In the language of the standards it is a Type III environmental declaration (ISO 14025), which is the most rigorous of the three types: unlike a vague 'eco-friendly' claim (Type II self-declaration) or a pass/fail eco-label (Type I), a Type III declaration reports quantified data across many impact categories and, crucially, is independently verified by a qualified third party. That verification is what separates an EPD from marketing: someone other than the manufacturer has checked that the numbers were calculated correctly and honestly.

Three things give an EPD its authority, and you should look for all three. First, a common rulebook: construction EPDs are made to EN 15804, which defines a Product Category Rule (PCR) for each product type so that, say, two insulation products are assessed with the same boundaries, stages and assumptions - without a shared PCR, two EPDs are not comparable even if both are honest. Second, third-party verification: an accredited verifier and an EPD programme operator (the body that publishes the EPD) stand behind it. Third, transparency: the EPD states its boundary, its declared unit, its data sources and vintage, and reports impacts module by module, so you can see exactly what it does and does not include.

Why does this scaffolding matter to a designer? Because it is the only thing that turns a carbon figure from an assertion into evidence. When a product's carbon comes from a verified EPD to a common PCR, you can put it beside a competitor's EPD and make a defensible comparison; when it comes from a sales sheet, you cannot. EPDs are the reason the modules and boundaries of 2.2 are usable in practice - they are where the trustworthy, module-by-module numbers actually live. They are not perfect (they carry uncertainty, and generic ones can be broad), but they are the recognised, verifiable data source, and using them - rather than unverified figures - is a basic mark of carbon-literate practice. When a binding number matters, it should trace back to a verified EPD, read with the eyes this lesson is training.

Open an EPD and it can look like a wall of tables; the skill is knowing the handful of fields that decide what the numbers mean. Start with the declared or functional unit - is this per kilogram, per cubic metre, per square metre, and (for a functional unit) delivering what performance for how long? Every figure in the document is *per that unit*, so misreading it is the fastest way to a wrong answer. Next find the system boundary and the modules reported: many product EPDs are cradle-to-gate (A1-A3) only, and you must know that so you remember to add transport, installation, replacement and end of life yourself where they matter (2.2). An EPD that reports A1-A3, A4-A5, B, C and D separately is telling you a fuller story - read the module table, do not just grab the top-line number.

Then go to the impact results, where your headline is the global warming potential figure in kg CO2e per declared unit (often now split into fossil, biogenic and land-use GWP - a nuance that matters for timber). Note that a full EPD reports other categories too (acidification, eutrophication, water use); carbon is one column (2.1). Check the reference service life if quoted, because it drives how often the product is replaced over a building's life. Finally, read the metadata: who made the EPD, to which PCR, verified by whom, published and valid until when. EPDs expire (typically after around five years) because data and processes change; an out-of-date EPD is a warning sign.

Two habits make this reliable. First, read the scope before the number - exactly as in 2.1 - so you never quote a figure without knowing its unit, boundary and vintage. Second, compare only within a PCR and boundary: two EPDs for the same product category, same modules, same declared unit, are comparable; anything else must be aligned first. If that sounds like work, it is - which is why for a binding project comparison you brief a specialist to select and reconcile the EPDs. But you should be able to open any EPD and, within a minute, state its unit, its boundary, its carbon figure and whether you would trust it. That single competence makes you a far more dangerous - in the good sense - specifier.

Not all EPDs are equal, and the most important distinction is between generic and product-specific data. A product-specific EPD describes one named product from one manufacturer (sometimes one factory), built on that maker's own production data. A generic or industry-average EPD (sometimes a sector or association EPD) describes a typical product of a category - 'average Indian OPC cement', say - built from pooled or representative data. Both are legitimate and both have their place, but they answer different questions and carry different uncertainty, and confusing them is a common error.

Product-specific data is more accurate and discriminating: it can reveal that Manufacturer A's steel is genuinely lower carbon than Manufacturer B's, which is exactly the signal you want when specifying, and it rewards manufacturers who actually cut their emissions. But it is only available where a maker has invested in an EPD, it applies strictly to that product, and it can be cherry-picked (a firm publishes an EPD only for its greenest line). Generic data is more conservative and robust: it will not single out a good product, but it gives a defensible typical figure when you do not yet know the supplier, which is most of early design. The professional pattern is to use generic data early, when you are comparing material strategies and have no supplier, and switch to product-specific EPDs late, once real products and suppliers are chosen and the differences between them matter.

There is a subtlety worth internalising: precision is not the same as accuracy. A product-specific EPD gives a precise number for one product, but if you apply it to a different product, or a different region's version, or assume the whole building will use that best-case line, your precise number is inaccurate for your actual project. Conversely a generic figure is imprecise but honestly so. Matching the data to the decision - and to your project's real context, which in India often means the data is thinner and more generic than in Europe - is itself a skill. And the deferral rule stands: which specific EPDs to use, how to reconcile generic and product-specific data, and how to handle the gaps are judgement calls for a qualified assessor. Your job is to know the difference exists, ask which kind a number is, and never mistake a best-case product-specific figure for what your whole building will actually emit.

Every carbon result inherits the quality of the data beneath it, and this is where good intentions most often produce bad numbers. Garbage in, garbage out is not a slogan here but the central risk of the whole method: a beautifully executed LCA built on wrong, outdated, mismatched or geographically irrelevant data produces a confident, precise, official-looking figure that is quietly meaningless - and arguably worse than no number, because it invites false certainty. The discipline of carbon work is therefore, above all, a discipline of data quality.

A few dimensions decide whether data is fit for purpose. Representativeness: does the data actually describe your product, from your region, made your way? A European EPD for a material may badly misdescribe the Indian version made with a different fuel mix or grid. Temporal validity: is the data current, or from a decade ago before a process changed? EPDs expire for exactly this reason. Technological match: same process route (electric-arc versus blast-furnace steel, for instance) - a mismatch here can swing carbon by large factors. Completeness and consistency: are the modules and boundaries the same as the rest of your assessment (2.2)? And underlying it all, provenance: is it verified third-party data, or an unverified figure someone typed into a spreadsheet? Good assessors record and, where the method requires, score these data-quality dimensions rather than hiding them.

The honest conclusion is a humble one, and it is the through-line of this whole course. Carbon numbers are estimates carrying real uncertainty, and their reliability depends entirely on the data behind them. That is not a reason to ignore them - a well-founded estimate is vastly better than a guess or a marketing claim - but it is a reason to treat any single figure with appropriate scepticism, to prefer verified EPDs over unverified numbers, to match data to context (especially in data-thin settings like India), and to be transparent about quality and uncertainty rather than presenting a number as more precise than it is. It is also exactly why binding assessments belong with a qualified specialist working to the standards and verified data: sourcing, judging and reconciling data is the hardest and most consequential part of the job. Master the habit of asking 'how good is the data behind this number?' and you have learned the thing that separates real carbon work from carbon theatre.