Calculating CBAM embedded emissions: direct, indirect and precursor

A CBAM declaration turns on one number: embedded emissions per tonne of product. That number is the sum of three components, each with its own measurement rule, data source and verification difficulty. Most companies that get the calculation wrong do not get the formula wrong — they draw the boundary wrong.

Component 1 — direct emissions

Direct emissions are combustion and process emissions occurring inside the installation’s own boundary. Process CO₂ from clinker calcination in cement, coke and coal combustion in steel, fuel consumption in ammonia synthesis for fertilisers all sit here. There are two sub-items: emissions from fuel combustion, and emissions from the chemical transformation in the process itself.

Process emissions are what people coming from LCA most often miss. In clinker production the CaCO₃ → CaO + CO₂ reaction releases roughly 0.52 tonnes of CO₂ per tonne of clinker even if no fuel is burned at all. That emission does not fall with fuel efficiency; it only falls by reducing the clinker factor or using alternative binders.

Component 2 — indirect emissions (electricity)

Emissions from generating the electricity the installation buys are indirect emissions. CBAM differentiates by sector here: for cement, fertilisers and electricity, indirect emissions are in scope from the start; for iron, steel and aluminium, inclusion is conditional. The distinction is not arbitrary — it mirrors indirect cost compensation mechanisms inside the ETS.

The choice of electricity emission factor is decisive. Grid electricity uses the country average grid factor; where there is a power purchase agreement with a specific generator and a demonstrable physical link, that source’s factor may be used. Turkey’s grid factor sits around 0.40–0.45 kg CO₂e/kWh depending on year and generation mix; for an electricity-intensive EAF steel plant that is the dominant line in the footprint.

Component 3 — precursor emissions

A precursor is an input to your product that is itself a CBAM good. For a plant producing hot-rolled coil, slab is a precursor; for a cement plant, purchased clinker is a precursor. The embedded emissions of those inputs carry through into your product. The chain extends upstream, and your supplier’s data quality becomes your declaration’s data quality.

This is where the process most often jams. You can measure your own installation; you cannot make your supplier measure theirs. Contracting for precursor data, standardising the measurement methodology and screening alternative suppliers on data maturity should therefore be among the earliest steps in CBAM preparation.

Installation boundary and allocation

In a multi-product installation, how total emissions are split across products is a separate problem. CBAM solves it by mapping production processes to goods: a distinct production process is defined for each CBAM good, and energy and material flows attributable to that process are assigned to that good. Shared utilities — steam boilers, air separation, wastewater — are allocated on a measurable physical key, typically energy or mass flow.

This is structurally the same problem as allocation in LCA, but solved with a narrower rule set. In LCA you make a justified choice between physical, economic and system expansion approaches; in CBAM you are expected to follow the prescribed rules. When feeding both from one inventory, documenting that difference is the question verifiers ask most often.