On POD-based modal analysis in simulations of granular flows

Abstract

Understanding the dynamics of granular flows is of great significance in various engineering fields but challenging because of flow complexity. The discrete element method (DEM) is often employed in simulations of various granular systems. Recently, modal analysis has been shown to remarkably deepen the understanding of these systems. The principle of the modal analysis is to extract a set of dominant structures from experimental or numerical datasets. In this study, an advanced modal analysis technique, i.e., the Lanczos-based proper orthogonal decomposition (LPOD), is incorporated into numerical simulations of a particle system. Two types of applications are presented. First, the modal analysis based on the LPOD technique is adopted to evaluate particle mixing in a rolling drum. It is demonstrated that the POD-mode distributions of particle movements can reveal dominant mechanisms, such as convection/diffusion. Second, the LPOD technique based modal analysis is employed to validate the adequacy of a coarse-grained DEM in a spouted bed. The results show that similar patterns of POD modes and related energy spectra are observed in the original particle and the coarse-grained particle systems. Consequently, not only can POD analysis be used to identify main mixing mechanisms, but also POD results are shown to be new effective parameters in the validation of the coarse-grained DEM. • An advanced Lanczos-based POD (LPOD) was employed to analyze granular phenomena. • Applicability of the LPOD was shown in simulations of a drum and a spouted bed. • LPOD was demonstrated to effectively identify main mixing mechanisms. • LPOD provided new parameters to effectively validate a coarse-grained DEM.