Key influence of clusters of Geldart Group B particles in a circulating fluidized bed riser

Abstract

The clustering phenomenon is an important characteristic of fluidized bed systems, so much attention has been given to understanding such unstable, transient features through both experiments and simulations. A review has pointed out that, because of the interplay of multiple factors, relationships are at times unclear even within the same study. For such non-linear and multi-dimensional problems, machine learning tools are proficient. In this study, self-organizing map (SOM) analysis was harnessed to classify 1188 circulating fluidized bed (CFB) riser cluster datasets of Geldart Group B particles into potential smaller data assemblies, in order to determine the key influence(s) responsible for the demarcation. Two distinct data assemblies were identified, with one constituted by the monodisperse particle systems (i.e., three narrow particle size distributions (PSDs)), while the other by the non-monodisperse particle systems (i.e., two binary mixtures and one broad PSD). Specifically, the clusters formed by the non-monodisperse systems were distinctively smaller than those of monodisperse ones. This suggests that multiple particle types hindered the growth of clusters, which has been tied to hydrodynamic screening, unequal charging and unequal damping effects that are unique to particle mixtures. More studies are needed to unveil the underlying mechanisms of such different clusters between the monodisperse versus non-monodisperse particle systems.