Unlocking the Potential of Recycled Aluminium through Machine Learning, High-Throughput Microanalysis, and Computational Mechanics

Forskningsprojekt, 2025 – 2030

Aluminium is vital to the green transition and is therefore classified by the EU as a critical material. Using recycled aluminium significantly enhances sustainability. However, during the recycling process, a wide range of tramp elements (impurities) accumulate and often negatively affect the material's properties. A recent disruptive manufacturing technology in large-scale high pressure die casting -- commonly known as mega-casting, promises the wide adoption of recycled aluminium in high-value applications. However, the interplay of numerous tramp elements and diverse thermomechanical parameters creates a high-dimensional space that is challenging to fully explore using traditional experimental and computational methods. In this project, we aim to develop advanced machine learning frameworks to gain a fundamental understanding of the microstructure and mechanical properties of cast recycled aluminium, as well as to accurately predict the performance of large components made of  recycled aluminium. Specifically, we will: i) develop high-throughput methods to acquire multi-length scale and multi-modal microstructure information and probe the underlying material mechanisms using advanced microscopy; ii) develop novel deep learning-aided computer vision methods to accelerate microstructural analysis and generate high-fidelity three-dimensional models to represent the inhomogeneous microstructure; iii) establish a finite element analysis framework using crystal plasticity to accurately predict  mechanical properties based on various microstructures; iv) discover neuro-symbolic models for the effective behaviour of the complex microstructure, realising virtual validation of recycled aluminium cast components. This project will significantly accelerate the adoption of recycled aluminium in various applications, unlocking its full potential.