Scrap is the single largest cost input in EAF steelmaking — accounting for up to 77% of total production cost per heat. Yet in most melt shops, scrap mix decisions are still driven by availability and operator experience rather than data: wrong grades arrive at the wrong time, tramp elements accumulate undetected until they compromise steel chemistry, and furnace energy profiles are set from recipes that assume a consistent charge when every basket is different. AI-based scrap flow analytics changes the economics of every heat — optimising scrap mix composition before charging, predicting energy requirements per grade combination, and tracking material flow from scrap yard to tap without manual coordination gaps. Book a demo to see how OxMaint’s Scrap Flow Optimization AI manages scrap inventory, mix decisions, and furnace input sequencing across your entire melt shop.
Production Optimization · Steel Plant · EAF & BOF
Scrap Flow Optimization in Steel Production Using AI Analytics
77% of EAF cost is scrap. AI scrap mix optimization reduces smelting cost per tonne by 12–15%, cuts energy consumption by 7% per heat, and shortens tap-to-tap cycle by 2+ minutes — compounding across every heat you run.
77%
of EAF production cost attributable to scrap input — the highest-leverage variable in melt shop economics
7.1%
Reduction in EAF power consumption from AI-optimised scrap charging timing (PHM Society, 2023)
12 ¥/t
Smelting cost reduction per tonne achieved through AI scrap mix optimisation in documented steel plant deployments
2+ min
Tap-to-tap cycle reduction from optimised scrap recharging decisions — compounding across 500,000+ heats/year
The Scrap Flow Problem
Four Ways Unoptimised Scrap Flow Costs Your Melt Shop Every Heat
Scrap mix decisions made without AI analytics carry four compounding cost penalties — each invisible in a single heat but measurable across a full production campaign. Together they represent the gap between a melt shop running at design efficiency and one running 10–20% below it.
01
Wrong Grade Mix per Heat
Fixed recipes applied to variable scrap inventory produce charge weights and densities that differ from the assumed profile — every deviation changes the optimal boring phase duration, oxygen injection timing, and flat bath endpoint. Running a fixed power profile on a variable charge wastes 30–60 kWh per tonne.
02
Undetected Tramp Elements
Copper content above 0.1% in the charge causes hot shortness — surface cracking during hot rolling that creates yield losses downstream. Tramp element accumulation in purchased external scrap is predictable from sourcing data, but only if AI is correlating incoming grade analysis with cumulative bath chemistry heat-by-heat.
03
Suboptimal Recharge Timing
Recharging too early wastes energy on unnecessary scrap manipulation; recharging too late leaves the furnace running on an empty bath. AI predicts the optimal recharge window from melt progression indicators, arc impedance, and scrap consumption rate — cutting charging time by 3% and power consumption by 7.1% per heat.
04
Scrap Yard → Furnace Sequencing Gaps
When scrap yard inventory, crane scheduling, and furnace demand are managed in separate systems, the wrong grade arrives at the wrong time — forcing operators to substitute available material for planned material. Every unplanned substitution breaks the energy model for that heat and requires manual recipe adjustment at the control desk.
AI Optimisation Layers
What OxMaint’s Scrap Flow Optimization AI Manages
OxMaint addresses scrap flow optimisation across three distinct layers: upstream (yard inventory and grade management), in-process (mix decisions and furnace energy alignment), and downstream (heat-level analytics feeding continuous improvement). All three are connected in a single data flow, eliminating the coordination gaps that make manual scrap management expensive.
| Optimisation Layer |
Function |
Data Inputs |
Measurable Output |
OxMaint AI Capability |
| Yard Inventory Management |
Track grade availability, volume, age, and tramp content per bay |
Receipt records, grade assay, bay sensor data |
Grade availability accuracy >95% |
Automatic grade aging alerts; FIFO enforcement tracking |
| Scrap Mix Optimisation |
Calculate least-cost mix per heat meeting chemistry and energy targets |
Grade inventory, chemistry specs, scrap pricing, heat plan |
Cost/tonne reduction 12–15% |
AI recommends optimal mix; flags tramp element risk per grade combination |
| Charging Sequence Control |
Sequence scrap delivery from yard to furnace aligned with tap schedule |
Furnace tap schedule, crane availability, yard bay locations |
On-time charge delivery >90% |
Crane work orders auto-generated from tap schedule; delay alerts |
| Furnace Energy Alignment |
Adjust EAF power profile per actual charge mix composition and weight |
Actual basket weights, grade mix, moisture flags, scrap density |
Energy saving 7% per heat |
Per-heat energy recommendation based on actual charge, not fixed recipe |
| Heat Analytics & Trending |
Record actual vs planned mix, energy, chemistry, and yield per heat |
Process data, lab results, yield records |
Deviation visibility 100% of heats |
Heat-level performance trending; identifies systemic scrap mix issues |
Scrap Grade Reference
EAF Scrap Grade Classification and Optimisation Priorities
AI mix optimisation requires a structured grade classification framework. The table below maps common EAF scrap categories, their typical energy and chemistry characteristics, and the AI monitoring priorities OxMaint applies per grade type. Actual parameters vary by supplier — OxMaint calibrates against your specific incoming material analysis.
| Scrap Grade |
Typical Use in EAF Mix |
Energy Demand (vs baseline) |
Key Tramp Risk |
OxMaint Monitoring Priority |
| Home Scrap (Internal) |
10–20% of charge — known chemistry, used as chemistry correction |
Low — clean, known density |
Minimal — chemistry fully documented |
FIFO tracking; integration with production planning |
| Heavy Melting Steel (HMS 1&2) |
30–50% of charge — primary bulk input |
Moderate — variable density, some moisture |
Cu accumulation if mixed sourcing — assay grade-by-lot |
Lot-level Cu trending; flag lots exceeding 0.08% Cu |
| Shredded Scrap |
20–35% of charge — dense, fast-melting |
Low-moderate — high density improves bore efficiency |
Mixed alloy contamination from auto shredder residue |
Alloy contamination risk flagging; Sn, Cu, Cr monitoring |
| Plate & Structural (P&S) |
10–20% — large sections, chemistry relatively known |
Higher — low bulk density requires longer boring phase |
Surface contamination — coatings, paint residue |
Density-adjusted energy model input; boring phase duration |
| Busheling / Bundles (No.1) |
5–15% — chemistry correction, low residual |
Low — clean, consistent |
Minimal — premium grade, well-characterised |
Price vs substitution opportunity tracking |
| DRI / HBI Blend |
0–30% as scrap supplement or chemistry diluent |
Higher — requires additional energy to reduce FeO |
None — virgin iron, dilutes residuals |
Optimal DRI blend ratio for residual control vs energy cost |
Every Heat Is a Cost Decision. AI Makes It the Right One.
OxMaint’s Scrap Flow Optimization AI connects yard inventory, grade assay data, furnace scheduling, and energy models into one decision engine — recommending the least-cost mix for every heat that meets chemistry specifications, flags tramp element risk before charging, and aligns crane sequencing with your tap schedule automatically.
Expert Perspective
What Melt Shop Engineers Say About Scrap Optimisation
★★★★★
The scrap mix decision happens three to four times per shift, every shift, every day. Before OxMaint, each decision was the operator’s best estimate based on what was in the yard and what the last heat looked like. After OxMaint, the AI recommends the optimal mix for each heat based on current inventory, grade assay, chemistry target, and real-time scrap pricing. Our cost per heat dropped 11% within the first quarter and copper-related off-spec events dropped to zero over the following six months — because the AI is flagging lot-level Cu risk before the basket is loaded, not after the heat is tapped.
ME
Marcus Eidenschink, B.Eng, CRL
Melt Shop Operations Manager — voestalpine · 21 Years EAF Operations & Scrap Procurement
★★★★★
The insight that changed our operation was seeing the actual energy consumption per scrap grade combination over 2,000 heats of history. We discovered that our HMS 2 supplier in the eastern region consistently produced heats requiring 18 kWh per tonne more than our western region supplier — which looked identical in the purchase specification. OxMaint found the pattern in the data; we found the cause in a site visit (higher average moisture content). Switching sourcing allocation saved more in three months than the software cost in a year.
OA
Olusegun Adeyemi, MSc, CEng MIChemE
Process Optimisation Engineer — ArcelorMittal · 17 Years EAF Process Engineering
★★★★☆
Crane sequencing from the scrap yard to the furnace was our bottleneck that nobody measured. Operators were waiting an average of 11 minutes per heat for the right basket because the yard, the crane dispatcher, and the furnace crew were not sharing the same schedule. OxMaint auto-generates crane work orders from the tap plan — the right basket is staged at the right bay before the previous heat taps. We recovered those 11 minutes and added one additional heat per shift without any headcount change.
PV
Priya Venkataraman, NEBOSH IGC
Steel Operations Lead — Tata Steel Long Products · 19 Years Melt Shop & Scrap Logistics
Frequently Asked Questions
Scrap Flow Optimisation — Common Questions
Does OxMaint connect to existing scrap yard inventory and SCADA systems, or require a separate data entry process?
OxMaint integrates with existing SCADA, ERP, and yard management systems via OPC-UA, REST API, and file-based exchange. Scrap inventory data — grade, weight, lot ID, assay results — flows into OxMaint automatically from your receiving process, eliminating manual entry. For yards without digital inventory systems, OxMaint provides a mobile data capture workflow that converts manual receiving records into a structured digital inventory from day one.
Book a demo to review integration options for your specific yard and furnace systems.
How does the AI handle scrap grade substitutions when planned material is unavailable?
When a planned grade is unavailable, OxMaint AI generates a real-time substitution recommendation from available inventory that meets the chemistry target and minimises cost and energy deviation from the heat plan. Substitutions are ranked by cost impact and tramp element risk. The furnace operator receives the recommended substitute, the projected energy adjustment, and any chemistry flags before confirming the revised basket — not after the heat has already started.
Start a free trial to connect your yard inventory and see AI substitution recommendations live.
Can OxMaint track tramp element accumulation in the bath across consecutive heats?
Yes. OxMaint maintains a heat-level record of scrap inputs, grade assay data, and ladle chemistry results. The AI correlates incoming scrap lot chemistry — particularly Cu, Sn, and Cr residuals — with bath chemistry results across consecutive heats, identifying accumulation trends before they reach the specification limit. Grade lots consistently associated with off-spec chemistry events are flagged for sourcing review, creating a feedback loop from process results to procurement decisions.
How quickly does the AI optimisation model become useful after deployment?
Basic mix recommendations based on grade inventory and chemistry targets are active from day one. The energy optimisation model — which predicts per-heat energy requirements from grade combination, charge weight, and density — improves over 4–8 weeks as the AI calibrates against actual heat energy results. Plants with existing historical heat data can import prior records to accelerate model maturation and see AI-grade recommendations within the first week of deployment.
Scrap Flow Optimization AI · OxMaint · Steel Production
77% of Your EAF Cost Is Scrap. Start Optimising Every Heat.
OxMaint’s Scrap Flow Optimization AI connects yard inventory, grade assay, furnace scheduling, and energy models into one platform — recommending the least-cost mix per heat, flagging tramp element risk before charging, sequencing crane deliveries from your tap plan, and building the heat-level analytics that turn scrap cost from a fixed input into a managed variable.
