Shape properties validation of virtual 3D particles created from AIMS 2D images for railway ballast DEM modeling

Abstract

Laboratory-scale research on railway ballast often fails to produce parameters reflecting real-world conditions, while real-scale research incurs high costs. Advancements in computational capacity allowed for discrete element method (DEM) to simulate ballast behavior with three-dimensional, irregularly shaped particles. This research focuses on developing virtual 3D particles for DEM based on digital image processing (DIP) from the use of the Aggregate Imaging Measurement System (AIMS). This can potentially provide a rationale for taking full advantage of databases of aggregate properties obtained with this equipment over more than a decade across various regions worldwide. Quarry-produced aggregates were characterized in terms of shape properties in three orthogonal positions using AIMS. Virtual 3D particles were generated from one, two, or three real 2D images, with strong correlations between real and virtual particles results obtained for sphericity, flatness, elongation, and flatness/elongation ratio. This study shows that generating virtual 3D particles from one single real 2D image from AIMS is an effective and time-efficient process. Furthermore, shape properties classification of virtual particles closely matched real ones, with minimal variation near classification boundaries, confirming the method’s consistency. This approach can be an alternative to more computationally expensive 3D modeling, as well as allowing for the virtual reproduction of aggregates not locally available by sharing AIMS databases. Finally, numerical simulations were proven to be sensitive to real particle shapes, allowing for better understanding of ballast performance, leading to optimization of maintenance and reducing track wear and elements’ failure.