A 120-ton capacity mini mill operating four electric arc furnaces was constrained by tap-to-tap cycle time averaging 56 minutes per heat — 18% above industry benchmark. Electrode consumption varied wildly (1.8–2.4 kg/ton across heats), oxygen injection lances clogged frequently, and carbon injection timing was inconsistent. The mill was producing 180,000 tons annually at 7 heats per furnace per day. After deploying Oxmaint's AI-powered oxygen-carbon optimization engine integrated with electrode tracking, the plant reduced tap-to-tap time to 42 minutes within 120 days, producing 1,820 additional heats annually, eliminating lance clogs through predictive service intervals, and reducing electrode consumption to 1.5–1.7 kg/ton. This freed up 8 additional production slots per week and delivered $2.4M in incremental revenue. Start free — optimize your EAF cycle time today.
Mini Mill Cuts EAF Tap-to-Tap Time 25% in 120 Days — From 56 to 42 Minutes
Case study: 120-ton mini mill deploys Oxmaint AI-powered oxygen-carbon injection optimization across four furnaces. Cycle time reduction from 56 to 42 minutes, 1,820 additional annual heats, $2.4M incremental revenue.
The Challenge — Inconsistent Melting & Equipment Degradation
The mini mill's four-furnace operation was constrained by inconsistent tap-to-tap times that varied 8–14 minutes between heats. The root causes were three: (1) oxygen and carbon injection timing was manual — operators adjusted based on visual assessment of foam height and slag color, leading to inconsistent melting rates; (2) oxygen injection lance tips eroded from thermal cycling and splash, degrading jet coherence and driving electrode consumption upward; (3) carbon injection lances clogged from moisture back-splash, requiring emergency cleanout that added 4–6 minutes to cycle time. Because electrode consumption was high and highly variable, the mill couldn't optimize batch schedules — they had to buffer 1-2 additional heats per week as insurance against out-of-spec charge weights. The furnace reliability manager described the operation as reactive maintenance in a capacity-constrained business — we can't optimize what we can't predict.
AI-Powered Oxygen-Carbon Injection System — Real-Time Foamy Slag Optimization
Oxmaint's EAF optimization engine integrates real-time burner oxygen flow, carbon injection lance timing, and electrode position data to dynamically adjust melting conditions for maximum efficiency. The system operates on a core principle: foamy slag practice — maintaining a controlled slag foam layer on the liquid surface that dissipates electrical arc radiation uniformly, reduces refractory wear, and accelerates melt completion. In manual operation, operators rely on visual assessment from observation ports — inherently laggy and inconsistent. Oxmaint's system uses four sensor inputs: (1) electrical energy input tracking (kWh per minute), (2) electrode slip velocity trending, (3) burner oxygen flow rate, and (4) carbon injection lance back-pressure as a proxy for lance cleanliness. A machine learning model trained on 6+ months of furnace-specific operational data predicts optimal oxygen injection rate and carbon timing ±3 minutes ahead, allowing operators to anticipate transitions from charging phase to melting phase to refining phase.
Electrode Consumption Optimization — 16% Reduction Through Real-Time Control
Electrode consumption in electric arc furnaces is driven by two mechanisms: oxidation at the tip (unavoidable, ~1.0 kg/ton) and side-surface erosion from flame impingement and electrical resistance heating (highly controllable, typically 0.6–1.0 kg/ton). The mill's baseline consumption of 1.8–2.4 kg/ton indicated poor electrode practice — side-surface erosion was running 20–50% above industry benchmarks. Oxmaint's system addresses this through three mechanisms. First, constant electrode voltage control — the system maintains arc voltage within ±2V of target, preventing the wild swings (±10–15V) that occur in manual operation and cause electrical resistance heating spikes. Second, slag chemistry management — lime-to-silica ratio is monitored continuously and adjusted via additive timing to maintain a slag melting point within 1,400–1,450°C, preventing slag that's either too fluid (excessive slip, side erosion) or too stiff (incomplete fluidity, refractory attack). Third, oxygen and carbon timing precision — over-oxidation causes rapid electrode consumption; under-oxidation extends melt time, also consuming electrodes. Oxmaint's AI finds the exact injection timing within each heat that minimizes total electrode consumption while meeting tap-readiness time targets. Within 120 days, consumption fell from 1.8–2.4 kg/ton (average 2.1) to 1.5–1.7 kg/ton (average 1.6), representing 24% reduction. At 180,000+ tons annual throughput, this saves approximately 80 tons of electrodes annually — a cost savings of $400K at current graphite electrode pricing. Start free — reduce your electrode consumption today.
Oxygen Injection Lance Management — Predictive Clogging Detection
Oxygen injection lances are critical process components with limited service life. They erode from thermal cycling and occasional splash of liquid slag. Once erosion exceeds 15–20% of original orifice diameter, back-pressure rises, jet coherence degradates, and oxygen injection efficiency drops by 30–40%. In the mill's prior operation, lance clogs were discovered reactively — operators noticed furnace taking longer to melt, checked the oxygen flow (down from 95 Nm³/min to 70 Nm³/min), confirmed clogging via visual inspection, and scheduled emergency lance replacement. Each event added 4–6 minutes to tap-to-tap time and required furnace shutdown for 30 minutes. The mill experienced 8–12 clog events per month (2–3 per furnace monthly). Oxmaint's predictive system monitors lance back-pressure trending on every heat. Because lance back-pressure is a proxy for orifice diameter degradation, Oxmaint can predict clogging 4–6 heats in advance. The plant now schedules lance replacement every 120 heats (preventive maintenance interval) rather than waiting for failure. With this schedule, clogs have been virtually eliminated — zero emergency lance replacements in the past 6 months post-deployment. This single improvement recovered 3–4 minutes per week in total furnace downtime and enabled the mill to schedule the additional 1,820 heats annually.
We were running EAFs like electricians, not metallurgists. Oxygen timing, carbon injection, electrode voltage — all manually tweaked based on what operators could see from observation ports. Every operator's judgement was slightly different, which meant tap-to-tap times varied wildly — 48 to 64 minutes depending on who was running the shift. Our electrode consumption was 20–30% above benchmark. Oxmaint changed that completely. Within 120 days, the system optimized oxygen and carbon timing to the heat, predicted lance failures 4 heats in advance, and reduced tap-to-tap time from 56 to 42 minutes consistently. We produce 1,820 additional heats per year now. At $1,200 per ton selling price, that's $2.4M of incremental revenue. More than that — the consistency freed up our planning team to schedule with confidence. No more buffer heats. Every slot is filled. Electrode consumption is now 1.58 kg/ton, down from 2.1. That's $400K saved annually. The system has paid for itself 3 times over in the first year. Every mini mill in North America should have this.
Mini Mill EAF Productivity Maturity — Benchmark Your Operation
Mini mill EAF productivity maturity spans from fully manual operation constrained by operator consistency to AI-optimized systems that produce repeatable cycle times and minimum electrode consumption. The framework below assesses current-state productivity drivers. The mill in this case study moved from Level 2 (manual oxygen-carbon injection, inconsistent tap-to-tap) to Level 4 (AI-optimized cycle management, predictive lance maintenance) within 120 days.
Technical Integration: EAF PLC Data Collection & CMMS Cycle Tracking
Oxmaint's EAF optimization engine ingests real-time data from furnace PLCs at 5-second intervals: electrode position (mm), arc voltage (V), electrical power (kW), oxygen flow rate (Nm³/min), carbon injection back-pressure (bar), and furnace temperature via thermocouples. Data flows through edge computing gateways that calculate derived metrics — electrode slip velocity, power balance, oxygen utilization efficiency — and feed these into machine learning models trained on furnace-specific baseline performance. The models predict optimal oxygen and carbon timing for each heat based on charge composition (if scrap type varies), grade being produced, and current electrode wear state. Recommendations are displayed to operators; in advanced implementations, control signals can be sent directly to proportional flow controllers. All decisions and heat outcomes are logged in the CMMS for traceability and continuous model refinement. Start free — integrate your EAF data into CMMS today.
Frequently Asked Questions — EAF Cycle Optimization & Mini Mill Operations
Deploy EAF Cycle Optimization Across Your Mini Mill — 120 Days to Full Integration
Real-time oxygen-carbon injection control, electrode tracking, lance predictive maintenance, and heat data logging. Free to start, live in 120 days.
