Abstract
Shape of blasted rock particles is a critical, yet often overlooked, factor in mining operations. While blast-induced particle size distribution provides essential information, shape characteristics such as sphericity, elongation, and angularity significantly influence material flow, energy consumption, and equipment wear. Irregularly shaped particles lead to increased interlocking and reduced flowability in underground ore passes, chutes, and conveyors. Irregular or elongated rock fragments can cause jamming in the crusher, leading to clogging in the feed area and additional stoppage to clear the blockage. Angular particles also accelerate equipment wear (e.g. shovel nails, crusher’s components and mail liners) and increase maintenance costs. Despite these critical implications, conventional fragmentation analysis methods remain limited in their ability to accurately characterize rock shape in three dimensions.
This study introduces a 3D shape analysis framework that leverages LiDAR technology and point cloud deep learning-based segmentation to provide a robust and lighting-independent solution for quantitative rock shape characterization. Unlike traditional 2D image-based approaches that depend on lighting conditions and often oversimplify particle geometries by qualitative analysis, our method captures the full 3D morphology of rock fragments, enabling precise quantification of shape metrics. By integrating shape descriptors, such as sphericity, angularity, aspect ratio and ferret diameter, this approach offers more profound insights into the impacts of particle shape on material handling and processing.
A key challenge in applying deep learning to shape analysis is the scarcity of labeled 3D datasets. To address this, we developed a synthetic labeled dataset platform that generates diverse training data for model development. This enables the automation of shape prediction while reducing the time and cost associated with manual labeling.
Accurate prediction and analysis of rock shape offer significant operational benefits. By understanding and mitigating the impact of irregular shapes on material flow and equipment wear, we can improve efficiency in loading, hauling, crushing, and grinding operations, leading to reduced energy consumption, lower maintenance costs, and increased throughput. Also, the lighting-independent capabilities of our LiDAR-based approach ensure consistent and reliable performance in underground and nighttime environments, where conventional image-based techniques are less effective.
In conclusion, this study highlights the importance of integrating shape analysis into rock fragmentation studies and provides a 3D scalable solution for mining applications. This approach enhances operational efficiency by offering precise shape characterization, paving the way for more informed decision-making in modern mining practices.