Spheres vs. rods in fluidized beds: Numerical and experimental investigations

I. Mema

Research output: ThesisDissertation (TU Delft)

27 Downloads (Pure)


For the past century, fluidized beds have been standard equipment in many branches of industry. In most applications they are used to manipulate granular and powder-like materials, whose particles can roughly be approximated as spheres. Therefore, numerical models and investigations have focused mainly on fluidized beds with spherical particles. Recent decades witnessed an increase in the use of fluidized beds in biomass processing. Unlike other materials typically used in fluidized beds, biomass is characterized by relatively large and elongated particles. For the sake of simplicity, numerical models for simulating fluidization of elongated particles have so far neglected a lot of specifics that can occur during this process and even applied the same models and conclusions that were developed for fluidization of spherical particles. The goal of this thesis is to define what is necessary for performing physically correct Computational Fluid Dynamics - Discrete Element Model (CFD-DEM) simulations of elongated particles fluidization. This thesis emphasizes the difference in fluidization between spherical and elongated particles and looks into ways to include specific particle and fluid interactions related to elongated particles into numerical (CFD-DEM) model. Results fromCFD-DEMsimulationswere validated using two experimental techniques, magnetic particle tracking (MPT) and X-ray tomography (XRT). This thesis is part of larger project of multi-scale modeling of fluidized beds with elongated particles and is focusing on the middle scale, bridging fully resolved, direct numerical simulations (DNS) with large scale, two fluid model (TFM) or multi-phase - particle in cell (MP-PIC) models, capable of simulating industrial sized fluidized beds. This thesis first looks in to the effect of including shape induced lift force and hydrodynamic torque, which were so far neglected in CFD-DEM simulations of elongated particles. It is shown that including lift force and hydrodynamic torque leads to considerable changes in the particle vertical velocity and particle preferred orientation in the fluidized bed. Looking into the mixing characteristics, as one of the most important parameters of fluidized beds, also considerable differenceswere found. Further differences in fluidization behaviour of spherical and elongated particles, as well as the effect of increasing particle aspect ratio, were shown experimentally, using MPT. Clear differences between spherical and elongated particles were found concerning the particle velocity and rotational velocity distributions. The effect of increasing particle aspect ratio and gas inlet velocity on fluidization of elongated particles was shown. Using XRT, the difference in bubbling and slugging fluidization between spherical and elongated particles was shown. In the end, the effect of newly developed multi-particle correlations for hydrodynamic forces and torque was tested, and it is concluded that they can improve the accuracy of simulations of dense fluidized beds containing elongated particles. The findings of this thesis clearly show that the models and assumptions developed for fluidization of spherical particles cannot simply be transferred to the fluidization of elongated particles. The results presented here give a new insight in the fluidization of elongated particles. They are also valuable for validation and development of larger scale models capable of simulating industrial size fluidized bed with elongated particles.
Original languageEnglish
Awarding Institution
  • Delft University of Technology
  • Padding, J.T., Supervisor
  • de Jong, W., Supervisor
Award date3 Dec 2020
Publication statusPublished - 2020


  • Fluidized bed
  • Non-spherical particles
  • MPT
  • XRT
  • Hydrodynamic forces


Dive into the research topics of 'Spheres vs. rods in fluidized beds: Numerical and experimental investigations'. Together they form a unique fingerprint.

Cite this