EU ETS in Steelworks: Define Data Roles and Training Needs

Not everyone needs the same ETS course

A plausible emissions report can rest on an implausible handoff: procurement supplies tonnes, the laboratory supplies a carbon content on a different basis, and production reports a stock figure that is later corrected. Reviewing only the annual total may hide the break. For a German steelworks, useful training needs therefore start with the data journey. Who produces each value, who understands its limitations, and who may accept it into reporting after an independent check? A list of courses cannot answer those questions.

This guide develops a practical operating approach for a stationary installation already identified as subject to emissions trading. It does not determine whether a particular installation is covered or which calculation method is permissible. The permit, approved monitoring plan and current requirements remain decisive. The European Commission guidance describes the annual monitoring, reporting and verification cycle. Our example focuses on the people between primary data capture and technical approval, not allowance trading, free allocation or a complete course in emissions trading law.

1. Establish the legal basis and system boundary

For installation operators, section 21 TEHG requires emissions to be determined under the approved monitoring plan and reported by the end of 31 March of the following year. Section 5 requires the information to be verified by a verification body. This does not create a blanket requirement for a particular annual online course for every employee. The relevant competence needs arise from actual duties. Someone matching sample identifiers needs different practice from someone evaluating corrections or taking responsibility for the completeness of the report.

First map the monitored installation using its actual source streams and emission sources. The factory gate is not automatically the correct accounting boundary. Section 21 contains specific provisions for certain iron and steel activities treated as a single installation and for associated power plants. Have the technical owner confirm the allocation before teams collect data. At every handoff, record the installation, reporting period and monitoring-plan version to which the information relates. This prevents later disputes about figures that only appear to contradict one another.

Name neighbouring processes explicitly as separate workstreams. A CBAM task matrix for steel imports may involve similar departments but serves a different regulatory purpose. Likewise, a template for EU ETS and MRV in shipping does not replace a stationary installation's monitoring plan. Borrow organisational ideas, such as a named recipient and a documented query, only where appropriate. Do not transfer calculation logic, deadlines or evidence categories into the steelworks without checking. Shared terminology is not evidence that the requirements are identical.

2. Give every data journey a clear owner

Article 58 of the Monitoring Regulation requires documented procedures for data flows. Article 59 adds a control system, including competence management and segregation of duties. Article 62 is particularly practical: responsibilities must be assigned to separate conflicting duties, and necessary competencies include training and performance reviews. In the absence of other control activities, the provision requires risk-appropriate confirmation by at least one person not involved in determining and recording the information. A learning-system tick is therefore neither an effective data control nor technical approval.

Start a pilot with five working roles rather than five new positions: data capture, methodology ownership, independent checking, technical system support and reporting coordination. Assign existing people and deputies. Capture explains origin and units; the methodology owner assesses compatibility with the plan; the checker reviews independently; system support protects processing and recovery; coordination tracks unresolved items through submission. This is a design aid, not a legally prescribed organisation chart. Small teams still need a concrete control arrangement that appropriately separates conflicting responsibilities.

A useful handoff card identifies the data object, period, unit, measurement basis, origin, plan reference, responsible person, control step and storage location. Add a status such as open, under review or released. This is a recommended local template, not an official form. A laboratory value should, for example, identify its sample and batch and show whether it has subsequently been corrected. A quantity record should say whether it represents delivery, consumption or stock movement. These distinctions belong in the first practical exercise.

Test deputies beyond the organisation chart. A deputy must find the correct folder, recognise the current procedure version and return an unresolved data set to the appropriate owner. A deputy matrix for key personnel helps distinguish knowledge, access rights and approval authority. A shared user account is not a substitute. Ask the deputy to complete a prepared case while the usual owner only observes. Missing permissions or unfamiliar escalation routes then become a specific improvement task rather than a surprise just before the reporting deadline.

AI-generated

3. Identify learning needs through observable mistakes

Use three short diagnostic tasks: trace a raw value back to its source, reject a deliberately flawed handoff and identify the correct owner for a methodology question. Assess observed behaviour rather than confidence. Someone confusing units needs practice with the data object; someone spotting a discrepancy but unable to contact anyone needs a clarified reporting route. If the required control cannot be performed in the system, more training will not solve the problem. Process owners must first correct the workflow or permissions.

Next assign small learning units by role. Data collectors practise identifiers, periods and status; checkers practise completeness, plausibility and justified rejection. Methodology owners address changes and data gaps using the approved procedure. IT and administration review access, backups and recovery within their assigned responsibilities. Documented training records connect the person, role, content version and exercise outcome. Add an outstanding corrective task when someone cannot yet perform reliably. A passed knowledge quiz must not conceal an observed error in applying the process at work.

Trigger refresher work through concrete events: a role change, altered measurement, new software, new source streams or an adverse review finding. A company can also establish regular internal refreshers, but should label that frequency as its own organisational decision. Separate ETS data competence from workplace safety instruction. Someone who safely collects a sample does not necessarily understand its emissions allocation; someone who understands reporting is not automatically instructed for work on equipment. The learning paths need different owners, practical contexts and permissions even where people hold both responsibilities.

4. Three cases for the next team exercise

Case one: procurement and production report different monthly quantities. Give the team a delivery shortly before month-end, later consumption and a stock correction. The correct response is not to choose the more convenient figure. Participants identify the period and meaning of each record, check the required allocation and document the difference. Article 63 includes comparisons between purchasing data, stock movements and consumption where applicable. The exercise outcome is a traceable reconciliation with any outstanding question, not an invented calculation detached from the monitoring plan.

Case two: a measuring instrument fails and leaves a gap in the file. The collector flags the affected period and preserves available evidence rather than silently entering zero. The methodology owner checks the required approach. Article 66 requires suitable conservative surrogate data for missing installation data; if no written estimation method exists, the procedure and necessary monitoring-plan modification must be addressed. Train recognition, stopping and escalation here, not improvised estimation formulas. Approval should show who performed the technical assessment and the basis for that decision.

Case three: an external provider supplies a corrected analysis after the initial version has already been processed. The team must connect the original and revised versions, identify affected calculations and trigger a fresh review. Article 65 retains operator quality-control duties for outsourced data processes. A structured supplier-data request and verification workflow offers organisational ideas, but does not replace ETS requirements. Similarly, steel origin evidence does not automatically establish the required emissions parameters. Practise asking for the correct evidence rather than accepting whichever certificate happens to be available.

AI-generated

5. Keep changes and evidence under control

Create a clear change pathway: report the technical change, assess its effect on data and the monitoring plan, clarify necessary steps with the responsible function and brief affected people before use. Articles 14 and 15 address monitoring-plan modifications and their notification or approval. Section 6 TEHG adds the German framework. Not every new spreadsheet view automatically constitutes a significant plan modification, but a new source stream must not disappear into a training-only discussion. The decision belongs with the technical owner and must remain retrievable with its rationale.

Keep evidence where the corresponding work can be examined. A learning record explains what a person practised; a control record shows what they checked in a particular data set. The two may be linked but must not be confused. Article 67 requires at least ten years of retention for the relevant data and information it covers. Do not automatically transfer that period to every personnel record or learning account. Define purpose, access and retention by document class and coordinate personal training records with the responsible functions.

A small pilot produces a more defensible training plan

Start with one selected source stream and one routine handoff plus one error case. Agree a short internal pilot period, identify the collector, independent reviewer and deputy, and ask everyone to work from the same evidence. Then check whether origin, unit, status, correction and approval can be fully reconstructed. Only after that should the approach be extended to other data journeys. This sequence avoids rolling out a comprehensive course when the real bottleneck is an unclear owner or an unsuitable filing system.

Do not evaluate the pilot solely through completed learning modules. Count unresolved handoffs, missing source references and corrections without renewed review, for example. These internal indicators are not statutory thresholds, but they make improvements visible. Discuss each error with its owner: was knowledge, time, access or a clear instruction missing? Only the first is directly a training problem. Record the agreed action, responsible person and internal due date, then check the same process again.

ConformBase can be considered as a framework for role-based learning content, not a replacement for the approved monitoring plan, a verification body or the company's methodology owner. Its current steel-package offer still describes final content review and expressions of interest rather than bookings. Confirm availability and the specific scope before planning a rollout. The most important next step does not depend on the tool: ask a responsible person to explain the journey of one real data object and have the deputy reconstruct the same process. That is where a verifiable competence plan begins.