U.S. Steel
Cutting energy costs by 5% across steel annealing operations with predictive AI
- ~5%
- Energy & emissions cost reduction
- <1 yr
- ROI on project investment
- $20B+
- Revenue company transformed
Sudolabs expertise
- Industrial process optimization
- Multi-constraint logistics algorithms
- Predictive modeling for manufacturing
- Large-scale industrial data analysis
- Integration of AI with existing industrial systems
- Customized solution development for industry-specific challenges
A $20B+ steel giant betting on AI to reshape its furnace floor
U.S. Steel is one of the world’s largest integrated steel producers. Its European subsidiary runs large-scale annealing furnaces — the energy-intensive heat treatment process that determines the final mechanical properties of steel coils.
With rising energy costs, tightening emissions regulations, and growing ESG pressure, the subsidiary needed a fundamentally different approach to controlling the time-temperature curves that govern every batch.
Two years of furnace data. Zero predictive intelligence.
The steel industry faces dual pressures of environmental concerns and rising energy costs. Our client’s complex logistical planning was entirely human-operated, leading to inefficiencies in resource management and energy usage.
Operators relied on conservative static profiles — proven safe, but far from optimal across thousands of batch configurations. The subsidiary had accumulated two years of detailed operational data but lacked the analytical infrastructure to turn it into actionable process changes.
- Conservative static furnace profiles wasting energy on every batch
- No predictive capability to tailor time-temperature curves to batch-specific parameters
- Two years of rich operational data sitting unused in legacy systems
- Regulatory and ESG pressure demanding measurable emissions reductions
Custom predictive models turning raw furnace data into optimised production plans
Sudolabs conducted a comprehensive analysis of the client’s processes and historical data. We developed two core AI systems.
The first — an ML-driven storage and production planning engine — automates steel coil grouping and annealing batch planning. The system considers multiple steps ahead and adheres to numerous constraints.
The second generates insights and recommendations for optimising temperature curves, trained on two years of historical data including weight, height, and furnace filling parameters.
- AI-powered batch planning — ML-driven optimisation of steel coil combinations and annealing schedules
- Predictive temperature curve modelling — batch-specific recommendations based on operational parameters
- Six-week POC proving model accuracy before committing to production build
5% energy cost reduction. ROI in under a year.
The deployed system delivered approximately 5% reduction in energy and emissions costs — translating to millions in annual savings at the scale of a $20B+ revenue manufacturer. The project paid for itself in less than a year.
For U.S. Steel’s European subsidiary, this was proof that AI-driven process control could be deployed in heavy manufacturing quickly, safely, and with measurable returns. This success demonstrates the transformative potential of AI in heavy industry, paving the way for more sustainable and cost-effective steel manufacturing.
- ~5% reduction in energy and emissions costs
- Full ROI on project investment in less than 12 months
- POC to production in under 5 months total
Tech stack