A steel plant reports its emissions several times a year, to several audiences: the regulator, customers, lenders and the group. Every report draws on the same four data sets — production, fuel, electricity and stack monitors — and each data set belongs to a different team. When the numbers disagree, the verifier usually finds it before you do. This playbook is a working checklist of 24 cross-checks that catch those gaps before anything is submitted, with a starting tolerance and the usual cause behind each one. It also shows how iFactory runs the checks every day instead of once a year. To test them against your own data, book a reconciliation review.
Steel Plant Emissions Data Reconciliation Playbook
Twenty-four cross-checks across production, fuel, electricity and stack data, so that every report you submit tells the same story.
- Three independent totals that must agree
- A tolerance and a usual cause for every gap
- Checked every day, not only at year end
A mass balance over the whole site.
Each fuel and material, added up.
Measured stacks plus estimated sources.
The whole playbook in one line: three routes to the same number. If they agree, you can defend it. If they do not, find out why before the verifier does.
Why Steel Emissions Numbers Disagree
Gaps between data sets are normal. Gaps nobody can explain are the problem.
Each number in an emissions report starts life somewhere else: a weighbridge ticket, a lab sheet, a utility bill, a stack analyser. They are recorded by different teams, on different dates, in different units. Reconciliation simply means lining them up and asking whether they describe the same plant. Our support team can map your data sources with you.
Four data sets, four owners
- Production. Planning and the melt shop: tonnes of hot metal, crude steel and coke.
- Fuel and materials. Stores, the weighbridge and the lab: tonnes, moisture and carbon content.
- Electricity. Utilities and finance: bills, incoming meters and sub-meters.
- Stack monitors. The environment team: concentration, flow and availability.
Each team trusts its own figure. Nobody is asked to compare them until a report is due.
Seven usual causes of a gap
- Dates. A meter read on the 28th, set against a calendar month.
- Wet or dry. Tonnes as received, treated as dry tonnes.
- Net or gross. Calorific value on one basis, emission factor on the other.
- Stock. Purchases reported as consumption.
- Factors. A default carbon content where a lab value exists.
- Process gas. The same carbon counted in two places.
- Boundary. Two reports that cover different parts of the site.
Researchers applied five recognised accounting methods to the same plant and got totals from 13.5 to 19.9 million tonnes of CO2 equivalent. Nothing was wrong with the data. The methods draw the boundary in different places. So a difference between two of your reports is fine, as long as you can explain it line by line.
Checks 1–12: Production, Fuel and Materials
Most reporting errors start here, long before any emission factor is applied.
A tonne counted twice, a stock change ignored or a wet tonne treated as dry flows straight into the total. Work through these twelve at every month end, not once a year, while people still remember what happened. To see them run against your own ERP and weighbridge data, book a working session.
If not: look for re-graded heats and month-end cut-off times.
If not: one of them is being estimated from a ratio.
If not: intensity figures will differ by the yield loss.
If not: check slab, billet and coke yards against the last survey.
If not: either the furnace had a bad month or a number is wrong. Find out which.
If not: purchases are standing in for consumption somewhere.
If not: check moisture terms in the contract before blaming the scale.
If not: a few percent of moisture becomes a few percent of error in carbon.
If not: list each default, and why no analysis was used.
If not: natural gas alone can be out by about a tenth.
If not: two meters will disagree while both are right.
If not: the same carbon can appear twice, or not at all.
Process gases carry carbon from one unit to another, and often to an on-site power plant. If that power plant reports separately, the same carbon can land in two reports or in neither. Draw one gas balance for the whole site: what was made, what each unit used, what was flared and what was exported. Accounting standards differ on this point, so write down which rule you followed.
Checks 13–18: Electricity and Scope 2
The bill, the meters and the report should all describe the same megawatt-hours.
Electricity looks simple because there is an invoice. In an integrated plant it is not: power is bought, generated from process gas, sometimes exported, and spread across hundreds of sub-meters. Scope 2 errors usually come from mixing these streams. Ask our energy specialists how your meters would map.
If not: align the read dates first, then look at the meters.
If not: find the unmetered load or the failed sub-meter.
If not: its carbon is already inside your direct emissions.
If not: a quiet netting-off will not survive an audit.
If not: last year's factor is the most common Scope 2 error.
If not: a certificate may have been counted for the wrong period.
54 of 1,600 GWh is 3.4%. Against a 2% band, this check stays open until the missing load is found.
Two Scope 2 totals, not one
The location-based total uses the average factor of the grid you are connected to. The market-based total uses the factors of the contracts and certificates you hold.
Where such contracts exist, the GHG Protocol expects both to be reported. Keep them side by side, and never let a customer report quote one while the annual report quotes the other without saying so.
One Year, Three Totals
This is what a clean three-way match looks like. The value is not in the totals agreeing. It is in knowing why, on the day somebody asks.
Checks 19–24: Stack Monitors and the Three-Way Match
A monitor measures what leaves the stack. It cannot see what never reached one.
Continuous stack monitors give an independent reading, which is exactly why they are useful for reconciliation. They also fail quietly: a drifting analyser, a flow sensor in the wrong place, a week of missing data filled with substitute values. These six checks test the monitors first, then use them. To review your own stack data, book a monitoring review.
If not: a low-availability month needs an explanation in the report.
If not: readings since the lapse may not stand.
If not: flow, not concentration, is usually the weak measurement.
If not: the monitored total will always read low.
If not: go back through checks 1 to 22 for the stream that moved.
If not: something large is missing or doubled.
The third bar is 6.08 Mt measured at stacks plus 0.21 Mt estimated for flares and small sources. Leave the estimate out, and the gap looks like 3.5%.
Check 24: reference ranges
- 2.34 t per tonne of crude steel: blast furnace and converter route
- 1.47 t: direct reduced iron with electric furnace
- 0.69 t: scrap-based electric furnace
- 1.92 t: global average across all routes
worldsteel figures for 2024. They include purchased electricity and some upstream emissions, so a direct-only plant figure will be lower.
Under the EU monitoring rules, emissions measured by stack monitors must be corroborated by a calculation for the same sources. Where the two differ considerably, the measured figure stands if the approved monitoring plan says so, but the verifier will look much harder at production and consumption data. In the United States, the reporting rule for iron and steel accepts stack monitors, a carbon mass balance or a site-specific factor.
Before You Submit: Tolerances, Bridges and Sign-Off
Every gap ends in one of three ways: explained, corrected or escalated.
A checklist only works if somebody decides in advance how big a gap matters, and what happens next. Set the tolerances once, write them down, and treat every breach the same way. Then build a bridge between each pair of reports, so that two different totals are never a surprise. Our reporting team can share a starting set.
Starting points for discussion, not regulatory limits. Agree your own with your verifier.
The same plant, 2.10 t per tonne in one report and 2.38 t in another. Both are right. The bridge is what makes that easy to show.
Three ways to close a gap
- Explained. The gap has a documented reason, such as a boundary or a timing difference. Keep the note with the report.
- Corrected. The source data was wrong. Fix it at the source, and keep a log of what changed and who approved it.
- Escalated. No cause found yet. Give it an owner and a date, and tell your verifier before they find it.
The person who owns the data, the environment lead and someone from finance. Finance matters more each year: carbon numbers now sit beside money in allowances, border charges and customer contracts, and they deserve the same month-end discipline as the accounts.
One set of checks, three views
- Data owners. Their own checks, with yesterday's breaches at the top.
- Environment lead. All 24 checks, open items and the bridge between reports.
- Plant head and finance. The three totals, the intensity and what is still open before submission.
Everyone sees the same figures, traced back to the meter, ticket or lab sheet they came from.
Turnkey AI: Delivered, Connected and Live in 6–12 Weeks
You do not build this. It arrives ready.
iFactory ships as a pre-configured NVIDIA AI server, racked and ready, with the software pre-loaded. Rack it, plug in power and Ethernet, and the AI is live on your network.
Our team handles cabling, network setup, PLC and SCADA integration, operator training and 24×7 remote monitoring. The server sits inside your own network, so production and energy data stay on site. For a scope matched to your plant, request a turnkey quote.
Ship, network and data
Server installed. Production, weighbridge, lab, energy meters, utility bills and stack monitors connected. Boundaries and the list of reports agreed.
Model training and pilot
Last year's data loaded and the 24 checks run against it. Tolerances tuned with your environment and finance teams.
Go-live and training
Daily checks, the month-end pack and the report bridges go live. Data owners trained. 24×7 remote monitoring begins.
Frequently Asked Questions
What is emissions data reconciliation in a steel plant?
It is checking that production, fuel, electricity and stack monitor data all describe the same plant before any report goes out. In practice that means a fixed list of cross-checks, a tolerance for each, and a record of how every gap was explained or corrected.
How close should the three totals be?
A common starting point is 2% for the site total, which is also the materiality level verifiers apply to the largest sites under the EU scheme. Individual stacks and sub-meter balances usually need a wider band. Set your own tolerances with your verifier and tighten them as data quality improves.
Which figure is right when the stack monitor and the calculation disagree?
The one your approved monitoring plan names as the reporting method. The other is evidence. A large gap still needs a cause: most often flow measurement, an unmonitored source, or a stock change missing from the calculation. Do not average the two.
Why does our customer figure differ from our regulator figure?
Because they cover different things. A regulator report usually counts direct emissions inside the site boundary. A customer or benchmark figure often adds purchased electricity and upstream materials. Keep a short bridge that walks from one to the other, so the difference is explained before anyone asks.
How often should we reconcile?
At least at every month end, while people still remember what happened. Checks that depend only on meters and production records can run daily. A gap found within a day costs a phone call. The same gap found at year end can cost weeks.
Do we need stack monitors for CO2?
Not always. Many integrated plants report by carbon mass balance, and the US rule for iron and steel accepts a mass balance or a site-specific factor as well as monitors. Where monitors exist for other reasons, they are a valuable independent check even if they are not your reporting method.
How long does it take to go live?
Six to twelve weeks from delivery. We need a place for the server with power and Ethernet, read access to production, weighbridge, lab, energy and stack data, and time with the people who own each data set. To check your set-up first, contact our team.
Bring Last Year's Numbers. Leave With the Gaps Listed.
In thirty minutes we walk through the 24 checks against the data you already have, and mark which pass, which fail and which cannot be run yet. You keep the list either way.
- 1Your last submitted emissions report
- 2Monthly production and despatch figures
- 3Fuel purchases, stock surveys and lab results
- 4Electricity bills and a list of meters
- 5Stack monitor availability for the year
