The electric arc furnace panel already talks — without touching its PLC
The electric arc furnace is the jewel of the melt shop, and its SCADA panel is ten or twenty years old, isolated by design, with data that dies on screen every minute. With an iLEAN Connect industrial camera the furnace stays the furnace, but the plant now has every heat curve in central memory. Digital TPM with no retrofit.
The asset that sets the OEE of the whole melt shop is also the most isolated in the plant.
The EAF decides the pace, the energy cost and the quality of everything that comes after — billet, rebar, wire rod. Its SCADA panel displays in real time the data that explains every heat, but that data lives and dies on the glass of the screen, refreshing every 30-60 seconds without leaving a record.
- Proprietary SCADA isolated by design — active power, power factor, temperature, specific consumption and the refining curve are shown on screen and disappear every 30-60 seconds. When the heat is tapped and the next basket comes in, the previous curve no longer exists anywhere.
- Replacing the EAF SCADA is not realistic in the short term — it means more than half a million euros per furnace and weeks of melt shop shutdown. As long as the furnace keeps melting within parameters, that capex does not get signed.
- A direct consequence for quality and for TPM — nobody correlates the heat curve with scrap or with a customer claim, because the data does not exist outside the panel. And furnace TPM is executed with handwritten notes per shift.
Investigating a heat with high scrap, or a claim about a shipment already dispatched, becomes manual archaeology with a high risk of never reconstructing the root cause — because the only witness to what happened during refining is a screen nobody photographs or records in structured form.
Connect in photo-over-panel mode — the camera reads the curve, the furnace stays the furnace.
The EAF needs no new PLC, no communications gateway the manufacturer no longer supports, and no intervention on its original electronics. It needs the curve that is already displayed on screen — power, power factor, temperature, specific consumption, refining — to be recorded per heat without depending on somebody writing it down by hand. That is what Connect does in photo-over-panel mode.
A fixed industrial camera watches the EAF panel without touching the original SCADA network, with interlocking and verified calibration. Connect reads the screen through OCR with an anchored LLM, structures each heat's time series and cross-references it with the heat order and the lab certificate. Root cause analysis per heat goes from weeks to minutes. The SCADA stays as it is, but now the melt shop can see the curve.
How Connect works on the EAF panel:
- Fixed industrial camera on an external bracket — installed with interlocking and calibration, pointing at the panel from outside, without touching the furnace network or wiring. It captures the screen frame by frame throughout the heat, from basket charge to tapping.
- OCR with an anchored LLM interprets every frame — each capture is read by OCR specialized in industrial panels, anchored to that specific SCADA layout, to extract active power, power factor, temperature, specific consumption and the refining curve without misinterpretation.
- It structures the time series per heat and ties it to the order — every reading is ordered into a time series per heat and automatically associated with the heat order and the lab certificate for that same heat.
- Dashboards per heat and per electrode — the melt shop lead and the maintenance lead see each heat's curve and the consumption drift per electrode, so the SCADA renewal can be deferred indefinitely without going blind to the data.
Isolated EAF vs. EAF with Connect on the panel
| Aspect | Before | With iLEAN Connect |
|---|---|---|
| Heat curve (power, power factor, temperature, specific consumption, refining) | Visible only on screen, gone every 30-60 seconds | Captured by camera and structured as a per-heat time series in central memory |
| Curve ↔ scrap and customer claim correlation | Zero — the data does not exist outside the panel | Root cause analysis per heat in minutes, cross-referenced with heat order and lab certificate |
| Furnace TPM (transformer, electrodes) | Handwritten notes per shift | Dashboards per heat and per electrode — the basis for predictive TPM |
| Intervention on the original EAF SCADA | Any connection means touching proprietary electronics | None — external camera, no network, wiring or control logic touched |
| SCADA renewal (>€500K per furnace, weeks of shutdown) | Pending, undated, on the committee table | Deferrable indefinitely without going blind to the data |
| Specific consumption (kWh/t) for ISO 50001 and ResponsibleSteel | No structured record — estimates and notes | Recorded per heat, traceable and auditable |
Impact estimate for your plant — to be validated with your own numbers.
The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.
- Long steel melt shop with an EAF and a ten or twenty year old SCADA panel, isolated by the manufacturer's design, with furnace TPM executed on handwritten notes.
- Connect photo-over-panel pilot on the EAF — only an external camera with interlocking and calibration, without touching the original SCADA or the furnace electronics.
- Indicative payback of 5 to 10 months, depending on the frequency of heats with unexplainable scrap and the cost of each claim investigation in your plant.
- Avoided SCADA renewal capex of more than €500,000 per furnace, plus the enabling of predictive TPM for the transformer and the electrodes, and the monetization of specific consumption for ISO 50001 and ResponsibleSteel. Estimate to be validated against your data.
And the fair question from the melt shop lead
"What if the camera misreads a digit on the curve of a 15-year-old SCADA and gives me a false figure in a scrap investigation?" — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely recontextualizes a specific figure from one medium to another — reading a fixed field on the panel curve and moving it into a structured series — the best models brought the error below 1.5% [1]. And even then, the critical call is not made alone: Connect discards the reading when it does not reach the confidence threshold, and a person reviews the history before a correlation counts. The three safety rings exist precisely for this.
[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.
What people ask about digitizing an EAF panel without touching the SCADA
Does the camera touch the electric arc furnace SCADA network?
No. The industrial camera is an independent device installed on an external bracket pointing at the EAF panel from outside. It does not connect to the proprietary SCADA network, does not share wiring with the furnace and does not write to any point of its control logic. It simply photographs the screen the pulpit operator already watches on every heat, exactly as a person writing the curve down by hand would. The furnace remains exactly the same machine, with the same melting and refining profiles and the same original electronics, without a single change to its configuration.
How does the camera survive the EAF environment — heat, dust, vibration?
The camera is not installed next to the furnace shell but pointing at the SCADA panel inside the tapping pulpit, a space already conditioned for people and electronics to work in. Even so, the device sits in a sealed industrial housing against dust and vibration, and calibration is verified continuously: if a frame arrives degraded — dust on the optics, glare, anomalous lighting — Connect discards it below the confidence threshold instead of forcing a value. Maintenance comes down to periodic cleaning of the optics, folded into the area's TPM routine as one more point on the autonomous schedule.
What happens if the SCADA panel layout changes?
It is rare on a ten or twenty year old SCADA with no firmware changes, but if the operator navigates to another view, the language changes or the screen is updated, Connect detects it as a layout deviation and stops structuring that reading automatically until the model is re-anchored to the new field arrangement — a documented configuration adjustment, not a reinstallation and not an intervention on the furnace. In the meantime it invents no data: the capture is flagged for review and the previous curve history stays intact and traceable per heat and per sequence.
How is the curve-to-scrap correlation used in practice?
Faced with a heat with high scrap or a customer claim about a rebar or wire rod shipment, the quality team opens the per-heat history and sees the full curve — active power, power factor, temperature, specific consumption and refining curve — cross-referenced with the heat order and the lab certificate. Instead of reconstructing by hand what happened weeks ago from shift notes, they filter the affected heat and check in minutes which stretch of the curve deviated from the conforming heats — for example, a refining stage that ran long or a tapping temperature outside the window. That root cause analysis, which does not exist today because the data dies on screen every 30-60 seconds, is the basis for correcting the process before the next sequence.
What does Connect contribute to ISO 50001 and ResponsibleSteel?
The EAF is the melt shop's largest electricity consumer, and specific consumption (kWh/t) per heat is the indicator both frameworks require evidence for. With Connect that figure stops dying on the screen: it is recorded per heat, traceable and exportable, so the ISO 50001 energy performance indicators are fed by real series rather than estimates, and the ResponsibleSteel audit finds continuous evidence instead of handwritten notes. The history also makes it possible to detect consumption drift per tonne pointing at electrode or transformer wear — the starting point for predictive furnace TPM. Impact estimate to be validated with your plant's data.
More cases from this series
- Every heat ships with its audit dossier already madeEvery heat ships with its auditable dossier: charge mix, chemistry, rolling record and certificates,…
- The coil marker stitched to the active order, with no retypingHeat number, grade and customer travel from the active order to the coil marker with no retyping: the…
- The customer alert, read within seconds instead of hoursOrder changes, quality alerts and scrap supply delays are read, classified and cross-checked against the…
- Heat startup in the melt shop, digitized with a photographThe heat startup sheet is photographed and the heat exists in the central system at zero latency: scrap…
- The lab spreadsheet poured into the ERP, without retypingChemistry and mechanical readings move from the lab spreadsheet into the ERP with no retyping — and…
- Scrap receiving enters the ERP with a photo, not with typed linesA photo of the weighbridge ticket and the load classification puts grade, weight and radiological check…
How we digitize your SCADA without touching it — external camera on the EAF panel, pilot in weeks.
We work on your plant's real data, not ours. Assessment with no commitment.
Request estimated ROI within 48h ‹ See all 12 long steel cases See steel