Concrete: The Hardest Line Item in Australian Scope 3
Two registered Australian EPDs, both for 32 MPa normal-class concrete, report 340 and 181 kg CO2-e per cubic metre. The delivery docket for either load says N32. That gap is why concrete, not fuel, is where Scope 3 reporting actually breaks.
Two Environmental Product Declarations, both registered with EPD Australasia, both covering 32 MPa normal-class concrete supplied in Australia.
Boral's Tasmanian pre-mix declaration puts Normal Class GP Blend 32 MPa at 340 kg CO2-e per cubic metre cradle to gate for the Hobart region. A Holcim declaration registered in April 2026, covering the Western Harbour Tunnel project mix designs in Sydney, puts its N32-20 normal-class product at 181 kg CO2-e per cubic metre across the same A1-A3 modules.
Same strength grade, same Australian standard, and close to a factor of two apart.
The delivery docket for either load says N32.
Those two numbers are not a fair head-to-head. Different states, different plants, different aggregate sources, different versions of the underlying product category rules, and the Holcim figure is a project-specific mix designed down for a tunnel. All true. It is also the point. Comparing two verified declarations properly takes an afternoon. Reading the docket takes three seconds and tells you nothing about carbon at all.
A fuel docket is the easy case, and we think the industry has been fooling itself by treating it as representative. One fuel, one unit, one factor from the NGA workbook, unambiguously Scope 1. Diesel at 2.7 kg CO2-e per litre. Concrete is the opposite on every axis, and it is usually the single largest line in a builder's Scope 3 Category 1 inventory.
What the docket actually tells you
Under AS 1379, the supplier issues an identification certificate with every load. Most people call it the docket or the ticket. It identifies the batch, and normal-class concrete is specified by a short set of parameters: characteristic compressive strength at 28 days (N20, N25, N32, N40, N50, N65), maximum nominal aggregate size, slump, and exposure classification.
That is a structural performance specification. It says the concrete will reach 32 MPa. It says nothing about how the supplier got there.
Here is the actual product code from that Holcim declaration: CONC,NSW,N32-20-T-R53,RMS. Grade 32, 20 mm nominal aggregate, then internal mix and product identifiers. Rich data, none of it carbon.
Strength grade does correlate with carbon, because higher strength generally needs more binder. Within Boral's Hobart range, 20 MPa sits at 291 and 50 MPa at 552, a spread of roughly 90% across the normal-class ladder. So grade matters. It just doesn't get you all the way there, and treating it as though it does is the mistake we come across most.
The carbon lives in the binder, and the binder is invisible
Cement is a small fraction of concrete by weight and most of its footprint. The Holcim declaration's content table gives general purpose cement at 5-21% by weight, aggregate at 67-84%, water at 11.6-12%, and supplementary cementitious materials at 0-11%.
Note what that last figure is measured against. SCM content is quoted as a percentage of the whole product, not of the binder. When a specifier says "30% fly ash replacement", they mean 30% of the cementitious fraction, which is somewhere around 4% of the concrete by weight. We have watched people transcribe one into the other and produce a number that is wrong by an order of magnitude before any factor is applied.
Fly ash, ground granulated blast furnace slag and silica fume are all treated as co-products under PCR 2019:14, with economic allocation. The environmental burden of fly ash is largely allocated back to the coal-fired electricity that produced it. That is the governing methodology, and it is also genuinely contested. If Australian coal generation retires on schedule, the fly ash supply and its allocated burden both change. We do not have a confident view on how that resolves, and anyone who tells you they do is guessing.
None of this appears on the docket. Two loads with the same grade, from the same supplier, batched at different plants with different SCM availability, are different numbers.
Same grade, different plant, different answer
The Boral Tasmanian declaration models plants in two regions and reports them separately.
| Boral TAS pre-mix, A1-A3 (kg CO2-e per m³) | Hobart | Launceston |
|---|---|---|
| Normal Class GP Blend 20 MPa | 291 | 290 |
| Normal Class GP Blend 25 MPa | 308 | 305 |
| Normal Class GP Blend 32 MPa | 340 | 337 |
| Normal Class GP Blend 40 MPa | 422 | 410 |
| Normal Class GP Blend 50 MPa | 552 | 541 |
| Envisia 25 MPa | 280 | 273 |
| Envisia 32 MPa | 300 | 292 |
| Envisia 40 MPa | 371 | 359 |
The Hobart-to-Launceston gap is small, a couple of percent. Tasmania is a small network. But look down the columns instead of across. In Hobart, Normal Class GP Blend at 32 MPa is 340 and Envisia at 32 MPa is 300. Same supplier, same region, same strength grade, 12% apart, because it is a different product line with a different binder.
The Holcim declaration names its production sites explicitly: Lidcombe, Alexandria and Artarmon. So "Holcim N32 in Sydney" resolves to a figure scoped to named plants and to one project's mix designs. Your docket records which plant the load came from. Almost nobody carries that field through into their emissions ledger.
Cubic metres in, tonnes out
Both declarations use one cubic metre of concrete as the declared unit. Most published material factors, and most of the general-purpose databases people fall back on, are expressed per tonne.
Concrete is ordered, delivered and invoiced in cubic metres. Bills of quantities sometimes carry tonnes. Weighbridge records definitely carry tonnes.
The Holcim mix declares a gross weight of 2,383 kg per cubic metre. The rule-of-thumb 2,400 kg/m³ that most people use is 0.7% off for that mix, which is fine. The failure mode that actually hurts is not converting at all: applying a per-cubic-metre factor to a tonnage quantity understates the result by roughly 2.4 times, and the reverse overstates it by the same. Both produce a number that looks completely plausible in a spreadsheet cell. We wrote about this class of unit conversion error separately, because it is the single most common cause of restatement we come across.
The EPD is a PDF, and it has an expiry date
Plant-specific and project-specific declarations do exist in Australia, and coverage for concrete is better than for almost any other construction material. Holcim publishes bespoke declarations for individual mixes. Boral publishes by region and product line.
They arrive as PDFs. The Holcim document runs 3.5 MB across dozens of pages, with the indicator tables sitting past page 20 in multi-column layouts. Getting 181 out of it means finding the right table, the right module column, and the right row. Multiply that by every mix on a project and you have an afternoon of research where you wanted a data feed. When we designed the document engine's handling of EPDs, this was the shape of the problem we were solving: the number exists, is third-party verified, and is effectively locked inside a document.
Then there is validity. The Tasmanian declaration we pulled carries a validity window that closed on 1 May 2026. Suppliers re-register, and a newer edition may well be live. But the copy sitting in someone's project folder is the copy that gets handed to the assurance provider, and a lapsed declaration used for a period it does not cover is an audit finding.
The same page of the Holcim declaration also reports GWP-Total for A1-A3 at 181 kg CO2-e and an additional AR5-aligned GWP-GHG line at 185. NGER runs on AR5 global warming potentials. AASB S2 requires AR6. On 40,000 m³ of concrete, that characterisation choice alone moves your reported figure by about 160 tonnes. Not huge, but you have to record which line you took, because your NGER return and your climate disclosure will not want the same one.
Why spend-based fails on this material specifically
Look at Envisia 32 MPa at 300 against Normal Class GP Blend 32 MPa at 340 in Hobart. A 12% physical reduction at the same structural performance.
Map dollars to an environmentally extended input-output factor and that reduction disappears. Every dollar spent in the ready-mix sector carries the same coefficient. Buying the lower-carbon product shows zero improvement in your reported number. If the low-carbon mix costs more per cubic metre, which it often does, your reported emissions go up while your physical emissions go down.
That is a method structurally incapable of measuring the thing you are trying to change. Spend-based estimation earns its place for professional services and IT, where activity data is meaningless and the amounts are small. On your largest material category, it is a placeholder.
The commercial pressure is heading the other way. In December 2025 the Cement Concrete & Aggregates Australia adopted the Global Cement and Concrete Association's Low Carbon Ratings framework for Australian conditions, with AA to G bands running across roughly 5 to 100 MPa. The ratings are calculated from third-party verified EPD data, A1-A3, in kg CO2-e per cubic metre, normalised against cylinder strength. A procurement team can now ask for a band on a tender. A spend-based inventory cannot answer that question, and will not be able to demonstrate improvement against it next year.
What NGER and AASB S2 each want from you
These are two different asks and they get conflated constantly.
NGER does not want your concrete at all. The scheme covers operational Scope 1 and Scope 2, and the NGA Factors workbook has no row for concrete. Fuels, grid electricity by state, waste, a set of industrial processes. Embodied carbon in a purchased material is out of scope by design. If you are a builder tripping the 50 kt corporate threshold, that is diesel and site power, not the slab.
AASB S2 is where concrete lands, in Scope 3 Category 1, with the categories and activity types disclosed and the methodology and inputs described. Group 2 entities get Scope 3 relief in their first year, which for most means the obligation bites in the financial year starting 1 July 2027. That sounds distant. It is one procurement cycle away, and the assurance provider in year two will walk two years of data.
Head contractors feel this first because they buy the most of it. They are not the only ones holding the problem. Property developers, councils funding civil works, water utilities replacing pipe and tank assets, and mining companies pouring pads and foundations all book concrete into Category 1, and all of them get their quantities from the same dockets.
Limited assurance under ASSA 5010 does not recalculate your inventory. It samples it. Someone picks one concrete line out of your ledger and asks where the number came from. "A national average table" is a weak answer. "The Hobart column of this registered declaration, valid over our reporting period, matched to this docket from this plant on this date" is not.
What it actually takes
Capture the docket, not the invoice. The invoice gives you dollars and a supplier name. The docket gives you grade, plant, volume, batch and date, which is everything the factor decision depends on. If your process starts at the accounts payable feed, you have already thrown away the data.
Build a factor library keyed to grade and plant, not to the word "concrete". Hold the declaration alongside the factor, with its registration number and validity dates, so the provenance travels with the number instead of living in someone's inbox.
Then push it upstream. Ask your top three concrete suppliers for current declarations covering the specific plants serving your sites, and write the requirement into the subcontract. Suppliers are generally willing. The declarations already exist. Nobody has asked for them in a structured way.
This is the workflow Carbonly is built around. The document engine reads batching-plant dockets across the formats they actually arrive in, and five-tier material matching maps a raw text like "N32-20-T-R53" to a specific factor rather than a generic one. The material library holds EPD-sourced factors with provenance flagged, the AR5 and AR6 toggle sits at report render time so NGER and AASB S2 can draw from one ledger, and every emission record keeps its source document, so the sampling question has a one-click answer.
Some of this we have not solved. Matching gets harder as the mix code gets more bespoke, and special-class designs still need a human to confirm the factor. Where no plant-specific declaration exists, you are on a regional average and your uncertainty is real. Module A4, the delivery leg from batching plant to site, is rarely documented at docket level and is usually an estimate. And the co-product allocation debate around SCMs is unsettled enough that we would not want to promise today's numbers survive the next PCR revision unchanged.
Concrete is still worth doing first. It is the biggest number, the data exists, and the suppliers have already published it.
Pull one month of concrete dockets and check whether your system recorded the batching plant. If it didn't, that is the fix, and it costs nothing but a process change.
Related Reading:
- Tier 1 Construction Carbon: 50+ Material Types
- Embodied Carbon in Buildings: What Australian Builders Need
- Spend-Based Emissions Accounting for Australian Companies
- Unit Conversion in Carbon Accounting: The 1,000x Error
- ASRS Assurance: What Your Auditor Will Actually Ask For
- Scope 3 Double Counting: The Same Tonne of Concrete, Counted Twice - Getting the factor right still leaves the boundary question: whether that tonne belongs in your ledger once or twice