Figure 10. Box plots of local segregation data in terms of percentage mass of the more massive species for binary mixtures and broad PSDs. The fast fluidized bed had 324 datasets 14, while the turbulent fluidized bed had 190 datasets 21.
Figure 11 presents the random forest analyses of the relative influence of the variables on segregation extents. While both axial and radial segregation are well acknowledged for risers14,21,33-36, two interesting observations are made with respect to the contrast between the fast and turbulent fluidized beds. Firstly, while r/R was the most influential in the fast fluidized bed, it was the least in the turbulent bed. This is because the core-annulus profile is generally more significant in the fast fluidized bed relative to that in the turbulent one, and the radial profiles of chaotic parameters have been reported to be flatter for the turbulent bed 11. Notably, this is in contrast to that for local mass flux (Figure 7), in which r/R was the most dominant influence for both regimes. More studies are needed to understand why r/R has such significant effect on the local mass flux but not segregation in the turbulent fluidization regime. Secondly, while h/H had a negligible influence in the fast fluidized bed, it was more significant in the turbulent bed. This is to be expected as the particle concentration gradient is more significant than that of the fast fluidized bed. Therefore, although Figure 10 shows the two segregation datasets from the two fluidization regimes were statistically similar, Figure 11 suggests that different mechanisms govern the segregation behaviors.