Quantitative Solids Mixing Analysis Using Interpretable Image Processing Methods

Abstract

This study presents a quantitative framework for evaluating lateral mixing in a bubbling fluidized bed based on high-speed image analysis and Lacey mixing index estimation. Experiments were conducted in a pseudo-two-dimensional fluidized bed using optically distinguishable glass beads, enabling detailed tracking of mixing dynamics over time. Recorded images were processed through a pipeline comprising contrast enhancement, noise reduction, and segmentation via adaptive thresholding to isolate tracer particles. Spatial concentration fields were then extracted from the segmented images and used to estimate local variances necessary for computing the Lacey index as a time-resolved measure of mixing progression from a fully segregated to a randomly mixed state. By providing clear, quantitative indicators of mixing evolution, the proposed methodology offers a potential basis for integrating interpretable measurements into digital models of fluidized bed systems. These image-based insights could contribute to improving the transparency of digital representations of fluidized beds and support future efforts toward explainable digital twins for process monitoring and control. The results suggest that combining classical image processing with statistical metrics provides a reliable means to analyze and understand complex mixing dynamics in a way that may be incorporated into interpretable digital frameworks.