A Digital Twin Framework for Continuous Casting With Integrated AI-Based Surface Inspection

Predictive first-principle models, AI-driven surface inspection and advanced visualizations have been integrated into a digital twin framework for continuous casting. The system provides real-time process representation with state predictions, surface inspection overlays and synchronized parameter tracing to analyze defect root causes. It also visualizes the impact of casting condition changes on predicted quality outcomes. Trials on slab casters show that the framework delivers valuable insights for operators and engineers to improve process stability and product quality. Ongoing development focuses on expanding predictive capabilities, refining adaptive feedback, and integrating broader upstream and downstream quality data for full-process optimization.

Authors:

Dr Hannu Suopajärvi | Sapotech Oy
 

Saku Kaukonen | Sapotech Oy
 

Professor Seppo Louhenkilpi | Casim Consulting Oy
 

Risto Vesanen | Casim Consulting Oy
 

Session Chairs:

Norbert Strobl | ArcelorMittal Dofasco G.P.
 

Tathagata Bhattacharya | ArcelorMittal Global R&D - East Chicago
 

Joydeep Sengupta | ArcelorMittal Global R&D - Hamilton