‍

Application of AI Lifts Quality Strip Production to the Next Level – ArcelorMittal Eisenhüttenstadt Significantly Reduces Surface Defects

Dr. Thorsten Müller | Head of Quality, ArcelorMittal Eisenhüttenstadt
Dr. Nadine Decker | Head of Internal Quality, ArcelorMittal Eisenhüttenstadt
Jens Gellert | Head of Casting Department, ArcelorMittal Eisenhüttenstadt
Dr. Falk-Florian Henrich | Founder & CEO, Smart Steel Technologies
Dr. Otmar Jannasch | VP Metallurgy, Smart Steel Technologies
Dr. Jan Daldrop | Team Lead Machine Learning, Smart Steel Technologies

A substantial improvement of surface quality in production of galvanized strip for automotive exposed applications has been achieved by process control through Artificial Intelligence and Machine Learning technologies. This progress was achieved by blending steelmaking expertise and machine learning knowledge. The main innovation is not to predict surface quality, but to reduce, respectively to avoid, formation of surface defects. Thus, defects can be prevented before they occur and don’t have to be -cost intensively -repaired or cut out. Key to this success is a highly dynamic process optimization for each casting sequence, each heat and each single slab which was implemented based on methods from Artificial Intelligence and Machine Learning.

The algorithmic core of SST Casting AI finds the optimal combination of casting and melt shop parameters for optimal surface quality based on historic data. The optimizer takes into account all relevant process constraints and uses a combination of metallurgical modelling and artificial intelligence. The high-quality live data transformation forms the basis for the automated data-driven casting optimization. Through live integration with the production planning system and MES of the melt shop and caster an optimized casting sequence plan is computed dynamically for each production campaign. The casting optimization is complemented by specific live modules from SST Temperature AI to supply the optimum predicted steel superheat for the caster by calculating the correct heat temperature in melt shop. The implementation and validation in production has been successfully completed within only a few months at ArcelorMittal Eisenhüttenstadt, one of the advanced flat steel producers of the ArcelorMittal group.