ReselScale

The main objective of ReselScale is to provide actionable guidance to the steel industry for managing surface quality during the shift to carbon-neutral steel production.

The project will:

• Produce and study up to 40 laboratory-cast and 3 industrial steel grades with varying levels of residual elements (DP, HSLA, low-carbon, structural steels).

• Investigate primary oxide scale formation using multiple reheating technologies: conventional furnaces, Thin Strip Casting and Rolling (TSCR), and Endless Strip Production (ESP), including electrification and hydrogen-based reheating.

• Conduct descaling and hot rolling simulations (over 60 trials) to assess how residuals influence descalability and surface quality.

• Analyse pickling performance through 120 laboratory and pilot-scale tests, quantifying contamination levels of Cu, Ni, and Sn.

• Study heat transfer coefficients in relation to scale composition and porosity using over 100 targeted experiments.

• Develop an automated SEM-EDS image analysis tool and a machine learning model (XGBoost) to predict oxide behaviour, growth, and adhesion based on alloy and process variables.

• Disseminate and communicate results across the European steel community, ensuring widespread access to the developed industrial guidelines for green steel processing.

The global steel sector faces an unprecedented shift toward green and circular production. With the rise of Electric Arc Furnace (EAF) and Direct Reduction (DRI) routes, residual elements from scrap are becoming a structural component of future steelmaking. These elements, while promoting circularity, complicate the control of oxide scale formation, adhesion, and removal.

ReselScale will experimentally reproduce the different industrial reheating and rolling conditions to quantify these effects, combining traditional approaches with innovative low-carbon technologies such as hydrogen combustion and induction preheating.

The project’s methodology includes large-scale laboratory and pilot simulations covering the full chain, reheating, descaling, rolling, and pickling, supported by advanced surface characterisation and data-driven modelling.

• Reheating trials will determine how residual elements affect oxide morphology and adhesion across different atmospheres and heating profiles.

• Descaling experiments will reproduce industrial hydraulic descaling, exploring the combined effects of multiple residuals, a first in this field.

• Hot rolling simulations will evaluate scale cracking, rolling-in defects, and surface roughness changes induced by element synergies.

• Pickling trials will investigate oxide composition and dissolution mechanisms to optimise cleaning processes and bath recyclability.

• Heat transfer studies will quantify how scale structure affects cooling efficiency and temperature control, feeding back into process modelling.

Data Analytics and Machine Learning

Through advanced image processing and AI-assisted analysis, ReselScale will transform oxide characterisation into quantifiable, predictive datasets. Automated tools will measure scale thickness, porosity, sublayers, and interfaces, correlating these with process parameters and residual concentrations.

Machine learning models (based on XGBoost) will enable accurate predictions of oxide growth, descalability, and mechanical behaviour, ultimately allowing steelmakers to simulate process outcomes for future steel compositions.