EU Carbon Border Adjustment Mechanism

Your EU customers pay for your emissions. Actual data usually costs them less.

From 2026 every tonne of fertilizer, steel, aluminum or cement entering the EU carries a carbon price. How much depends on your numbers: measured, or the EU's conservative defaults.

Why now

Free allocation falls every year from 2026

Certificates cover what free allocation does not. Each year less of the benchmark is free, so each tonne of embedded CO₂ costs more.

  • 1 January 2026Definitive period starts. Importers above 50 t a year need authorized-declarant status.
  • February 2027Certificate sales open for 2026 imports, priced at the quarterly average of EU ETS auctions.
  • 30 September 2027First annual CBAM declaration for 2026 imports, with verified emissions where actual values are used.
  • 2034Free allocation reaches zero.

Dates per Reg. (EU) 2023/956 as amended by Reg. (EU) 2025/2083.

97.5%202695%202790%202877.5%202951.5%203039%203126.5%203214%20330%2034
CBAM factor: share of the benchmark still covered by free allocation · Directive 2003/87/EC Art. 10a(1a)
Worked example

What actual data is worth: one tonne of urea, 2026

Data used in the calculation

Basis: 1 t urea · import year 2026 · illustrative plant

SEE, actual (direct + indirect)
1.117 t CO₂/t
Ammonia in the urea
0.564 t NH₃/t
Benchmarks, column A: urea · ammonia
0.053 · 1.522 t CO₂/t
EU default (Egypt) × 1% mark-up
1.39 × 1.01 = 1.404
Benchmark, column B · CBAM factor 2026
0.902 · 97.5%

Working

  1. Free allocation, actual = (0.053 + 0.564 × 1.522) × 97.5% = 0.889
  2. Certificates, actual = 1.117 − 0.889 = 0.228 t CO₂/t
  3. Free allocation, default = 0.902 × 97.5% = 0.879
  4. Certificates, default = 1.404 − 0.879 = 0.524 t CO₂/t

Difference: 0.296 certificates per tonne of urea

With actual data
0.228
With EU defaults
0.524

certificates per tonne of urea (t CO₂/t)

57% fewer certificates · about €22 per tonne of urea at €75/t CO₂

Benchmarks IR 2025/2620 · CBAM factor Dir. 2003/87/EC Art. 10a(1a) · CSCF = 1 for 2026–2030 (Decision 2026/1862) · Egypt default IR 2025/2621 as corrected by IR 2026/1740. Plant figures illustrative.

How we do it

Seven steps, each signed off before the next

StepWhat happens
1 · ScopeGoods, CN codes, the data request
2 · BoundaryInstallation and production processes
3 · Process modelMass, carbon and heat balances traced to the design documents
4 · MeteringEvery input mapped to a meter, or listed as a gap
5 · AttributionHeat, power and precursors per the annexes
6 · CalculatorOne workbook in; embedded emissions and checks out
7 · ReportingMonitoring plan, operator's report content, verifier pack

Most of the work is steps 2 to 5. Get the boundary and the allocation right and the calculator is the easy part.

What you receive

What lands on your desk at each gate

Typical scope: one installation, one reporting period, up to four product groups, about ten weeks in all. The technical lead runs every phase, with an associate engineer from your sector where it helps.

PhaseEffortWe need from youYou receive
0 · Applicability and reporting route3–5 daysCN codes and tonnes shipped to the EU, a block diagram, the data you already haveA short memo: goods in scope, actual against default values with the certificate cost both ways, the recommended route, what your EU importer needs, and a fixed quote for Phase 1
1 · Data and monitoring plan8–12 daysDesign documents, meter list, one year of production, fuel and power figures, a contact at the plant for data questionsData request, installation boundary, meter map and a monitoring plan with every open item marked
2 · Calculation6–10 daysAnswers to the open items, lab results and meter readings for the periodResults workbook: specific embedded emissions per CN code, direct and indirect, monthly and annual, with balances that close
3 · Pre-verification4–6 daysAccess to the people the verifier will talk toMock audit findings, operator's report content with an uncertainty estimate, and an evidence pack where every number points to its source

Samples on an illustrative plant can be shown on the first call. Client files are never used as samples.

Where numbers go wrong

Five allocation traps in fertilizer plants

  1. Shared steam network. Charged by the heat each process uses, at the network's own factor.
  2. Steam leaving a process. At the fuel mix of the process that made it; 62.3 t CO₂/TJ if that cannot be shown.
  3. Reaction heat. Steam from the ammonia synthesis loop counts as zero, if it is metered on its own.
  4. Own generator. The grid default factor applies unless there is hourly evidence.
  5. Precursors. Ammonia takes its full embedded emissions into the urea or nitrate made from it.
Illustrative dataSteam and power view of an ammonia-urea site: turbines, headers and power split, each with its CO2 share
Steam and power view in our CBAM tool: every turbine, header and power split carries its CO₂ share.
Illustrative dataResults screen of the CBAM tool with embedded emissions per product against the EU default
Results screen: intensity per CN code against the EU default, and where each tonne comes from.
The tool

One workbook in, a verifier pack out

  • Every row checked before anything is calculated; every run saved
  • Placeholders stay visible on every screen until measured data replaces them
  • Scenarios: carbon price, tonnage, import year, method choices
  • Results workbook, boundary page, monitoring plan and operator's report from the same run
  • Runs offline on your own computer: plant data never leaves the site
Quality assurance

Checked before the verifier sees it, including the misses

TestResultWhat it means
Legal text72 / 72Rule statements in the method checked against the official text; 8 deliberately false statements put in as controls, all rejected
Commission examples28 / 28Checks passed against the Commission's printed results: Guidance 5A cement (12), 5D blast furnace steel (9), 5D electric arc furnace steel (7). Four printing slips in 5D turned up on the way
Recalculation0.0025%Largest difference between an independent recalculation and the tool, over 22 items
Uncertainty± 2.2%On urea embedded emissions at the recommended metering; Monte Carlo, 20,000 runs. With weak metering, closer to ± 5%
Mock verification7 of 14Planted errors caught by the first version of the checks. The seven misses (drifting meters, swapped months, kWh for MWh) went straight onto the fix list

Test cases: a synthetic ammonia-urea plant, synthetic cement and steel works, and the Commission's worked examples. Results on a real plant remain subject to its verifier.

Start with Phase 0: 3–5 days, fixed price.

You get the reporting route that costs your EU customers least, and a list of what it takes to get there.

Website software developed by HA Egypt · © 2026