• The IRIDISCENTE project aims to improve the production of more sustainable steels and achieve net-zero emissions.
• In its first 12 months, it has made progress in developing AI models to optimise scrap composition combinations that yield the best properties in the final 3D-printed parts.

The automation of defect detection processes in the microstructure of 3D printed components is just one of the advances in sustainable steel manufacturing that the European project IRIDISCENTE is working on.

With a budget of nearly €10 million and coordinated by IMDEA Materials Institute and ArcelorMittal, one of the world’s leading integrated steel companies, the project recently celebrated its first anniversary.

IRIDISCENTE seeks to improve the recyclability of steel through the integration of artificial intelligence (AI) and additive manufacturing. By maximising the reuse of scrap and waste, the project’s researchers aim to reduce carbon emissions and dependence on critical raw materials.

The process involves melting and refining recycled steel before atomising it into powder. This processed powder can be used in additive manufacturing techniques, such as laser powder bed fusion (LPBF) or direct energy deposition (DED).

“One of the biggest challenges in promoting scrap reuse in additive manufacturing is ensuring the quality of the produced powder,” explains Laura del Río Fernández, Additive Manufacturing Scientist at ArcelorMittal. “Ensuring the quality and homogeneity of the powder is as critical as it is complex, and the variability of recycled material presents a significant challenge.”

“However, its impact on the transition to net-zero emissions steel is enormous. Although this technology is still evolving, the recyclability of steel and its by-products is key to achieving a circular economy.”

Recycling metal scrap for steel additive manufacturing presents several key challenges. The atomisation process, where molten metal is turned into powder, is highly sensitive to the quality of the raw material, as is the 3D printing process, which is highly sensitive to the quality of the powder.

Therefore, maintaining strict quality control is crucial when there is significant potential for variability in the nominal chemical composition of the raw material and a high potential for contamination.

Additionally, the integration of AI and machine learning to optimise compositions faces challenges due to the lack of extensive, validated datasets required to predict how different scrap mixtures affect final properties.

*”Currently, we do not know the optimal scrap metal mix that would allow us to improve the properties of the final part while also ensuring a smooth atomisation and printing process,” explains David Noriega from TheNextPangea, one of the project’s partners.

“To bridge this knowledge gap, we use machine learning to generate models that establish relationships between all the factors involved.”

“Additionally, we are developing models aimed at identifying defects in the characterisation of the printed parts’ microstructure, with the goal of automating their detection,” he adds.

This work is part of a broader effort to optimise processes and reduce the costs associated with material production. “By filling these knowledge gaps with our models, we can propose new processing routes that facilitate the creation of new products,” Noriega concludes.*

The annual IRIDISCENTE meeting, held at ArcelorMittal’s R&D facilities in Avilés, brought together representatives from each of the project’s partners, which also include Universidad Carlos III de Madrid, Universidad de Burgos, AIMEN Technology Centre, Renishaw Ibérica, Blesol Tech, mim-tech ALFA, AENIUM, IDAERO SOLUTIONS, and Syspro Automation. They presented progress and defined key objectives for the next 12 months.

“We will continue researching AI algorithms for the automatic processing of data on raw materials, manufacturing processes, physical simulation models, monitoring devices, and experimentally produced materials characterisation,” states del Río Fernández.

“We will also develop simulation models to predict material properties derived from the powder production process and subsequent manufacturing processes, establishing relationships between process, microstructure, properties, and performance,” she adds.

IRIDISCENTE (PLEC2023-010190) is funded by MCIN/AEI/10.13039/501100011033, and the MIG-20232094 project is funded by the Science and Innovation Missions Program within the TransMisiones 2023 initiative, with funding from MCIN/CDTI.