Unveiling particle mixing from non-destructive 3D XCT imaging with machine learning aided spatial distribution analysis

Efficient mixing of binary particle systems is essential in process engineering, as it directly impacts product quality, stability, and cost. Traditional evaluation methods rely on empirical modeling from the theoretical assumptions or macroscale characterizations, both remaining time- and cost-intensive. In this study, X-ray computed tomography (XCT) is employed to perform non-destructive three-dimensional imaging of particulate systems. This advanced technique enables detailed characterization of microstructural features and spatial arrangements, yielding critical insights into mixing conditions at the microscale, and is quantified by tailored evaluation metrics. Machine learning-enhanced image segmentation enables efficient particle identification in 2D cross-sections from 3D XCT data. This framework enhances XCT-based feature extraction, enabling simultaneous qualitative observation and quantitative analysis to optimize and improve chemical engineering processes and product quality.