Micro/meso simulation of a fluidized bed in a homogeneous bubbling regime

Abstract

Due to the wide range of spatial scales and the complex features associated to fluid/solid and solid/solid interactions in a dense fluidized bed, the system can be studied at different length scales, namely micro, meso and macro. In this work, we select a flow configuration relevant of a homogeneous liquid/solid fluidization and compare computed results from Particle Resolved Simulation (PRS) with those from locally averaged Euler/Lagrange simulation. PRS at the micro-scale is carried out by a parallel Distributed Lagrange Multiplier (DLM) solver in the framework of fictitious domain methods (Wachs, 2011a, 2015). For meso-scale simulations, the set of mass and momentum conservation equations is averaged in control volumes encompassing few particles and momentum transfer between the two phases is modeled using appropriate drag laws. Both methods are coupled to a Discrete Element Method (DEM) combined with a soft-sphere contact model to solve the Newton–Euler equations with collisions for the particles in a Lagrangian framework (Wachs et al., 2012). A test case of intermediate size with 2000 spheres is chosen as a sensible compromise between size limitations of the meso-scale model for an appropriate averaging process and computational resources required to run micro-scale simulations. These two datasets yield new insight on momentum transfer at different spatial scales in the flow, and question the validity of certain approximations adopted in the meso-scale model. Results demonstrate an acceptable agreement between the micro- and meso-scale predictions on integral measures as pressure drop and bed height. Investigating more detailed features of the flow, it has been shown that particles fluctuations are considerably suppressed in meso-scale simulations and in particular the particles transverse motion is underestimated, regardless of the selected drag law. The origin of these dependencies is carefully investigated by reconstructing the closure laws based on PRS results and comparing them to the closure laws proposed in the literature.