A Generalized Rule for a Grid Independent Solution for TFM Simulations of Geldart B Bubbling Fluidized Beds

Abstract

Bubbling fluidized beds (BFBs) of Geldart B particles are commonly encountered in combustion and gasification applications. Recently, they have been used as fuel reactor in the Chemical-Looping Combustion (CLC) process, a leading alternative to reduce the economic cost of CO₂capture. Computational Fluid Dynamics (CFD) modeling can aid in design and scale-up of fuel reactor used in the CLC process. The Eulerian–Eulerian CFD approach, commonly known as kinetic theory-based two fluid model (TFM), is extensively used for small and large-scale simulation of gas-solid fluidized beds due its relatively low computational cost. However, application of TFM to industrial-scale BFBs is limited due to requirement of fine spatiotemporal discretization to solve the mesoscale structure. The characteristic properties such as bubble formation, bubble size and shape of bubbling fluidized bed are largely dependent on grid resolution, and poor resolution leads to inaccurate prediction of hydrodynamic behavior and inaccurate prediction of reactor performance. We have reviewed several published reports showing grid independent solutions for Geldart B BFBs, and it is apparent that grid independence depends on particle size.

In this study, a large number of two dimensional CFD simulations were performed for two pilot-scale BFBs at two diameters: 30.8 cm and 58.0 cm. We have evaluated the effect of particle diameter on mesh requirements for grid independence. All simulations were performed using MFIX, an open source code developed by NETL, USA. To identify a grid independent solution for the TFM simulation of BFBs, carefully computed bed expansions from CFD were compared with available experimental result (Geldart, D. Powder Technology 1968, 1, 355) and empirical predictions. Grid independent solutions are achieved with a mesh size directly proportional to particle size, irrespective of the scale of the fluidized beds. Our results show that, for the range of Geldart B systems studies, grid independent solutions are achieved with mesh size approximately equal to 19 particle diameters. Results are compared to numerous published results, and are in good agreement. Up until now, common practice is to use a rule of mesh size equal to 10 particle diameters, a rule derived for riser flow simulations. With a somewhat relaxed constraint for mesh size, our conclusion will allow for simulation of industrial scale reactors. This will serve as guideline for choosing the appropriate grid size and for avoiding the time consuming grid independent study that requires for all numerical simulations.