Monday, 12 October 2026 |
4–6 p.m. | Registration |
Tuesday, 13 October 2026
|
7 a.m. | Registration and Breakfast |
7:50 a.m. | Opening Remarks |
8 a.m. | Keynote: Deploying Digital at Scale: How ArcelorMittal Flat Europe Drives Safety, Quality and Productivity Across a Multicountry Operation
Peter D'haese, ArcelorMittal
A view on how, in Europe, digital (smart automation, data, Internet of Things, AI...) is supported and managed from transvasl to the shop floor to reach business goals. This presentation will also discuss ArcelorMittal Flat Europe’s position within the ArcelorMittal group, as well as the digital organization in ArcelorMittal across the globe, at business segments and in production sites. It will review the roll-out of digital and digitalization in “multicountry and language and maturity” Europe at both local and transversal businesses from the customer to the shop floor. The discussion will cover the main digital axes in Europe supporting different business goals: IT and OT integration, cooperation and organization in Europe, Reuse-Make-Buy strategy on digital, data lake orientation on production sites, and launch of AI in the steelmaking business. |
8:50 a.m. | AI in Action: Enhancing Cold Rolling Stability for Electrical Steels, Adnan Husakovic
Primetals Technologies Austria GmbH
This presentation examines strip breaks during the cold rolling of electrical steel, utilizing AI to forecast their occurrence. By predicting these events, the technology provides actionable insights that enhance operator efficiency and process stability. |
9:25 a.m. | Building a Sustainable GenAI Foundation: Lessons From ArcelorMittal Calvert's Journey
Piotr Pedziwiatr, ArcelorMittal
ArcelorMittal Calvert is on a pragmatic GenAI journey — not a polished success story, but a work in progress shaped by real constraints, trade-offs and lessons learned. Early digital and AI efforts began without a clean architecture, clear standards or a GenAI road map. Those came later, driven by experience, failures and the growing realization that scaling AI in an industrial environment requires discipline, not just experimentation. This presentation shares how ArcelorMittal Calvert is gradually moving from isolated proofs of concept toward a more resilient GenAI foundation built around AI‑ready infrastructure, scalable data pipelines and governance that enables progress without blocking it. The talk is intentionally honest about challenges, including legacy systems, cybersecurity and data‑protection constraints, and the tension between moving fast and doing things responsibly. It also addresses the human side — building skills, resetting expectations around GenAI, and learning to experiment within clear architectural and governance boundaries — offering a realistic view of what sustainable GenAI adoption actually looks like on the plant floor. |
10 a.m. | Break |
10:15 a.m. | AI-Enabled Automated Analysis of Steel Microstructural Evolution During CLSM Experiments
Yanan Song, ArcelorMittal
Confocal laser scanning microscopy (CLSM) enables real-time observation of steel microstructural transformations during thermal cycles, but manual video analysis is labor-intensive and inefficient. This study introduces an AI-assisted workflow that combines contrastive language–image pretraining (CLIP) embeddings with uniform manifold approximation and projection (UMAP) for dimensionality reduction and clustering-based classification of microstructural phases. Applied to steel samples under controlled heating and cooling conditions, the method successfully tracked phase evolution, including austenite and bainite, by correlating cluster distributions with temperature and time. Compared to traditional manual review, the proposed approach reduces processing time by approximately 70%, completing analysis in under 20 minutes per video. While effective for most phases, distinguishing bainite from martensite remains challenging due to morphological similarities, highlighting the need for dedicated segmentation models trained on metallographic data sets. The workflow incorporates expert validation to ensure accuracy, offering a scalable solution for accelerating microstructural analysis in steel research. |
10:50 a.m. | Leveraging AI to improve Reliability and Optimize the Process at a Coking Facility
Keith Hiles, United States Steel Corporation
This presentation highlights how United States Steel Corporation has applied AI and GenAI to improve operations at its cokemaking facility. AI-driven solutions have enhanced the reliability of critical compressors, reducing downtime and lowering maintenance costs. In the plant’s cryogenic section, advanced algorithms optimize operating parameters to boost efficiency and throughput. Additionally, GenAI tools provide operators with real-time, actionable insights, enabling faster decision-making and improved responsiveness. By integrating these technologies, U. S. Steel has achieved measurable gains in operational stability, productivity and overall performance. The presentation demonstrates how AI and GenAI are driving continuous improvement and digital innovation in traditional steel manufacturing, showcasing the transformative impact of data-driven technologies on industrial operations |
11:25 a.m. | Enhancing Safety in the Metals Industry Through AI
Edgardo Labruna, Janus Automation
This presentation explores the transformative role of AI in improving safety across the metals industry, focusing on the integration of advanced detection and predictive technologies into daily operations. Industrial plants are dynamic, high-risk environments where heavy equipment, elevated temperatures and complex material flows converge — creating conditions where safety must be proactive, not reactive. AI-based personnel and pedestrian detection systems use computer vision and sensor fusion to identify people in restricted or high-risk zones in real time, triggering immediate alerts and automated safety responses. Environmental event detection enables the system to recognize hazards such as smoke, sparks, fire or gas emissions, allowing operators to act before an incident escalates. AI algorithms play a crucial role in cobble detection and prevention, identifying abnormal conditions in rolling or processing lines that may lead to material breakouts or equipment damage. Abnormal operation detection leverages machine learning to continuously monitor equipment behavior, enabling early intervention when deviations occur. In addition, incursion monitoring helps secure critical areas such as crane aisles, furnace zones and transfer paths — automatically detecting unauthorized or unexpected movements. Ladle-handling AI modules improve operational discipline and reduce the risk of severe incidents by ensuring proper transport paths, temperature control and interlock compliance. By combining these technologies under a unified safety strategy, AI enables a layered, intelligent protection system that minimizes human error, increases response speed and enhances operational visibility. This approach is not only a safety improvement but also contributes to production reliability, regulatory compliance and cost reduction. Ultimately, AI is reshaping how the metals industry approaches safety — shifting from passive monitoring to active, intelligent prevention. Real successful application cases are presented. |
Noon | Lunch |
1 p.m. | Producer Panel Discussion: Advanced AI Applications and Process Optimization Nathan Settlemire, Charter Steel
Piotr Pedziwiatr, ArcelorMittal |
3 p.m. | Keynote: From Islands to Ecosystems: Interoperability as the Steel Industry’s Productivity Engine
John Dyck, CESMII - The Smart Manufacturing Institute
As steelmakers invest in new capacity and upgrades, the biggest productivity gains will come from interoperability — not equipment size or speed. This keynote argues that standardization and connected architectures can deliver step-change improvements in yield, throughput, quality, energy performance and traceability. It outlines a practical path from siloed systems to vendor-neutral, plug-and-play operations using common identifiers, shared data semantics and open interfaces. Standardized information models reduce integration time, ease technical debt and enable tighter supplier connectivity — critical in capital-constrained, talent-tight environments. By creating a consistent data foundation, interoperability also makes AI scalable, turning legacy assets and process data into reliable, production-ready insights. The result is a coordinated, learning production system where every reshoring investment delivers amplified value. |
3:50 p.m. | Optimizing Byproduct Gas Networks in Integrated Steel Plants Using an Advanced Energy Management System
Tarun Mathur, ABB
In the steel- and ironmaking processes, integrated steel mills generate large volumes of energy-rich byproduct gases such as blast furnace gas, coke oven gas and oxygen furnace gas. These gases are typically used in cogeneration power plants, for reheating furnaces, boilers and in other process units. However, because generation and consumption rates are highly variable and dependent on production, significant imbalances arise. During surplus, gases may be flared, representing wasted fuel and an impact on the environment. While gas shortages may force the plant to buy supplementary fuel or electricity from the grid. This presentation will describe a digital approach for implementation of an energy management system (EMS) to balance and optimize plantwide byproduct gas networks. The method follows a three-step sequence: (1) real-time monitoring of the gas network parameters, (2) short-term forecasting of gas generation and consumption using statistical, regression, and machine-learning models, and (3) optimization through an economic flow network model solved with a mixed-integer linear programming method. Once these steps are integrated the tool produces an economical gas distribution plan that stays within operating limits, reduces flaring and satisfies internal demand. Two case studies illustrate the approach. First, a European integrated mill that showed measurable reductions in flaring and improved power purchase forecast accuracy. The second, a large Asian mill currently under engineering and design where machine learning forecasting combined with optimization is expected to support robust planning even under variable production schedules. Key lessons from these implementations highlight the importance of reliable data for gas generation and consumption, realistic modeling of power generation constraints, and flexible scheduling logic to adapt to shifting operations. Attendees will gain a practical understanding of how modern EMS platforms and optimization models can support steelmakers in reducing flaring, maximizing internal energy utilization and contributing directly to corporate sustainability goals. The session will also highlight lessons learned, data availability considerations and a road map toward an autonomous utility network balancing. |
4:25 p.m. | Integrated Data-Driven Digital Application Grounded in Comprehensive Process Knowledge and Advanced Automation: A Holistic Approach to Optimizing the Hot Strip Mill Process
Franck Adjogble, SMS group
The metals industry faces challenges due to a fragmented and often incrementally grown software architecture with isolated or repetitive functionalities, which hinder comprehensive data management and seamless integration between level 2 and level 3 applications. To address this, a platform convergence approach is proposed, developing a modular architecture that enhances integration, configurability and scalability, facilitating seamless communication between applications. This transformation aims to create a unified user experience and streamline operations, supporting the advancement to holistic production optimization. The solution must address the need for a homogenous software environment, fostering sustainable partnerships and ongoing innovation. |
5 p.m. | Reception |
Wednesday, 14 October 2026
|
7 a.m. | Breakfast |
8 a.m. | Keynote: From Islands to Ecosystems: Interoperability as the Steel Industry’s Productivity Engine
John Dyck, CESMII-The Smart Manufacturing Institute
As steelmakers invest in new capacity and upgrades, the biggest productivity gains will come from interoperability — not equipment size or speed. This keynote argues that standardization and connected architectures can deliver step-change improvements in yield, throughput, quality, energy performance and traceability. It outlines a practical path from siloed systems to vendor-neutral, plug-and-play operations using common identifiers, shared data semantics and open interfaces. Standardized information models reduce integration time, ease technical debt and enable tighter supplier connectivity — critical in capital-constrained, talent-tight environments. By creating a consistent data foundation, interoperability also makes AI scalable, turning legacy assets and process data into reliable, production-ready insights. The result is a coordinated, learning production system where every reshoring investment delivers amplified value. |
8:50 a.m. | Break |
9:05 a.m. | Safety Panel Discussion: Safety and Digitalization Technologies Ed LaBruan, Janus Automation
Gianluca Maccani, Polytec NA |
10:35 a.m. | Break |
10:50 a.m. | Enhancing EAF Temperature Accuracy: A Hybrid Approach Using Physical Bounds and Gradient Boosting,
Richard Jacinto Marquez Contreras, ECON Tech
Unreliable EAF dip-thermocouple measurements, such as false 1,700°C readings caused by setup issues or slag interference, lead to costly reblows and overheating. This case study details a novel "soft sensor" solution implemented at Gerdau Tultitlán. The framework integrates Industrial Internet of Things power profiles and physical constraints — like theoretical liquidus temperature anchors — with a dual-model AI system. By utilizing a gradient boosting regressor, a logistic classifier and evaluating temperature gradients, the system assigns a 0-to-1 reliability score to physical measurements. This approach instantly flags biased outliers, enabling precise, real-time bath temperature estimation and significant operational improvements. |
11:25 a.m. | Lessons Learned From a Decade of Developing Digital Twins for the Steel Industry
Matheus de Oliveira Mendonça, Enacom Group
Drawing on a decade of real-world industrial deployments, this session highlights why many digital initiatives fall short and what differentiates those that deliver sustained results. Through practical examples, it explores how the right technology choices, aligned processes and engaged people turn decision engines into embedded capabilities that generate measurable and lasting business impact. |
Noon | Lunch |
1 p.m. | How Generative AI Will Change Steel Production Optimization
Dirk Lieftucht, PSI Metals North America Inc.
This presentation will explore how machine learning and generative AI are transforming combinatorial optimization in the steel industry through examples that include: Generating hyperparameter settings for optimization algorithms; learning to branch within the search space; searching based on reinforcement learning; creating constraint models for process and material flow from historical data; and using large language models to generate user and parameter preferences integrating GenAI, machine learning and traditional optimization techniques will enable production management solutions to evolve into autonomous and adaptive systems. |
1:35 p.m. | A Data-Centric Strategy in Ironmaking Operations Brings Functional Intelligence and Building the Foundation for Generative AI
Suhas Rohit Mehta, Falkonry Inc.
While many steel plants deploy vibration sensors for predictive maintenance, broader data-centric operations remain underutilized despite extensive instrumentation and connectivity. Ironmaking operations continue to lose hundreds of production hours annually due to unstable furnace behavior, unexpected shutdowns and slowdowns. The industry’s critical challenge in the era of GenAI is not large language models, but extracting reliable, actionable insight from existing time-series data to drive measurable operational impact. This paper presents how a North American steelmaker digitally transformed ironmaking through an intentional, incremental adoption of a time-series intelligence platform. The deployment embedded AI into daily workflows designed by seasoned process experts, creating a structured, operations-driven monitoring discipline. The result has been measurable avoidance of production delays and downtime. By combining cultural alignment with a disciplined data strategy, the company established a trusted AI foundation that enhances operational stability and positions GenAI as an amplifier of real industrial value rather than a stand-alone solution. |
2:10 p.m. | Capturing Steel Expertise With Industrial AI
Bryan DeBois, RoviSys
The steel industry is grappling with an unprecedented loss of expertise as experienced workers retire or leave. This presentation will demonstrate how Industrial AI can capture and preserve critical knowledge before it’s lost. It will distinguish between Generative AI and Autonomous AI, share real-world examples of success in the steel industry, and explain why Industrial AI is the ultimate solution for empowering manufacturers to retain and scale their most valuable skills. If you are looking to embark on a digital transformation journey toward AI, this is a presentation that will guide your way. |
2:45 p.m. | Break |
3 p.m. | Panel Discussion: AI and GenAI Bryan DeBois, RoviSys
David Kober, Global Gauge Corp.
Tarun Mathur, ABB
Nathan Settlemire, Charter Steel – Cleveland, Ohio |
4:45 p.m. | Conference Adjourn |
