j m a t e r r e s t e c h n o l .

2 0 1 5;4(2):208216

Available online at www.sciencedirect.com

www.jmrt.com.br

Review Article

Fluidized bed modeling applied to the analysis of

processes: review and state of the art
Caterina Goncalves Philippsen a, , Antnio Cezar Faria Vilela a , Leandro Dalla Zen b
a
b

Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brazil

Fundaco de Cincia e Tecnologia, Porto Alegre, RS, Brazil

a r t i c l e

i n f o

a b s t r a c t

Article history:

The uidized bed is a technology that involves multiple phases, allowing for efcient con-

Received 13 August 2014

tact between them, therefore it is widely used in the chemical industry, metallurgy, oil and

Accepted 22 October 2014

thermal power generation. In uidized bed processes, the gassolid interactions and chem-

Available online 30 December 2014

ical reactions generate a large number of variables to be handled, making the process very
complex. Therefore, uidized bed modeling and simulation is widely used to predict and

Keywords:

analyze different processes, but it is possible to nd in the literature many mathematical

Mathematical modeling

correlations that describe this type of ow. Based on this, the present work presents a review

Numerical simulation

of the main mathematical models that describe the behavior of a uidized bed reactor, and

Fluidized bed

the state of the art regarding the use of modeling and simulation of the bed to predict and
analyze different processes. As a result of this review, we can observe the importance of further development of the hydrodynamic modeling of uidized beds, where understanding
the interactions between the phases and the inuence of this interaction is crucial for a better understanding and control of the processes. Generating experimental data of gassolid
and solidsolid interactions is also required for the validation of the numerical models.
2014 Brazilian Metallurgical, Materials and Mining Association. Published by Elsevier
Editora Ltda. All rights reserved.

Caterina Goncalves Philippsen graduated in physics from University of

Vale do Rio dos Sinos (UNISINOS)
in 2008, and master in Mechanical
Engineering from UNISINOS in 2012.
Now she is currently pursuing her
PhD in Mining Engineering, Metallurgical and Materials from Federal

University of Rio Grande do Sul (UFRGS). Her research, under

Dr. Antnio Cezar Faria Vilela and Dr. Leandro Dalla Zen,
studies the uid dynamics behavior and their inuences in
combustion processes in bubbling uidized beds. Since beginning his doctorate in 2012, Philippsen conducts this study in
the combustion laboratory of the Foundation for Science and
Technology, CIENTEC. Philippsen looks to complete her studies by 2016.

Corresponding author.
E-mail: catigp@yahoo.com.br (C.G. Philippsen).
http://dx.doi.org/10.1016/j.jmrt.2014.10.018
2238-7854/ 2014 Brazilian Metallurgical, Materials and Mining Association. Published by Elsevier Editora Ltda. All rights reserved.

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

Antnio Cezar Faria Vilela graduated in

Metallurgical Engineering from UFRGS
in 1977, master in Metallurgical Engineering from UFRGS in 1980 and doctorate (Dr.-Ing) in the Steel Institute
Rheinisch Technische Hochschule Westflischer/Aachen/Germany (1986). He is
currently associate professor, level 04, at
the UFRGS. Operates at graduation in the
Department of Metallurgy in the graduate
program, in Graduate in Mining Engineering, Metallurgical and Materials (PPGE3M), and coordinates
the Laboratory Steel since its foundation in the School of
Engineering. Has experience in guiding students and Technology Center, in conducting projects in Metallurgical and
Materials Engineering. Has served on the editorial board of scientic and technical magazines, in organizing annual events
in the areas of reduction and steelmaking and in supporting the activities of several promoters research bodies such
as the CNPq, CAPES, and FINEP FAPERGS. Was elected Director of ABM Brazilian Association of Metallurgy, Mining and
Materials for the term 2007/2009. Currently a Director of
ABM. In 2007, his work was recognized by the Foundation for
Research Support of the State of Rio Grande do Sul FAPERGS,
statewide, with FAPERGS AWARD RESEARCHER FEATURED
IN ENGINEERING. In 2012, his work has also been recognized
nationally by the Brazilian Association of Metallurgy, Mining
and Materials with MEDAL VICENTE Chiaverini MERIT IN
PROCESS.
Leandro Dalla Zen graduated in Mechanical Engineering from UFRGS in 1977,
master in Mechanical Engineering from
UFRGS in 1981 and doctorate in Mining
Engineering, Metallurgical and Materials
from UFRGS in 2000. Currently is an
adjunct professor at the UNISINOS, and
consultant of the City of Santa Vitria
do Palmar, a consultant for the Center
of Investigaciones Energticas Medioambientales Y Tecnologicas from Spain and
technical adviser of the Federation of Industries of Rio
Grande do Sul. Has experience in Mechanical Engineering
with emphasis on Harnessing of Energy, acting on the following topics: rice husk, combustion of biomass and waste
into uidized bed, energy conservation and alternative energy
sources. Also provides advice and design of passive energy systems in architecture: natural ventilation, thermal inertia, solar
energy.

1.

Introduction

The uidized bed technology consists in the combustion of the

particulate solid fuel in an inert material bed (usually sand),
which is uidized due to the ow of a gas. This type of ow
allows efcient gassolid contact, therefore it is widely used
in covering particles, drying, granulation, blending, combustion and gasication processes. In this ow, there are many
variables to be handled where the gassolid interaction and
chemical reaction characterizing a complex process. Because
of this, there is a growing demand for mathematical models
that allow the description and analysis for the development
of a better understanding of processes and for creating new
reactor projects.

209

The mathematical modeling of gas ow in uidized beds

began in the 60s with Davidson and Harrison [1] and Kunii and
Levenspiel [2], where they analyzed mainly bubble motion,
system instability and mass transfer. Based on these models,
many authors studied this subject [38], among others who
have contributed to the modeling of uidized bed reactors
with the aim of developing greater control of the system.
Computational Fluid Dynamics (CFD) is widely used to
predict physical and chemical phenomena during the uidized bed processes. CFD uses mathematical models based
on mass transport phenomena, energy and momentum, along
with theoretical and empirical correlations that require a long
processing time, because it uses more complex and broader
models, requiring more powerful computers.
In the literature, there are several mathematical correlations that describe a uidized bed process, but these models
depend on the application, since there is not a model with
universal applicability [9]. For a very precise description of the
process, there is a complex set of equations that must be analyzed before one can start the problem solving procedure. This
analysis must be congruent with the goals to be achieved and
with the available data [1013].
In thermochemical conversion processes, efciency is
mainly connected to chemical reactions and heat transfer, where the mixture between gas and solids has great
importance in mass and energy transfer [14]. Therefore, a
hydrodynamic study of the uidized bed is important to
improve the process, because it is what determines the distribution of the phases and the species involved.
Improving the hydrodynamic description is one of the
current challenges in order to improve understanding of
the processes in uidized beds [15]. A representation of the
physical or hydrodynamic characteristics of the system, as
realistically as possible, is required for modeling a chemical
reactor [16,17]. Carvalho [10] and Deen et al. [18] also claim
that improving the hydrodynamic description is necessary,
especially for particles from group B, according to the furtherpresented Geldart Classication. For Van Lare [19], the uid
dynamics parameters inuence the mass transfer between
the bubble and emulsion phases in the bed, which should be
the highest possible to maximize conversion in heterogeneous
reactions in uidized beds. This knowledge is important to
establish correct parameters of the reaction and mass transport, and it is useful for making decisions about the reactor
performance [12,20].
Based on this information, the present work presents a
review of the main mathematical models that describe the
hydrodynamic behavior of a uidized bed reactor, and the
state of the art regarding the use of modeling and simulation
to predict and analyze different processes in uidized beds,
followed by the main considerations about the models used.

2.

Review: uidized bed modeling

2.1.

Hydrodynamic model

The hydrodynamic models describe the motion and distribution of solids, gassolid mixture, size, velocity and growth

210

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

of bubbles, and the relation between bubble and emulsion

phases and mass and heat transfer phenomena [9].
The hydrodynamics of a uidized bed are given basically
by the balance of forces between particles and gas velocity. By
controlling the gas velocity, it is possible to set the required uidization regime (Fig. 1). The xed bed is characterized by a low
gas velocity, keeping the bed static. The minimum uidization
regime is the starting point of the uidization regime. The
bubbling regime arises when gas velocity exceeds the minimum uidization velocity, generating instability in the ow.
Pneumatic transport occurs when gas velocity is greater than
terminal velocity, and it is used in circulating uidized beds
[2,9,21].
The uidized bed is also characterized by the relation
between pressure drop and gas velocity. When the gas passes
through the porous bed, it loses pressure. In the xed bed,
the pressure drop increases linearly with increasing gas velocity until the pressure drop balances with the weight of the
particles. This equilibrium is characterized by a minimum uidization velocity (umf ). In the bubbling regime, the pressure
drop remains constant even with increasing gas velocity. The
bed pressure drop decreases when gas velocity is above the terminal velocity of the particles (ut ) and that is when pneumatic
transport starts [2,9,21].
Fluidization is largely inuenced by the characteristics of
the particles. Geldart [5] classied particle behavior in uidization into four groups, which are widely accepted and used in
uidized bed modeling.
Group A: Small particles (30150 m), and low density
(<1.4 g/cm3 ). The uidization is easy, smooth and homogeneous. It makes possible operating with low gas ows and
controlling the growth and speed of the bubbles.
Group B: Particles with medium diameter (40500 m) and
density between 1.4 and 4 g/cm3 . The uidization is good
for high gas ow rates. The bubbles tend to grow a lot and
appear at the beginning of uidization (umf umb ).
Group C: Very small particles (d < 30 m). Fluidization is difcult.
Group D: Dense and large particles (d > 500 m). Fluidization is difcult and non-uniform, ideal for spouted
beds.

Fixed bed

In uidized bed modeling, the minimum uidization condition is determined by the physical properties of the particles,
where porosity, pressure drop, bed expansion and gas velocity are dened. These characteristics allow determination of
bubble diameter and velocity, which inuence mass and heat
transfer between the bubble and emulsion phases. In Yang [9]
it is possible to nd many correlations about uidization of
particles.

2.1.1.

Gassolid interface model

In a gassolid inter phase model in a uidized bed it is

generally used a two-phase model (bubbleemulsion) or a
three-phase model (bubblecloud/wakeemulsion).
The two-phase model consists of one dense phase, or emulsion, formed by a big number of particles, and a dilute phase,
or bubble, without particles. Toomey and Johnstone [22] began
the two-phase model development, which consists of perfect
blending in emulsion phase, plug ow in bubble phase, heterogeneous reactions in emulsion phase, minimum uidization
condition in an emulsion phase. Solids motion occurs only
while the bubbles pass through the emulsion and the thermal
effects are reduced.
Based on this model, Davidson and Harrison [1] presented
the rst multiphase model, where they added that the bubble diameter is constant and mass transfer has a diffusion
and convective contribution. The authors also developed an
integrated model, which considers the emulsion and bubble
phases as a plug ow. In Partridge and Rowes [23] model,
the authors consider just diffusive transport, where the cloud
region belongs to emulsion phase or bubble phase. In Kato and
Wen [4], the uidized bed modeling considers compartments
with height equal to bubble diameter, which is a bed height
function.
According to Kunii and Levenspiel [2], depending on bubble
velocity, a third region called cloud arises, where the gas recirculation occurs around the bubble, mostly observed in fast
bubbles. This occurs mainly in reasonably high beds (>0.3 m),
in which the emulsion phase does not maintain the minimum
uidization conditions causing solids recirculation [24]. Based
on this, the authors [2] developed the three-phase model
(bubble cloud/wake and emulsion), where the wake phase
belongs to the cloud phase, making it more complex than the

Minimum
fluidization

Bubbling
bed

Pneumatic
transport

Ht
Hmf

Packed
bed height

Gas

Gas

Gas

Fig. 1 Hydrodynamic behavior of a uidized bed [21].

Gas

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

two-phase model. This model assumes that the emulsion

phase does not maintain the minimum uidization conditions, there is solids recirculation in the bed, the mass transfer
coefcient of bubbleemulsion has a contribution in mass
transfer bubblecloud and cloudemulsion, there are heterogeneous chemical reactions occurring in the emulsion and
cloud phase. Kunii and Levenspiels model represents uidized
beds more efciently, but it has disadvantages to computational implementation [24]. Following Kunii and Levenspiels
model, Grace [25] considers the existence of particles in bubble phase possibly reactive, and that all uidizing gas passes
through the bed in bubble form.

2.2.

Transfer phenomena

The mass and heat transfer phenomena are associated with

the contact between the phases involved. The higher the contact, the greater will be the coefcients of heat and mass
transfer in the bed, which affects the reactors performance.
The transfer phenomenon can happen in two ways. The
rst occurs between gas and particles, like many others
processes that involve gas and solids. The second way happens between the bed phases (bubble, cloud and emulsion),
and occurs only in uidized bed processes. Because of this,
understanding bubble hydrodynamics is important in bubbling uidized bed (BFB) modeling, since they are responsible
for the turbulence that favors the transfer phenomena, maintaining the supply of reactant gas for the emulsion, and they
move the solids. They also favor particles elutriation and bed
expansion [24].
Bubbles in uidized beds behave just like liquid bubbles
with low viscosity. Generally, they are assumed to have spherical shape, be dependent on bubble velocity and diameter, be
dependent on the increase in diameter, near the surface due
to decreased hydrostatic pressure and be dependent on the
coalescence phenomenon of adjacent bubbles.

2.3.

Numerical simulation

The numerical simulation of the uidized bed is based on

the classical equations of mass conservation, energy and
momentum, coupled with equations that describe the interactions between the phases. This is based on chemical kinetics,
in experimental correlations and models derived from the
Kinetic Theory of Granular Flows (KTGF) [15,26,27]. To describe
gassolid ow in uidized beds, some numerical models were
developed in the last years, such as LatticeBoltzmann (LBM),
Discrete Particle Model (DPM) and the Two Fluid Model (TFM).
Van der Hoef et al. [28] describe these models.
The hydrodynamic numerical model can be described in
two ways, with EulerEulers approach or EulerLagranges
approach [27]. The computational cost, in the latter approach,
is higher than in the Eulerian one, therefore, the rst approach
is more used in numerical simulation of uidized beds. Boemer et al. [29] stated that the EulerEuler approach with the
Kinetic Theory (KTGF) describes the processes in uidized
beds in agreement with experimental data.
Nowadays there is an availability of commercial codes like
CFX, FLUENT, ASPEN and MFIX, which use the nite volumes

211

method to model phenomena that involve uid mechanics,

heat transfer, combustion and gasication [30].

3.
State of the art: modeling and simulation
of BFB
Fluidized bed modeling and simulation studies can be divided
into two types, one that uses the CFD tool applied to the analysis of a process, and another applied to the analysis of the
model, based on the approach and the numerical method,
aiming at the improvement of the CFD tool.
Chavarie and Grace [31] conducted a study about catalytic
decomposition of ozone in a uidized bed, where they evaluated the results of some mathematical models comparing
with experimental data. The models analyzed were Davidson
and Harrisons [1], Partridge and Rowes [23], Kunii and Levenspiels [2] and Kato and Wens [4], and the last two presented
the best results. Kunii and Levenspiels [2] model is useful
to describe the chemical reactions, while Kato and Wens [4]
model is useful to represent the hydrodynamics of the uidized bed.
Van Lare [19] studied the inuence of particle size on mass
transfer in a uidized bed. The author used the two-phase
model along with Van Deemters model [32] and experimental data, to obtain a simple model that produces reasonable
predictions. The author indicated that to maximize heterogeneous reaction conversions, bubbleemulsion mass transfer
should be the highest possible. The results showed that group
B particles [5] are more efcient than the smaller particles.
Carvalho [10] studied methanol production in BFB. The
author modeled, scaled and simulated a reactor using the
operational conditions of the xed bed. The author also
used the two-phase model and the perfect blend model, and
concluded that the latter one is more advantageous for preliminary calculations of reactor performance. It generates a less
complex system of equations as well as required information.
Matos [33] studied coke combustion in a uidized bed,
based on Davidson and Harrisons [1] two-phase model,
which presented many numerical problems in fast reactions,
requiring better development of the numerical method. The
CSTR-PFR model, where bubbles have a sub model given by
Toomey and Jonhstone [22], proved suitable to describe the
ow in a uidized bed with injector holes distributor, since
the velocity is 23 times higher than the minimum uidization velocity. The inuence of many variables in uidized bed
models for rate constants of the rst order proved relatively
limited. The numerical model proved very applicable in tested
cases.
Mota [34] developed a mass transfer mechanisms based on
coke combustion in a uidized bed. The author analyzed coke
combustion with high ash content and uniform size, particle
combustion rate with varied initial sizes, and the O2 transfer
from bubbles to the dense phase. A theoretical model for each
subject was developed and then experimentally proved. The
results show that O2 transfer resistance depends on their diffusion that increases with the fraction of xed carbon. The
tortuosity (diffusion path in the particle porous matrix) varies
between 3 and 8 and has no relation with fraction xed carbon.
Another important conclusion shows that O2 transfer during

212

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

bubble formation has an important role in overall mass transfer (bubbleemulsion). The author says that it is important to
consider the interaction between consecutive bubbles and the
coalescence phenomenon.
Tarelho [17] studied the control of gaseous emissions during coal combustion in a uidized bed, and stated that for
chemical reactor modeling, before introducing the chemical parameters, rst it is necessary to represent the physical
and hydrodynamic characteristics of the system as realistically as possible. The numerical modeling used was based
on Rajan and Wens model [35], where mathematical simulations that describe particular aspects of combustion processes
have been used, by considering that these models simulations
should be simple enough not to require extensive calculations. The results showed that the model makes possible a
reasonable description of the qualitative behavior of gaseous
species along the reactor. The kinetic mechanism of some
gaseous species cannot be described by simple kinetics and
must include the radicals (O, H and OH). Thus, the inuence of
the solid particles present in the environment in the concentration of these radicals must be added. Although the model
reasonably simulates the NO along the reactor, the model
predicts a high decrease in concentration with increasing
temperature, which is not observed experimentally. In relation to the mathematical model used, Tarelho [17] suggested
a sensitivity analysis on the model to some hydrodynamic
parameters, such as diameter of the bubble uidization velocity, and gas exchange between bubble phase and emulsion
phase.
Gambetta [36] developed a simplied generic dynamic
model for polymerization reactors in a uidized bed to predict
operating conditions and product properties. The author used
Kunii and Levenspiels [21] and Choi and Rays models [37]. The
reduced model used other measures for the reactor without
changing its kinetics. The need for simultaneous estimation of
the kinetic parameters and for adjustment parameters of the
control mesh was removed, besides reducing iteration time
by at least ten times. The results showed that the method of
estimation of kinetic parameters using the reduced model is
valid.
Farias Jnior [38] modeled and simulated the dynamics
of a natural gas combustor in a uidized bed, where he
adopted three different approaches, two one-dimensional
(plug ow modied and the two-phase theory) and one twodimensional with the MFIX code (CFD). The one-dimensional
models are simpler and require less computational resources,
offering answers timely for advanced control systems. The
two-dimensional model is able to provide more detailed proles on several variables along the bed, which is useful for
assessing its behavior. The simulation of one-dimensional
models developed (Fluidization Simulator) showed a much
better performance in the FORTRAN version enabling a
real-time simulation. The two-dimensional model (CFD), as
expected, showed high computational cost.
Neves [24] studied the heterogeneous reactions in coal gasication in a BFB, in such study he analyzed the kinetic NO
reduction with experimental data from Matos [33]. The author
used Davidson and Harrisons model [1], which allows an evaluation of the kinetic and diffusive limitations of the boundary
layer of particles and of the hydrodynamics of the bed. The

methodology used is intended to correct the rate of chemical

reactions observed in association with effects of the reactor
hydrodynamics and mass transfer phenomena in the boundary layer and within the particle. This is an alternative model
for global constant of heterogeneous chemical reaction, based
on the concentration of NO in the entrance of the reactor,
originally developed by Matos [39], and it also allows us to
determine the intrinsic constant based on the contribution
of diffusive and hydrodynamic limitations modeling for the
global constant. The characterization of heterogeneous kinetics in uidized beds lacks a quantitative assessment of the
role of the various phenomena of mass transfer to the apparent rate of chemical reaction. The proposed model allows us
to conclude that the study of kinetics of heterogeneous reactions in a uidized bed should involve a review of the effects of
mass transfer and hydrodynamics of the reactor at the rate of
chemical reaction observed. This is especially relevant when
carrying out trials with large particles and high temperatures
for it is likely to cause a signicant conversion of the gaseous
reactant in the uidized bed.
Silva [40] used Tarelhos [17] model to understand how the
release location of volatile matter inuences during biomass
combustion in a uidized bed through mass balance. One of
the parameters analyzed in the simulations was the excess
of air, which proved to be very important in relation to the
behavior of gaseous species and carbonized matter along the
reactor. Species such as CO2 , CO, H2 and H2 O have, according to simulations, maximum concentration between the bed
surface and the secondary air feeding point and biomass. Due
to high turbulence in the surface of the bed, it becomes difcult to model this region. Compared with experimental data,
the simulation results showed that the kinetics used for CO
is not the best one. One of the limitations of the model lies
in the fact that it considers complete mixing in each compartment, which in reality does not happen. However, overall,
the model used is valid because it showed concordance in the
concentration proles.
Wanderley [41] modeled and simulated the BFB reactor
to obtain 1,2-dichloroethane from an oxychlorination reaction. The author used a phenomenological model based on
the two-phase theory, in which he considered a system of
one-dimensional ow in axial direction and steady state. The
effects of uid dynamics variables, the bubbleemulsion mass
transfer and the importance of the freeboard region in the
overall conversion were also considered. The study showed
that the model adequately represents an industrial reactor.
The simplifying assumptions adopted were also adequate,
because they did not inuence the results. The results showed
that the process is very sensitive to operating pressure, temperature of the cooling water, minimum uidization height,
and variation of the particle diameter as well as the bubble
and the reactor diameter. With the exception of the bubble
diameter, the increase in other parameters causes a rise in
the conversion reactor. The diffusion phenomenon exercises
a more important role than the residence time in the reactor
conversion. Increasing the diameter of the reactor causes a
signicant reduction in the height of the expanded bed and
a signicant increase in conversion, given the wider diffusion of species in the bubbleemulsion interface caused by
the reduction of the bubbles diameter. The air distributor is an

213

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

Table 1 Summary of studies which used modeling and simulation to analyze processes in a uidized bed.
Author
[10]
[17]
[19]
[24]
[31]
[33]
[34]
[36]
[38]
[40]
[41]
[42]

Institution/Country
Universidade Estadual de Campinas, Brazil
Universidade de Aveiro, Portugal
Technische Universiteit Eindhoven, The Netherlands
Universidade de Aveiro, Portugal
McGill University, Canada
Universidade de Aveiro, Portugal
Universidade do Porto, Portugal
Universidade Federal do Rio Grande do Sul, Brazil
Universidade Federal de Pernambuco, Brazil
Universidade de Aveiro, Portugal
Universidade Federal de Alagoas, Brazil
Universidade de So Paulo, Brazil

important step in the design of reactors, because it has a direct

inuence on bubble size. This should be designed to produce
bubbles initially of small diameter and low velocities of the
gases, favoring mass transfer in bubbleemulsion interface.
The catalyst particle diameter affects more signicantly the
conversion. The increase of particle diameter causes a reduction not only in the bubble size, but also in the expanded bed
height and in minimum uidization height. Wanderley [41]
stated that the diffusive effect exerts considerable inuence
on the process, where the contribution bubbleemulsion is
more important than the emulsionbubble one. The freeboard
region is very important in the overall performance of the reactor, and should not be neglected in modeling work of uidized
bed reactors, especially when it comes to real industrial plants
that operate at high ow rates. In this region, the particle size
is important to conversion, where the small particles are more
easily drawn into this region, resulting in higher conversion
rates. For particles with higher diameter, the contribution of
the freeboard region becomes less signicant.
Moraes [42] developed a modeling and simulation of atmospheric BFB combustion of coal with high ash and sulfur with
desulfurization by limestone. The author used a phenomenological approach that showed the need for improvement. In
comparison with experimental data generated in a pilot plant,
some differences and concordances in gas concentration and
particle size distribution proles were reported, as well as
quantitative results of sulfur absorption efciency compared
with the experimental data.
Table 1 summarizes the aforementioned studies that used
the modeling and simulation of a BFB applied to specic cases.

3.1.

Process/analysis

Computational uid dynamic CFD

The CFD tool is widely used to analyze uidized bed reactor

behavior, but the modeling still presents numerical instability
in the equations, and the gassolid interphase is transient and
known only in some regions [43]. Because of this, nowadays it
is possible to nd many studies in the literature on numerical
modeling of a uidized bed that seeks to improve the CFD tool.
The following are some of these relevant studies.
Kuipers et al. [44] analyzed numerically the hydrodynamics
of the uidized bed and the model used was EulerEulers, twouid 2-D, applying the nite differences method. The authors
observed that the bubbles shape is sensitive to the rheology
of the bed, but the size (growth) is not greatly inuenced by

Methanol production
Coal combustion
Particle size inuence in mass transfer
Coal gasication
Catalytic decomposition of ozone
Coke combustion
Mass transfer mechanisms in coke combustion
Polymerization process
Natural gas combustion
Biomass combustion
Oxychlorination reaction
Coal combustion with high ash and sulfur contents

this physical property. The model needs better development

of rheology and numerical method.
Goldschmitt et al. [45] analyzed the effect of the coefcient of restitution in hydrodynamic modeling of dense gas
uidized beds using the kinetic theory of granular ow (KTGF)
and the two-uid EulerEulers model. The authors concluded
that the hydrodynamics of the dense uidized bed depends
strongly on the amount of energy dissipated due to solidsolid
interactions. Goldschmitt et al. [45] used the discrete particle
model as a useful tool for the analysis of solidsolid interaction. However, it lacked experiments on interactions between
the particles for correct validation of the numerical models.
They also argue that it is necessary a further development of
KTGF multi-uid models.
Pain et al. [46] studied the application of chaos theory in
the transient uidized bed simulation, using the nite elements method. The authors claimed that the scheme is stable,
allowing large time steps, and has good exibility for complex
geometries, but the chaotic behavior requires further mesh
renement. The results of the average bed height and average
velocity vectors turned out to be qualitatively consistent with
the experimental ndings.
Huilin et al. [47] simulated the uidized bed to analyze the
motion of the particles, assuming a binary mixture as well as
the KTGF and EulerEulers approach. The authors concluded
from this study that, in order to obtain the correct dynamics of
the bed, it is important to consider the distribution of particle
size and the energy dissipation due to solidsolid interactions.
The effect of these interactions is as critical as the rheology
and particle collision parameters, but it lacks experimental
studies for validation.
Mineto et al. [48] studied the inuence of particle diameter
on the hydrodynamic simulation of the uidized bed using
the two-phase models by EulerEuler and KTGF. The authors
observed that in dense regions, the granular temperature is
minimal, and in the bubble region is maximum. For particles
with an average diameter of 500 m, the cohesive forces can
be neglected, where the central bubbles are well dened and
with lower speeds. In simulations of particles with an average
diameter of 125 m, the cohesive forces are considered the
main factor in the stability of the ow, where they show a
homogeneous ow and are well mixed with higher speeds.
Papadikis et al. [49] simulated the fast pyrolysis process
of biomass in a uidized bed using the EulerEuler approach
and the KTGF. The authors concluded that the drag force is

214

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

an important parameter that denes particle motion. In the

dense region, the drag is more important than the virtual mass
effect, and in the diluted region, the drag is induced by gas.
An increased gas velocity gradient increases the importance
of the effect of virtual mass. Particle behavior is similar in
both cases, 2-D and 3-D. The authors stated that for studies
concerning hydrodynamics of the bed, the 2-D simulation provides good results, but for heat, mass and momentum transfer,
the 3-D simulation is more suitable. The model was considered
valid by Papadikis et al. [49] for the design of reactors, since
it permits to analyze the motion of the particles. In Papadikis
et al. [13,50], there was a continuation of this study about modeling and simulation of fast pyrolysis of biomass in BFB. In
another work, Papadikis et al. [51] analyzed the effect of particle size on the drag of the carbonized in coal combustion in
BFB.
Souza [52] investigated the diffusion effects in the numerical discretization of the convection terms and the dependence
on computational mesh size of the uidized bed in the simulation using the two-phase EulerEuler model and KTGF. The
author noted that the rst order method FOUP (First Order
UPwind) is highly diffusive requiring rened meshes. The high
order method called Superbee, presented results of better
quality compared to the experimental results of Kuipers et al.
[44], and it allows the use of coarser mesh. The effects of
numerical diffusion decrease with increasing number of iterations the Superbee. The author further stated that high-order
methods tend to produce numerical oscillations near high gradients, requiring the application of a ow restrictor.
Philippsen [53] analyzed four gassolid drag models
[26,5456] in hydrodynamic simulation of a uidized bed,
which used the EulerEuler approach and KTGF. The results
showed that the Syamlal and OBrien correlation [26] presents
good results about volume fraction of the phases involved and
gas velocity prole. Additionally, it has the best mesh convergence and uses the lowest computational time; about half of
the time used by other models. For a bubble study, the models
by Gidaspow [54] and Hill-Koch-Ladd [56] are the most appropriate because their presented results were more congruent
with experimental data.
Studies about BFB simulation using the CFD tool may be
found in [11,43,57,58] among others. In Brazil, uidized bed
simulation has been studied by some research groups like the
Graduate Program in Mechanical Engineering from UNISINOS
RS, the CTCL group (Centro Tecnolgico de Carvo Limpo)
from SATC SC, the Faculty of Engineering from So Paulo
State University Jlio de Mesquita Filho SP, and So Carlos
School of Engineering of from USP, SP. These groups have some
publications on the use of the CFD tool [12,48,52,5961].

4.

Final comments

Modeling and simulation of a uidized bed is largely used

nowadays, not only to predict but also to analyze and improve
different processes, and it has presented good results even
though it still needs improvement. In the present work, the
hydrodynamic modeling state of the art works regarding the
use of simulation applied to the analysis of processes were
reviewed.

Based on the studies reviewed, it was concluded that the

hydrodynamic modeling of a uidized bed lacks important
details for a more precise analysis of the process. There
is also a need of experimental data, particularly relating
to gassolid and solidsolid interactions for the validation
of existing mathematical models. Understanding the interactions between the phases and the inuence of these
interactions in the process is crucial, since they affect the
dynamic behavior of the reactor and the chemical species conversion.
In the numerical simulation of a uidized bed, due to the
complexity of the equations that describe the ow, there is a
need for the improvement of numerical technique and more
powerful computers, because using the CFD tool for simulating processes in uidized beds still has a high computational
cost.

Conicts of interest
The authors declare no conicts of interest.

Acknowledgements
The authors wish to acknowledge CNPq and Rede Carvo for
supporting and sponsoring this research.

references

[1] Davidson JF, Harrison D. Fluidized particles. Cambridge:

Cambridge University Press; 1963.
[2] Kunii D, Levenspiel O. Fluidization engineering. New York:
Jonh Wiley; 1969.
[3] Wen YC, Yu YH. Mechanics of uidization. Chem Eng Prog
Symp Ser 1966;62:10011.
[4] Kato K, Wen CY. Bubble assemblage model for uidized bed
catalytic reactors. Chem Eng Sci 1969;24:135169.
[5] Geldart D. Types of gas uidization. Powder Technol
1973;7:28592.
[6] Mori S, Wen CY. Estimation of bubble diameter in gaseous
uidized beds. AIChE J 1975;21:10915.
[7] Darton RC, Lanauze RD, Davidson JF, Harrison D. Bubble
growth due to coalescence in uidized beds. Trans Inst
Chem Eng 1977;55:27480.
[8] Haider A, Levenspiel O. Drag coefcient and terminal
velocity of spherical and nonspherical particles. Powder
Technol 1989;58:6370.
[9] Yang WC. Handbook of uidization and uid-particle
systems. New York: Marcel-Dekker Inc., Taylor & Francis
Group LLC; 2003.
[10] Carvalho FO [Dissertaco de Mestrado] Modeling and
simulation of a uidized bed reactor for the methanol
production. So Paulo: Faculdade de Engenharia Qumica,
UNICAMP; 1994 [in Portuguese].
[11] Ravelli S, Perdichizzi A, Barigozzi G. Description, applications
and numerical modelling of bubbling uidized bed
combustion in waste-to-energy plants. Prog Energy Combust
Sci 2008;34:22453.
[12] Santos FLP [Tese de doutorado] Eulerian numerical
simulation of gassolid ow in riser with reduced
dimensions applying rened meshes. So Carlos: Escola de
Engenharia de So Carlos, USP; 2008 [in Portuguese].

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

[13] Papadikis K, Gu S, Bridgwater AV. CFD modelling of the fast

pyrolysis of biomass in uidised bed reactors: Part B. Heat,
momentum and mass transport in bubbling uidised beds.
Chem Eng Sci 2009;64:103645.
[14] Wu Z. Understanding uidized bed combustion. London: IEA
Clean Coal Centre, CCC/76; 2003, ISBN 92-9029-391-8, 42 pp.
[15] Milioli CC [Tese de Doutorado] Simulation of reactive ows
gassolid in circulating uidized beds. So Carlos: Escola de
Engenharia de So Carlos, USP; 2006 [in Portuguese].
[16] Ergudenler A, Ghaly AE, Hamdullahpur F, Al-Taweel AM.
Mathematical modeling of uidized bed straw gasier Part
I: model development. Energy Sources 1997;19:106584.
[17] Tarelho LAC [Tese de Doutorado] Gaseous pollutants
emissions control from coal combustion in uidized bed.
Aveiro: Cincias Aplicadas ao Ambiente, Universidade de
Aveiro; 2001 [in Portuguese].
[18] Deen NG, Van Sint Annaland M, Kuipers JAM. Detailed
computational and experimental uid dynamics of uidized
beds. Appl Math Model 2006;30:145971.
[19] Van Lare CEJ [Tese de Doutorado] Mass transfer in gas
uidized beds: scaling, modeling and particle size inuence.
Technische Universiteit Eindhoven; 1991.
[20] Paiva JLMS [Tese de Doutorado] The particle size inuence
on the hydrodynamic behavior of uidized beds. Coimbra:
Cincias da Engenharia Mecnica, Universidade de Coimbra;
2001 [in Portuguese].
[21] Kunii D, Levenspiel O. Fluidization engineering. Boston:
Butterworth-Heinemann; 1991.
[22] Toomey RO, Johnstone HF. Gaseous uidization of solid
particles. Chem Eng Prog 1952;48:2206.
[23] Partridge BA, Rowe PN. Chemical reaction in a bubbling
gas-uidised bed. Trans Inst Chem Eng 1966;44:33548.
[24] Neves DSF [Dissertaco de Mestrado] Heterogeneous
reactions of the carbonized gasication in a uidized bed.
Aveiro: Departamento de Ambiente e Ordenamento da
Universidade de Aveiro; 2007 [in Portuguese].
[25] Grace JR. Contacting modes and behaviors classication of
gassolid and other two phase suspensions. Can J Chem Eng
1986;64:35363.
[26] Syamlal M, Rogers W, OBrien T. MFIX documentation: theory
guide. Technical note, DOE/METC-94/1004. US Department of
Energy; 1993.
[27] Gidaspow D. Multiphase ow and uidization continuum
and kinetic theory descriptions. Chicago: Academic Press;
1994. p. 466. ISBN: 0-12-282470-9.
[28] Van Der Hoef MA, Van Sint Annalabd M, Deen NG, Kuipers
JAM. Numerical simulation of dense gassolid uidized beds:
a multiscale modeling strategy. Annu Rev Fluid Mech
2008;40:4770.
[29] Boemer A, Qi H, Renz U. Eulerian simulation of bubble
formation at a jet in a two-dimensional uidized bed. Int J
Multiphase Flow 1997;23:92744.
[30] Melo BA [Dissertaco de Mestrado] Computational
evaluation of a gasication system in uidized bed using
CSFB software. Itajub: Programa de P-Graduaco em
Engenharia Mecnica, Universidade Federal de Itajub; 2008
[in Portuguese].
[31] Chavarie C, Grace JR. Performance analysis of a uidized bed
reactor: II. Observed reactor behavior compared with simple
two-phase models. Ind Eng Chem Fundam 1975;14:7986.
[32] Van Deemter JJ. Mixing and contacting in gas-solid uidized
beds. Chem Eng Sci 1961;13:14354.
[33] Matos MAA [Tese de Doutorado] The formation and reducing
of the NOx in the coke combustion in uidized bed. Aveiro:
Departamento de Ambiente e Ordenamento, Universidade
de Aveiro; 1995 [in Portuguese].
[34] Mota ODS [Tese de Doutorado] Estudos sobre combusto de
coque e transferncia de massa em leitos uidizados. Porto:

[35]
[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

215

Departamento de Engenharia Mecnica, Universidade do

Porto; 1997.
Rajan RR, Wen CY. A comprehensive model for uidized bed
coal combustors. AIChE J 1980;26:64255.
Gambetta R [Dissertaco de Mestrado] Modeling and
simulation of polymerization reactors in uidized bed. Porto
Alegre: Programa de Ps-Graduaco em Engenharia Qumica,
UFRGS; 2001 [in Portuguese].
Choi KY, Ray WH. The dynamic behaviour of uidized bed
reactors for solid catalysed gas phase olen polymerization.
Chem Eng Sci 1985;40:226179.
Farias Jnior FS [Dissertaco de Mestrado] Modeling and
dynamic simulation of the natural gas combustor in
uidized bed. Recife: Programa de Ps-Graduaco em
Engenharia Qumica, Universidade Federal de Pernambuco;
2004 [in Portuguese].
Matos MA [Dissertaco de Mestrado] Kinetics of reduction of
nitric oxide (NO) by coke in uidized bed. Lisboa: Instituto
Superior Tcnico; 1988 [in Portuguese].
Silva ARM [Dissertaco de Mestrado] Simulation of biomass
combustion in uidized bed mass balance. Aveiro:
Engenharia do Ambiente, Universidade de Aveiro; 2009 [in
Portuguese].
Wanderley PRB [Dissertaco de Mestrado] Modeling and
simulation of an oxychlorination reactor in uidized bed for
the production of 1,2-dichloroethane. Macei: Engenharia
Qumica, Universidade Federal de Alagoas; 2010. p. 171 [in
Portuguese].
Moraes AAU [Tese de Doutorado] Modeling and simulation of
uidized bed combustion of coal with high ash content with
desulfurization by limestone. So Carlos: Escola de
Engenharia de So Carlos da USP; 2011 [in Portuguese].
Taghipour F, Ellis N, Wong C. Experimental and
computational study of gassolid uidized bed
hydrodynamics. Chem Eng Sci 2005;60:685767.
Kuipers JAM, Van Duin KJ, Van Beckum FPH, Van Swaaij
WPM. Computer simulation of the hydrodynamics of a
two-dimensional gas-uidized bed. Comput Chem Eng
1993;17:83958.
Goldschmidt MJV, Kuipers JAM, Van Swaaij WPM.
Hydrodynamic modeling of dense gas-uidised beds using
the kinetic theory of granular ow: effect of coefcient of
restitution on bed dynamics. In: Sixteenth international
symposium on chemical reaction engineering ISCRE-16.
2000.
Pain CC, Mansoorzadeh S, Gomes JLM, Oliveira CRE. A
numerical investigation of bubbling uidized bed dynamics
in 2-D geometries. Powder Technol 2002;128:5677.
Huilin L, Yurong H, Gidapow D. Hydrodynamic modeling of
binary mixture n a gas bubbling uidized bed using the
kinetic theory of granular ow. Chem Eng Sci
2003;58:1197205.
Mineto AT, Souza MP, Silva RC, Cabzas-Gomes L, Montagnoli
AN, Navarro HA. Particle diameter inuence on simulation of
gassolid ow in bubbling uidized bed. In: 7th Brazilian
conference on dynamics, control and applications. 2008 [in
Portuguese].
Papadikis K, Bridgwater AV, Gu S. CFD modelling of the fast
pyrolysis of biomass in uidised bed reactors: Part A.
Eulerian computation of momentum transport in bubbling
uidised beds. Chem Eng Sci 2008;63:421827.
Papadikis K, Gu S, Bridgwater AV. CFD modelling of the fast
pyrolysis of biomass in uidised bed reactors: modelling the
impact of biomass shrinkage. Chem Eng J 2009;149:
41727.
Papadikis K, Gu S, Bridgwater AVA. CFD approach on the
effect of particle size on char entrainment in bubbling
uidised bed reactors. Biomass Bioenergy 2010;34:219.

216

j m a t e r r e s t e c h n o l . 2 0 1 5;4(2):208216

[52] Souza MP [Dissertaco de Mestrado] Numerical effects on

the simulation of gassolid ow in bubbling uidized bed
using the kinetic theory of granular ow. So Paulo:
Faculdade de Engenharia da Universidade Estadual Paulista
Jlio de Mesquita Filho, UNESP; 2009 [in Portuguese].
[53] Philippsen CG [Dissertaco de Mestrado] Study of
performance of solidgas drag correlations on numerical
simulation of a bubbling uidized bed. So Leopoldo:
Programa de Ps-Graduaco em Engenharia Mecnica,
UNISINOS; 2012 [in Portuguese].
[54] Gidaspow D. Hydrodynamics of uidization and heat
transfer: supercomputer modeling. Appl Mech Rev
1986;39:123.
[55] Arastoopour H, Pakdel P, Adewumi M. Hydrodynamic
analysis of dilute gassolid ow in a vertical pipe. Powder
Technol 1990;62:16370.
[56] Hill RJ, Koch DL, Ladd JC. The rst effects of uid inertia on
ows in ordered and random arrays of spheres. J Fluid Mech
2001;448:21341.

[57] Huilin L, Gidapow D. Hydrodynamics of binary uidization in

a riser: CFD simulation using two granular temperatures.
Chem Eng Sci 2003;58:377792.
[58] Rozainee M, Ngo SP, Salema AA, Tan KG. Computational uid
dynamics modeling of rice husk combustion in a uidised
bed combustor. Powder Technol 2010;203:33147.
[59] Pedroso FA, Zinani F, Indrusiak MLS. Numerical simulation of
the uid dynamics of a circulating uidized bed using the
Euler-granular modeling. In: IV Congresso Brasileiro de
Carvo Mineral. 2013 [in Portuguese].
[60] Silva RC, Navarro HA, Cabezas-Gmez L. Analysis of the
numerical discretization schemes inuence on a gassolid
two-phase ow in a bubbling uidized bed. Far East J Appl
Math 2010;38:83100.
[61] Kestering DA, de Aquino TF, Spillere Ronchi JC. CFD
simulation of Brazilian coal gasication on a circulating
uidized bed reactor. In: International Pittsburgh coal
conference. 2012.