90
4 Modelling Dynamically Structured Fluidisation
4.1 Introduction
Bubbling fluidised beds provide good mixing and transport rate for a broad range of
gas-solid operations in the industry. Nevertheless, insufficient understanding of its
complex hydrodynamics often complicates system design, control as well as scale-up
practices. Direct experimental investigation of fluidisation dynamics is often challenging, as it is largely restricted to measurement techniques and analytical considerations [13]. Over the recent years, sophisticated, non-invasive tomographic techniques, such as magnetic resonance imaging (MRI) [5, 40], X-ray tomography [42,
57], electrical capacitance tomography (ECT) [33, 34] or capacitance volume tomography (ECVT) [46], have been successfully implemented to study bubbling columns
and retrieve useful insights, but events on a microscale scale, such as interparticle
contacts, are yet hardly accessible in direct experimentation.
In parallel to experimental techniques, computational fluid dynamics (CFD) techniques have been increasingly adopted to facilitate engineering design and provide
fundamental insights into fluidisation processes [38, 53, 55]. Over the last decades,
researchers proposed a decent amount of models for granule flows at different
levels of complexity [2, 14, 28, 52, 53]. Among those, two frameworks are prevalent: two-fluid models (TFM) or Eulerian-Eulerian method, and computational fluid
dynamics-discrete particle models (CFD-DEM) or Eulerian-Lagrangian approach.
Both approaches are broadly employed and attain tremendous success in predicting
behaviour of bubbling columns [3, 13, 30, 56].
When modelling particles as discrete elements, the Lagrangian approach tracks
trajectories of every single particle explicitly [51, 53]. To resolve interparticle contact,
particles are assumed mathematically as either hard-spheres [1] or soft-spheres [12].
Soft-sphere treatment allows resolving the entire process of inelastic collision using
a spring-dashpot model, preferable in handling dense flows, in which particles often
encounter multiply, sustained contacts. Therefore, it is commonly adopted for analysis of solid mechanics [7, 8, 48]. When applied to describe gas-solid suspensions,
discrete solids couple gas flow fields computed on Eulerian grids with a mesh size
greater than particle characteristic length [13]. Such a coupling implementation,
nevertheless, is computationally expensive, and yet uneconomical for large scale
applications, where characteristic spatial scales of the two phases differ by several
orders of magnitude. To this date, a direct CFD-DEM implementation for gas-solid
flows is widely limited to the scope of fundamental studies [4, 20, 53]. Moreover,
several Lagrangian approaches with reduced level of detail, such as Particle-in-Cell
(PIC) approach [31, 48] and coarse-grained DEM approach [43], have been developed to tackle large-scale applications. The PIC approach hybridises Eulerian and
Lagrangian descriptions for the particulate phase [47]. It resolves motions of particle
clouds, representative parcels of multiple neighbouring particles, whereas models
solid collisions on Eulerian grids. On the other hand, coarse-grained simulations also
group multiple particles as a single virtual parcel, but calculate collision processes
and dynamics of parcels directly using DEM.
4 Modelling Dynamically Structured Fluidisation
4.1 Introduction
Bubbling fluidised beds provide good mixing and transport rate for a broad range of
gas-solid operations in the industry. Nevertheless, insufficient understanding of its
complex hydrodynamics often complicates system design, control as well as scale-up
practices. Direct experimental investigation of fluidisation dynamics is often challenging, as it is largely restricted to measurement techniques and analytical considerations [13]. Over the recent years, sophisticated, non-invasive tomographic techniques, such as magnetic resonance imaging (MRI) [5, 40], X-ray tomography [42,
57], electrical capacitance tomography (ECT) [33, 34] or capacitance volume tomography (ECVT) [46], have been successfully implemented to study bubbling columns
and retrieve useful insights, but events on a microscale scale, such as interparticle
contacts, are yet hardly accessible in direct experimentation.
In parallel to experimental techniques, computational fluid dynamics (CFD) techniques have been increasingly adopted to facilitate engineering design and provide
fundamental insights into fluidisation processes [38, 53, 55]. Over the last decades,
researchers proposed a decent amount of models for granule flows at different
levels of complexity [2, 14, 28, 52, 53]. Among those, two frameworks are prevalent: two-fluid models (TFM) or Eulerian-Eulerian method, and computational fluid
dynamics-discrete particle models (CFD-DEM) or Eulerian-Lagrangian approach.
Both approaches are broadly employed and attain tremendous success in predicting
behaviour of bubbling columns [3, 13, 30, 56].
When modelling particles as discrete elements, the Lagrangian approach tracks
trajectories of every single particle explicitly [51, 53]. To resolve interparticle contact,
particles are assumed mathematically as either hard-spheres [1] or soft-spheres [12].
Soft-sphere treatment allows resolving the entire process of inelastic collision using
a spring-dashpot model, preferable in handling dense flows, in which particles often
encounter multiply, sustained contacts. Therefore, it is commonly adopted for analysis of solid mechanics [7, 8, 48]. When applied to describe gas-solid suspensions,
discrete solids couple gas flow fields computed on Eulerian grids with a mesh size
greater than particle characteristic length [13]. Such a coupling implementation,
nevertheless, is computationally expensive, and yet uneconomical for large scale
applications, where characteristic spatial scales of the two phases differ by several
orders of magnitude. To this date, a direct CFD-DEM implementation for gas-solid
flows is widely limited to the scope of fundamental studies [4, 20, 53]. Moreover,
several Lagrangian approaches with reduced level of detail, such as Particle-in-Cell
(PIC) approach [31, 48] and coarse-grained DEM approach [43], have been developed to tackle large-scale applications. The PIC approach hybridises Eulerian and
Lagrangian descriptions for the particulate phase [47]. It resolves motions of particle
clouds, representative parcels of multiple neighbouring particles, whereas models
solid collisions on Eulerian grids. On the other hand, coarse-grained simulations also
group multiple particles as a single virtual parcel, but calculate collision processes
and dynamics of parcels directly using DEM.